diff --git "a/checkpoint-5500/trainer_state.json" "b/checkpoint-5500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-5500/trainer_state.json"
@@ -0,0 +1,33016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.4,
+  "global_step": 5500,
+  "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": 1.9315,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.8391,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019996798975672214,
+      "loss": 1.8441,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019993597951344433,
+      "loss": 1.7656,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019990396927016646,
+      "loss": 1.6324,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019987195902688862,
+      "loss": 1.677,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019983994878361075,
+      "loss": 1.4647,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019980793854033291,
+      "loss": 1.4285,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019977592829705507,
+      "loss": 1.3052,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001997439180537772,
+      "loss": 1.2885,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019971190781049937,
+      "loss": 1.1584,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019967989756722153,
+      "loss": 1.106,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019964788732394366,
+      "loss": 1.2272,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019961587708066582,
+      "loss": 1.1007,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019958386683738798,
+      "loss": 0.9575,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019955185659411014,
+      "loss": 1.1255,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019951984635083227,
+      "loss": 0.9909,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019948783610755443,
+      "loss": 1.0567,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001994558258642766,
+      "loss": 1.1133,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019942381562099872,
+      "loss": 1.0576,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019939180537772088,
+      "loss": 1.0808,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019935979513444302,
+      "loss": 0.9458,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001993277848911652,
+      "loss": 1.0027,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019929577464788734,
+      "loss": 0.9723,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019926376440460947,
+      "loss": 0.9045,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019923175416133163,
+      "loss": 0.8118,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001991997439180538,
+      "loss": 0.9211,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019916773367477595,
+      "loss": 0.9402,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019913572343149808,
+      "loss": 1.0336,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019910371318822024,
+      "loss": 0.9593,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001990717029449424,
+      "loss": 0.9217,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019903969270166453,
+      "loss": 0.914,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001990076824583867,
+      "loss": 0.8461,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019897567221510885,
+      "loss": 0.9383,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019894366197183098,
+      "loss": 1.0103,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019891165172855314,
+      "loss": 0.7654,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001988796414852753,
+      "loss": 0.8492,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019884763124199746,
+      "loss": 0.9894,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001988156209987196,
+      "loss": 0.94,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019878361075544173,
+      "loss": 0.913,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019875160051216392,
+      "loss": 1.0549,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019871959026888605,
+      "loss": 0.8724,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001986875800256082,
+      "loss": 0.8876,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019865556978233034,
+      "loss": 1.013,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019862355953905253,
+      "loss": 0.9322,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019859154929577466,
+      "loss": 0.8533,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001985595390524968,
+      "loss": 0.8774,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019852752880921895,
+      "loss": 0.8907,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001984955185659411,
+      "loss": 0.914,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019846350832266327,
+      "loss": 1.0708,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001984314980793854,
+      "loss": 0.8649,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019839948783610757,
+      "loss": 0.9685,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019836747759282972,
+      "loss": 0.8426,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019833546734955186,
+      "loss": 0.8327,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019830345710627402,
+      "loss": 0.8601,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019827144686299618,
+      "loss": 0.98,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001982394366197183,
+      "loss": 0.8878,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019820742637644047,
+      "loss": 0.8317,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019817541613316263,
+      "loss": 0.8962,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001981434058898848,
+      "loss": 0.8735,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019811139564660692,
+      "loss": 0.8498,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019807938540332905,
+      "loss": 0.8897,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019804737516005124,
+      "loss": 0.8903,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019801536491677337,
+      "loss": 0.7908,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019798335467349553,
+      "loss": 0.8899,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019795134443021767,
+      "loss": 0.7803,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019791933418693983,
+      "loss": 0.9622,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019788732394366199,
+      "loss": 0.8514,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019785531370038412,
+      "loss": 0.8658,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019782330345710628,
+      "loss": 0.892,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019779129321382844,
+      "loss": 0.7928,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001977592829705506,
+      "loss": 0.9122,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019772727272727273,
+      "loss": 0.8282,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001976952624839949,
+      "loss": 0.9106,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019766325224071705,
+      "loss": 0.9137,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019763124199743918,
+      "loss": 0.8609,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019759923175416134,
+      "loss": 0.8198,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001975672215108835,
+      "loss": 0.805,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019753521126760564,
+      "loss": 0.8454,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001975032010243278,
+      "loss": 0.8345,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019747119078104995,
+      "loss": 0.9347,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019743918053777211,
+      "loss": 0.9628,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019740717029449425,
+      "loss": 0.8413,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019737516005121638,
+      "loss": 0.8286,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019734314980793857,
+      "loss": 0.8401,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001973111395646607,
+      "loss": 0.8555,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019727912932138286,
+      "loss": 0.8608,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000197247119078105,
+      "loss": 0.8512,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019721510883482715,
+      "loss": 0.8502,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001971830985915493,
+      "loss": 0.883,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019715108834827144,
+      "loss": 0.8998,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001971190781049936,
+      "loss": 0.8364,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019708706786171576,
+      "loss": 0.9417,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001970550576184379,
+      "loss": 0.7837,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019702304737516006,
+      "loss": 0.8604,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019699103713188222,
+      "loss": 0.8501,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019695902688860438,
+      "loss": 0.9183,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001969270166453265,
+      "loss": 0.9338,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019689500640204864,
+      "loss": 0.8868,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019686299615877083,
+      "loss": 0.7937,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019683098591549296,
+      "loss": 0.9089,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019679897567221512,
+      "loss": 0.9192,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019676696542893725,
+      "loss": 0.6853,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019673495518565944,
+      "loss": 0.8323,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019670294494238157,
+      "loss": 1.0199,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001966709346991037,
+      "loss": 0.8575,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001966389244558259,
+      "loss": 0.9167,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019660691421254803,
+      "loss": 0.8873,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019657490396927018,
+      "loss": 0.8711,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019654289372599232,
+      "loss": 0.8671,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019651088348271448,
+      "loss": 0.7837,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019647887323943664,
+      "loss": 0.9155,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019644686299615877,
+      "loss": 0.885,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019641485275288093,
+      "loss": 0.768,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001963828425096031,
+      "loss": 0.9048,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019635083226632522,
+      "loss": 0.8682,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019631882202304738,
+      "loss": 0.8946,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019628681177976954,
+      "loss": 0.9001,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001962548015364917,
+      "loss": 0.8769,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019622279129321383,
+      "loss": 0.8878,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019619078104993597,
+      "loss": 0.8588,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019615877080665815,
+      "loss": 0.8742,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001961267605633803,
+      "loss": 0.8681,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019609475032010245,
+      "loss": 0.7442,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019606274007682458,
+      "loss": 0.8789,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019603072983354674,
+      "loss": 0.907,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001959987195902689,
+      "loss": 0.8865,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019596670934699103,
+      "loss": 0.7627,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001959346991037132,
+      "loss": 0.9144,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019590268886043535,
+      "loss": 0.7996,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001958706786171575,
+      "loss": 0.8552,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019583866837387964,
+      "loss": 0.8462,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001958066581306018,
+      "loss": 0.9224,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019577464788732396,
+      "loss": 0.9863,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001957426376440461,
+      "loss": 0.8801,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019571062740076826,
+      "loss": 0.8503,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019567861715749042,
+      "loss": 0.7993,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019564660691421255,
+      "loss": 1.0245,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001956145966709347,
+      "loss": 0.8884,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019558258642765687,
+      "loss": 0.8398,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019555057618437903,
+      "loss": 0.947,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019551856594110116,
+      "loss": 0.9127,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001954865556978233,
+      "loss": 0.8593,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019545454545454548,
+      "loss": 0.922,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954225352112676,
+      "loss": 0.8535,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019539052496798977,
+      "loss": 0.9153,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001953585147247119,
+      "loss": 0.898,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019532650448143406,
+      "loss": 0.7648,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019529449423815622,
+      "loss": 0.8843,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019526248399487836,
+      "loss": 0.8734,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019523047375160052,
+      "loss": 0.8772,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019519846350832268,
+      "loss": 0.7813,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001951664532650448,
+      "loss": 0.8675,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019513444302176697,
+      "loss": 0.8901,
+      "step": 154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019510243277848913,
+      "loss": 0.7662,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001950704225352113,
+      "loss": 0.8703,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019503841229193342,
+      "loss": 0.8963,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019500640204865558,
+      "loss": 0.8979,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019497439180537774,
+      "loss": 1.0489,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019494238156209987,
+      "loss": 0.8099,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019491037131882203,
+      "loss": 0.9223,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948783610755442,
+      "loss": 0.8511,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019484635083226635,
+      "loss": 0.9503,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019481434058898849,
+      "loss": 0.8375,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019478233034571062,
+      "loss": 0.8652,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001947503201024328,
+      "loss": 0.7749,
+      "step": 166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471830985915494,
+      "loss": 0.7934,
+      "step": 167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946862996158771,
+      "loss": 0.8054,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019465428937259923,
+      "loss": 0.8301,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001946222791293214,
+      "loss": 0.8996,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019459026888604355,
+      "loss": 0.9442,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019455825864276568,
+      "loss": 0.8854,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019452624839948784,
+      "loss": 0.7999,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019449423815621,
+      "loss": 0.8808,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019446222791293213,
+      "loss": 0.7705,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001944302176696543,
+      "loss": 0.8195,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019439820742637645,
+      "loss": 0.8216,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019436619718309861,
+      "loss": 0.9044,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019433418693982075,
+      "loss": 0.9466,
+      "step": 179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019430217669654288,
+      "loss": 0.8257,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019427016645326507,
+      "loss": 0.8719,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001942381562099872,
+      "loss": 0.8018,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019420614596670936,
+      "loss": 0.9924,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019417413572343152,
+      "loss": 0.8662,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019414212548015368,
+      "loss": 0.8489,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001941101152368758,
+      "loss": 0.8398,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019407810499359794,
+      "loss": 0.8627,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019404609475032013,
+      "loss": 0.8292,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019401408450704226,
+      "loss": 0.8722,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019398207426376442,
+      "loss": 0.9201,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019395006402048656,
+      "loss": 0.9509,
+      "step": 191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019391805377720872,
+      "loss": 0.8291,
+      "step": 192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019388604353393088,
+      "loss": 0.757,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000193854033290653,
+      "loss": 0.8797,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019382202304737517,
+      "loss": 0.8004,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019379001280409733,
+      "loss": 0.7925,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019375800256081946,
+      "loss": 0.8024,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019372599231754162,
+      "loss": 0.8551,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019369398207426378,
+      "loss": 0.741,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019366197183098594,
+      "loss": 0.7821,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019362996158770807,
+      "loss": 0.7872,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001935979513444302,
+      "loss": 0.7361,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001935659411011524,
+      "loss": 0.9066,
+      "step": 203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019353393085787452,
+      "loss": 0.9637,
+      "step": 204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019350192061459668,
+      "loss": 0.7598,
+      "step": 205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019346991037131882,
+      "loss": 0.9266,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019343790012804098,
+      "loss": 0.9029,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019340588988476314,
+      "loss": 0.8283,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019337387964148527,
+      "loss": 1.0154,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019334186939820746,
+      "loss": 0.8852,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001933098591549296,
+      "loss": 0.8763,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019327784891165175,
+      "loss": 0.8474,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019324583866837388,
+      "loss": 0.8772,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019321382842509604,
+      "loss": 0.7933,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001931818181818182,
+      "loss": 0.8402,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019314980793854033,
+      "loss": 0.8051,
+      "step": 216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001931177976952625,
+      "loss": 0.806,
+      "step": 217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019308578745198465,
+      "loss": 0.7911,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019305377720870679,
+      "loss": 0.8357,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019302176696542895,
+      "loss": 0.8854,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001929897567221511,
+      "loss": 0.9124,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019295774647887326,
+      "loss": 0.7755,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001929257362355954,
+      "loss": 0.8193,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019289372599231753,
+      "loss": 0.8625,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019286171574903972,
+      "loss": 0.8731,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019282970550576185,
+      "loss": 0.8643,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000192797695262484,
+      "loss": 0.7401,
+      "step": 227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019276568501920614,
+      "loss": 0.7438,
+      "step": 228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001927336747759283,
+      "loss": 0.7978,
+      "step": 229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019270166453265046,
+      "loss": 0.8409,
+      "step": 230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001926696542893726,
+      "loss": 0.99,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019263764404609475,
+      "loss": 0.8783,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019260563380281691,
+      "loss": 0.9061,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019257362355953905,
+      "loss": 0.7678,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001925416133162612,
+      "loss": 0.8586,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019250960307298337,
+      "loss": 0.864,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019247759282970553,
+      "loss": 0.7929,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019244558258642766,
+      "loss": 0.8414,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019241357234314982,
+      "loss": 0.7847,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019238156209987198,
+      "loss": 0.8775,
+      "step": 240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001923495518565941,
+      "loss": 0.9297,
+      "step": 241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019231754161331627,
+      "loss": 0.7949,
+      "step": 242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019228553137003843,
+      "loss": 0.7288,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001922535211267606,
+      "loss": 0.701,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019222151088348272,
+      "loss": 0.8869,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019218950064020486,
+      "loss": 0.7423,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019215749039692704,
+      "loss": 0.8798,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019212548015364918,
+      "loss": 0.8075,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019209346991037134,
+      "loss": 0.8108,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019206145966709347,
+      "loss": 0.7848,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019202944942381563,
+      "loss": 0.9251,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001919974391805378,
+      "loss": 0.8289,
+      "step": 252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019196542893725992,
+      "loss": 0.7946,
+      "step": 253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019193341869398208,
+      "loss": 0.8757,
+      "step": 254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019190140845070424,
+      "loss": 0.9176,
+      "step": 255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019186939820742637,
+      "loss": 0.7873,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019183738796414853,
+      "loss": 0.8681,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001918053777208707,
+      "loss": 0.7723,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019177336747759285,
+      "loss": 0.9087,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019174135723431498,
+      "loss": 0.8274,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019170934699103714,
+      "loss": 0.7061,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001916773367477593,
+      "loss": 0.8029,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019164532650448144,
+      "loss": 0.9644,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001916133162612036,
+      "loss": 0.8109,
+      "step": 264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019158130601792576,
+      "loss": 0.7946,
+      "step": 265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001915492957746479,
+      "loss": 0.9144,
+      "step": 266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019151728553137005,
+      "loss": 0.8488,
+      "step": 267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019148527528809218,
+      "loss": 0.9137,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019145326504481437,
+      "loss": 0.8399,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001914212548015365,
+      "loss": 0.936,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019138924455825866,
+      "loss": 0.7787,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001913572343149808,
+      "loss": 0.7859,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019132522407170295,
+      "loss": 0.7855,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001912932138284251,
+      "loss": 0.7884,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019126120358514725,
+      "loss": 0.8351,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001912291933418694,
+      "loss": 0.81,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019119718309859157,
+      "loss": 0.8702,
+      "step": 277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001911651728553137,
+      "loss": 0.8879,
+      "step": 278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019113316261203586,
+      "loss": 0.8077,
+      "step": 279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019110115236875802,
+      "loss": 0.7394,
+      "step": 280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019106914212548018,
+      "loss": 0.8772,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001910371318822023,
+      "loss": 0.9009,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019100512163892444,
+      "loss": 0.8463,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019097311139564663,
+      "loss": 0.7462,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019094110115236876,
+      "loss": 0.7717,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019090909090909092,
+      "loss": 0.9058,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019087708066581308,
+      "loss": 0.899,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019084507042253521,
+      "loss": 0.8285,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019081306017925737,
+      "loss": 0.8832,
+      "step": 289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001907810499359795,
+      "loss": 0.835,
+      "step": 290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001907490396927017,
+      "loss": 0.8642,
+      "step": 291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019071702944942383,
+      "loss": 0.8272,
+      "step": 292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019068501920614596,
+      "loss": 0.7325,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019065300896286812,
+      "loss": 0.7843,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019062099871959028,
+      "loss": 0.8673,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019058898847631244,
+      "loss": 0.7734,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019055697823303457,
+      "loss": 0.9477,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019052496798975673,
+      "loss": 0.9324,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001904929577464789,
+      "loss": 0.8609,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019046094750320102,
+      "loss": 0.8278,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019042893725992318,
+      "loss": 1.0022,
+      "step": 301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019039692701664534,
+      "loss": 0.8121,
+      "step": 302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001903649167733675,
+      "loss": 0.8454,
+      "step": 303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019033290653008964,
+      "loss": 0.8791,
+      "step": 304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019030089628681177,
+      "loss": 0.8723,
+      "step": 305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019026888604353396,
+      "loss": 0.8677,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001902368758002561,
+      "loss": 0.8207,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019020486555697825,
+      "loss": 0.8039,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019017285531370038,
+      "loss": 0.8103,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019014084507042254,
+      "loss": 0.9038,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001901088348271447,
+      "loss": 0.7618,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019007682458386683,
+      "loss": 0.8253,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019004481434058902,
+      "loss": 0.7338,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019001280409731115,
+      "loss": 0.8628,
+      "step": 314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018998079385403328,
+      "loss": 0.8617,
+      "step": 315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018994878361075544,
+      "loss": 0.888,
+      "step": 316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001899167733674776,
+      "loss": 0.823,
+      "step": 317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00018988476312419976,
+      "loss": 0.9481,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001898527528809219,
+      "loss": 0.9014,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018982074263764406,
+      "loss": 0.9272,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018978873239436622,
+      "loss": 0.9695,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018975672215108835,
+      "loss": 0.8319,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001897247119078105,
+      "loss": 0.9359,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018969270166453267,
+      "loss": 0.7734,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018966069142125483,
+      "loss": 0.938,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018962868117797696,
+      "loss": 0.8277,
+      "step": 326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001895966709346991,
+      "loss": 0.8462,
+      "step": 327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018956466069142128,
+      "loss": 0.7983,
+      "step": 328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001895326504481434,
+      "loss": 0.9047,
+      "step": 329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00018950064020486557,
+      "loss": 0.8525,
+      "step": 330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001894686299615877,
+      "loss": 0.8341,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018943661971830987,
+      "loss": 0.8353,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018940460947503203,
+      "loss": 0.7323,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018937259923175416,
+      "loss": 0.8815,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018934058898847632,
+      "loss": 0.8846,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018930857874519848,
+      "loss": 0.7966,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001892765685019206,
+      "loss": 0.8179,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018924455825864277,
+      "loss": 0.772,
+      "step": 338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018921254801536493,
+      "loss": 0.8358,
+      "step": 339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001891805377720871,
+      "loss": 0.8171,
+      "step": 340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018914852752880922,
+      "loss": 0.9321,
+      "step": 341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018911651728553138,
+      "loss": 0.8696,
+      "step": 342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018908450704225354,
+      "loss": 0.8824,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018905249679897567,
+      "loss": 0.8447,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018902048655569783,
+      "loss": 0.9029,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018898847631242,
+      "loss": 0.8869,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018895646606914213,
+      "loss": 0.8039,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001889244558258643,
+      "loss": 0.9036,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018889244558258642,
+      "loss": 0.8788,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001888604353393086,
+      "loss": 0.9308,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018882842509603074,
+      "loss": 0.9206,
+      "step": 351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001887964148527529,
+      "loss": 0.8557,
+      "step": 352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018876440460947503,
+      "loss": 0.7907,
+      "step": 353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001887323943661972,
+      "loss": 0.8699,
+      "step": 354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018870038412291935,
+      "loss": 0.8015,
+      "step": 355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018866837387964148,
+      "loss": 0.86,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018863636363636364,
+      "loss": 0.7817,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001886043533930858,
+      "loss": 0.7862,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018857234314980794,
+      "loss": 0.8325,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001885403329065301,
+      "loss": 0.8379,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018850832266325226,
+      "loss": 0.8594,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018847631241997442,
+      "loss": 0.6996,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018844430217669655,
+      "loss": 0.7794,
+      "step": 363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001884122919334187,
+      "loss": 0.6593,
+      "step": 364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018838028169014087,
+      "loss": 0.7969,
+      "step": 365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000188348271446863,
+      "loss": 0.8293,
+      "step": 366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018831626120358516,
+      "loss": 0.9299,
+      "step": 367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00018828425096030732,
+      "loss": 1.0683,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018825224071702945,
+      "loss": 0.7665,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001882202304737516,
+      "loss": 0.8487,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018818822023047374,
+      "loss": 0.8032,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018815620998719593,
+      "loss": 0.8072,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018812419974391806,
+      "loss": 0.8926,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001880921895006402,
+      "loss": 0.7657,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018806017925736236,
+      "loss": 0.7209,
+      "step": 375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018802816901408452,
+      "loss": 0.8408,
+      "step": 376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018799615877080668,
+      "loss": 0.8641,
+      "step": 377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001879641485275288,
+      "loss": 0.7724,
+      "step": 378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018793213828425097,
+      "loss": 0.9203,
+      "step": 379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018790012804097313,
+      "loss": 0.6768,
+      "step": 380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00018786811779769526,
+      "loss": 0.9299,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018783610755441742,
+      "loss": 0.7752,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018780409731113958,
+      "loss": 0.8127,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018777208706786174,
+      "loss": 0.8628,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018774007682458387,
+      "loss": 0.9397,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000187708066581306,
+      "loss": 0.7555,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001876760563380282,
+      "loss": 0.8191,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018764404609475033,
+      "loss": 0.8863,
+      "step": 388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018761203585147249,
+      "loss": 0.7836,
+      "step": 389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018758002560819465,
+      "loss": 0.9117,
+      "step": 390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018754801536491678,
+      "loss": 0.8898,
+      "step": 391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018751600512163894,
+      "loss": 0.8702,
+      "step": 392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00018748399487836107,
+      "loss": 0.7765,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018745198463508326,
+      "loss": 0.8169,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001874199743918054,
+      "loss": 0.8381,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018738796414852752,
+      "loss": 0.7824,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018735595390524968,
+      "loss": 0.9157,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018732394366197184,
+      "loss": 0.9007,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000187291933418694,
+      "loss": 0.8198,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018725992317541613,
+      "loss": 0.827,
+      "step": 400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001872279129321383,
+      "loss": 0.7988,
+      "step": 401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018719590268886045,
+      "loss": 0.8396,
+      "step": 402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001871638924455826,
+      "loss": 0.8873,
+      "step": 403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018713188220230475,
+      "loss": 0.878,
+      "step": 404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001870998719590269,
+      "loss": 0.8117,
+      "step": 405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00018706786171574904,
+      "loss": 0.8518,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001870358514724712,
+      "loss": 0.7746,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018700384122919333,
+      "loss": 0.7945,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018697183098591552,
+      "loss": 0.935,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018693982074263765,
+      "loss": 0.8708,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001869078104993598,
+      "loss": 0.8271,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018687580025608194,
+      "loss": 0.899,
+      "step": 412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001868437900128041,
+      "loss": 0.8558,
+      "step": 413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018681177976952626,
+      "loss": 0.8582,
+      "step": 414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001867797695262484,
+      "loss": 0.8338,
+      "step": 415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018674775928297056,
+      "loss": 0.8864,
+      "step": 416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018671574903969272,
+      "loss": 0.7602,
+      "step": 417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00018668373879641485,
+      "loss": 0.8053,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000186651728553137,
+      "loss": 0.8215,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018661971830985917,
+      "loss": 0.9307,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018658770806658133,
+      "loss": 0.7872,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018655569782330346,
+      "loss": 0.9308,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018652368758002562,
+      "loss": 0.9111,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018649167733674778,
+      "loss": 0.8009,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001864596670934699,
+      "loss": 0.7599,
+      "step": 425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018642765685019207,
+      "loss": 0.8486,
+      "step": 426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018639564660691423,
+      "loss": 0.7983,
+      "step": 427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018636363636363636,
+      "loss": 0.8217,
+      "step": 428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018633162612035852,
+      "loss": 0.826,
+      "step": 429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018629961587708066,
+      "loss": 0.8371,
+      "step": 430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00018626760563380284,
+      "loss": 0.9477,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018623559539052498,
+      "loss": 0.9015,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001862035851472471,
+      "loss": 0.9626,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018617157490396927,
+      "loss": 0.9214,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018613956466069143,
+      "loss": 0.8551,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001861075544174136,
+      "loss": 0.8984,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018607554417413572,
+      "loss": 0.9231,
+      "step": 437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018604353393085788,
+      "loss": 0.889,
+      "step": 438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018601152368758004,
+      "loss": 0.8963,
+      "step": 439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018597951344430217,
+      "loss": 0.9209,
+      "step": 440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018594750320102433,
+      "loss": 0.7419,
+      "step": 441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001859154929577465,
+      "loss": 0.8324,
+      "step": 442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00018588348271446865,
+      "loss": 0.747,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018585147247119079,
+      "loss": 0.7793,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018581946222791295,
+      "loss": 0.7732,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001857874519846351,
+      "loss": 0.7829,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018575544174135724,
+      "loss": 0.7964,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001857234314980794,
+      "loss": 0.8326,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018569142125480156,
+      "loss": 0.7514,
+      "step": 449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001856594110115237,
+      "loss": 0.8751,
+      "step": 450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018562740076824585,
+      "loss": 0.9423,
+      "step": 451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018559539052496798,
+      "loss": 0.8569,
+      "step": 452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018556338028169017,
+      "loss": 0.8859,
+      "step": 453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001855313700384123,
+      "loss": 0.909,
+      "step": 454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018549935979513443,
+      "loss": 0.8874,
+      "step": 455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001854673495518566,
+      "loss": 0.8224,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018543533930857875,
+      "loss": 0.9106,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018540332906530091,
+      "loss": 0.8802,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018537131882202305,
+      "loss": 0.7875,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001853393085787452,
+      "loss": 0.9491,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018530729833546737,
+      "loss": 0.8323,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001852752880921895,
+      "loss": 0.6475,
+      "step": 462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018524327784891166,
+      "loss": 0.8548,
+      "step": 463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018521126760563382,
+      "loss": 0.8599,
+      "step": 464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018517925736235598,
+      "loss": 0.6655,
+      "step": 465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001851472471190781,
+      "loss": 0.8753,
+      "step": 466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018511523687580027,
+      "loss": 0.9359,
+      "step": 467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00018508322663252243,
+      "loss": 0.8909,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018505121638924456,
+      "loss": 0.7802,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018501920614596672,
+      "loss": 0.937,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018498719590268888,
+      "loss": 0.8469,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018495518565941102,
+      "loss": 0.7917,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018492317541613318,
+      "loss": 0.7483,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001848911651728553,
+      "loss": 0.7961,
+      "step": 474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001848591549295775,
+      "loss": 0.944,
+      "step": 475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018482714468629963,
+      "loss": 0.9443,
+      "step": 476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018479513444302176,
+      "loss": 0.9223,
+      "step": 477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018476312419974392,
+      "loss": 0.8869,
+      "step": 478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018473111395646608,
+      "loss": 0.696,
+      "step": 479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018469910371318824,
+      "loss": 0.8594,
+      "step": 480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00018466709346991037,
+      "loss": 0.7863,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018463508322663253,
+      "loss": 0.8353,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001846030729833547,
+      "loss": 0.7911,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018457106274007682,
+      "loss": 0.8551,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018453905249679898,
+      "loss": 0.9459,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018450704225352114,
+      "loss": 0.8402,
+      "step": 486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018447503201024328,
+      "loss": 0.9455,
+      "step": 487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018444302176696544,
+      "loss": 0.9049,
+      "step": 488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018441101152368757,
+      "loss": 0.868,
+      "step": 489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018437900128040976,
+      "loss": 0.8347,
+      "step": 490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001843469910371319,
+      "loss": 0.7836,
+      "step": 491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00018431498079385405,
+      "loss": 0.741,
+      "step": 492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001842829705505762,
+      "loss": 0.7387,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018425096030729834,
+      "loss": 0.7173,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001842189500640205,
+      "loss": 0.7886,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018418693982074263,
+      "loss": 0.9398,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018415492957746482,
+      "loss": 0.8796,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018412291933418695,
+      "loss": 0.987,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018409090909090909,
+      "loss": 0.8179,
+      "step": 499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018405889884763125,
+      "loss": 0.9324,
+      "step": 500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001840268886043534,
+      "loss": 0.7424,
+      "step": 501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018399487836107557,
+      "loss": 0.7985,
+      "step": 502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001839628681177977,
+      "loss": 0.7645,
+      "step": 503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018393085787451986,
+      "loss": 0.8283,
+      "step": 504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018389884763124202,
+      "loss": 0.9172,
+      "step": 505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00018386683738796415,
+      "loss": 0.7786,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001838348271446863,
+      "loss": 0.8414,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018380281690140847,
+      "loss": 0.8636,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001837708066581306,
+      "loss": 0.829,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018373879641485276,
+      "loss": 0.7562,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001837067861715749,
+      "loss": 0.9098,
+      "step": 511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018367477592829708,
+      "loss": 0.8473,
+      "step": 512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018364276568501921,
+      "loss": 0.7459,
+      "step": 513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018361075544174135,
+      "loss": 0.8395,
+      "step": 514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001835787451984635,
+      "loss": 0.7941,
+      "step": 515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018354673495518567,
+      "loss": 0.8841,
+      "step": 516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018351472471190783,
+      "loss": 0.713,
+      "step": 517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00018348271446862996,
+      "loss": 0.7585,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018345070422535212,
+      "loss": 0.8871,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018341869398207428,
+      "loss": 0.7748,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001833866837387964,
+      "loss": 0.874,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018335467349551857,
+      "loss": 0.8121,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018332266325224073,
+      "loss": 0.9474,
+      "step": 523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001832906530089629,
+      "loss": 0.8082,
+      "step": 524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018325864276568502,
+      "loss": 0.7965,
+      "step": 525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018322663252240718,
+      "loss": 0.7245,
+      "step": 526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018319462227912934,
+      "loss": 0.8257,
+      "step": 527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018316261203585148,
+      "loss": 0.7629,
+      "step": 528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018313060179257364,
+      "loss": 0.8382,
+      "step": 529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001830985915492958,
+      "loss": 0.8895,
+      "step": 530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00018306658130601793,
+      "loss": 0.8657,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001830345710627401,
+      "loss": 0.803,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018300256081946222,
+      "loss": 0.8395,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001829705505761844,
+      "loss": 0.8885,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018293854033290654,
+      "loss": 0.9239,
+      "step": 535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018290653008962867,
+      "loss": 0.8181,
+      "step": 536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018287451984635083,
+      "loss": 0.8911,
+      "step": 537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000182842509603073,
+      "loss": 0.9071,
+      "step": 538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018281049935979515,
+      "loss": 0.7655,
+      "step": 539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018277848911651728,
+      "loss": 0.8869,
+      "step": 540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018274647887323944,
+      "loss": 0.7825,
+      "step": 541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001827144686299616,
+      "loss": 0.8369,
+      "step": 542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00018268245838668374,
+      "loss": 0.8596,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001826504481434059,
+      "loss": 0.7925,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018261843790012806,
+      "loss": 0.9609,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018258642765685022,
+      "loss": 0.8913,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018255441741357235,
+      "loss": 0.7056,
+      "step": 547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001825224071702945,
+      "loss": 0.9277,
+      "step": 548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018249039692701667,
+      "loss": 0.8671,
+      "step": 549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001824583866837388,
+      "loss": 0.7622,
+      "step": 550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018242637644046096,
+      "loss": 0.8714,
+      "step": 551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018239436619718312,
+      "loss": 0.7377,
+      "step": 552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018236235595390525,
+      "loss": 0.7515,
+      "step": 553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001823303457106274,
+      "loss": 0.8823,
+      "step": 554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018229833546734955,
+      "loss": 0.8167,
+      "step": 555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00018226632522407173,
+      "loss": 0.9988,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018223431498079387,
+      "loss": 0.94,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000182202304737516,
+      "loss": 0.8409,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018217029449423816,
+      "loss": 0.7887,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018213828425096032,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018210627400768248,
+      "loss": 0.8411,
+      "step": 561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001820742637644046,
+      "loss": 0.8791,
+      "step": 562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018204225352112677,
+      "loss": 0.8165,
+      "step": 563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018201024327784893,
+      "loss": 0.9362,
+      "step": 564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018197823303457106,
+      "loss": 0.8369,
+      "step": 565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018194622279129322,
+      "loss": 0.8853,
+      "step": 566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018191421254801538,
+      "loss": 0.8202,
+      "step": 567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00018188220230473751,
+      "loss": 0.7873,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018185019206145967,
+      "loss": 0.8558,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018181818181818183,
+      "loss": 0.8611,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000181786171574904,
+      "loss": 0.9649,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018175416133162613,
+      "loss": 0.8515,
+      "step": 572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018172215108834826,
+      "loss": 0.9032,
+      "step": 573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018169014084507045,
+      "loss": 0.8443,
+      "step": 574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018165813060179258,
+      "loss": 0.784,
+      "step": 575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018162612035851474,
+      "loss": 0.8274,
+      "step": 576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018159411011523687,
+      "loss": 0.7932,
+      "step": 577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018156209987195906,
+      "loss": 0.7656,
+      "step": 578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001815300896286812,
+      "loss": 0.8199,
+      "step": 579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018149807938540332,
+      "loss": 0.8427,
+      "step": 580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00018146606914212548,
+      "loss": 0.8398,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018143405889884764,
+      "loss": 0.7899,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001814020486555698,
+      "loss": 0.8586,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018137003841229194,
+      "loss": 0.7978,
+      "step": 584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001813380281690141,
+      "loss": 0.8531,
+      "step": 585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018130601792573626,
+      "loss": 0.8284,
+      "step": 586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001812740076824584,
+      "loss": 0.9183,
+      "step": 587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018124199743918055,
+      "loss": 0.6815,
+      "step": 588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001812099871959027,
+      "loss": 0.7949,
+      "step": 589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018117797695262484,
+      "loss": 0.8758,
+      "step": 590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000181145966709347,
+      "loss": 0.8122,
+      "step": 591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018111395646606913,
+      "loss": 0.9532,
+      "step": 592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00018108194622279132,
+      "loss": 0.7524,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018104993597951345,
+      "loss": 0.827,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018101792573623558,
+      "loss": 0.7709,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018098591549295774,
+      "loss": 0.8297,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001809539052496799,
+      "loss": 0.8368,
+      "step": 597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018092189500640206,
+      "loss": 0.8493,
+      "step": 598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001808898847631242,
+      "loss": 0.7531,
+      "step": 599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018085787451984636,
+      "loss": 0.9589,
+      "step": 600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018082586427656852,
+      "loss": 0.8053,
+      "step": 601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018079385403329065,
+      "loss": 0.9307,
+      "step": 602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001807618437900128,
+      "loss": 0.7764,
+      "step": 603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018072983354673497,
+      "loss": 0.7977,
+      "step": 604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018069782330345713,
+      "loss": 0.745,
+      "step": 605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00018066581306017926,
+      "loss": 0.9461,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018063380281690142,
+      "loss": 0.8356,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018060179257362358,
+      "loss": 0.8973,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001805697823303457,
+      "loss": 0.9058,
+      "step": 609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018053777208706787,
+      "loss": 0.8199,
+      "step": 610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018050576184379003,
+      "loss": 0.9249,
+      "step": 611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018047375160051217,
+      "loss": 0.7857,
+      "step": 612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018044174135723433,
+      "loss": 0.8156,
+      "step": 613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018040973111395646,
+      "loss": 0.6936,
+      "step": 614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018037772087067865,
+      "loss": 0.8705,
+      "step": 615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018034571062740078,
+      "loss": 0.9653,
+      "step": 616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001803137003841229,
+      "loss": 0.8526,
+      "step": 617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00018028169014084507,
+      "loss": 0.8326,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018024967989756723,
+      "loss": 0.8805,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001802176696542894,
+      "loss": 0.8862,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018018565941101152,
+      "loss": 0.8578,
+      "step": 621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018015364916773368,
+      "loss": 0.8099,
+      "step": 622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018012163892445584,
+      "loss": 0.8893,
+      "step": 623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018008962868117797,
+      "loss": 0.8373,
+      "step": 624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018005761843790013,
+      "loss": 0.8194,
+      "step": 625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001800256081946223,
+      "loss": 0.879,
+      "step": 626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017999359795134443,
+      "loss": 1.058,
+      "step": 627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001799615877080666,
+      "loss": 0.9124,
+      "step": 628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017992957746478875,
+      "loss": 0.8791,
+      "step": 629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001798975672215109,
+      "loss": 0.8692,
+      "step": 630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00017986555697823304,
+      "loss": 0.8888,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001798335467349552,
+      "loss": 0.9036,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017980153649167736,
+      "loss": 0.9481,
+      "step": 633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001797695262483995,
+      "loss": 0.9276,
+      "step": 634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017973751600512165,
+      "loss": 0.7706,
+      "step": 635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017970550576184378,
+      "loss": 0.9047,
+      "step": 636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017967349551856597,
+      "loss": 0.9271,
+      "step": 637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001796414852752881,
+      "loss": 0.9238,
+      "step": 638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017960947503201024,
+      "loss": 0.8489,
+      "step": 639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001795774647887324,
+      "loss": 0.8193,
+      "step": 640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017954545454545456,
+      "loss": 0.7161,
+      "step": 641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017951344430217672,
+      "loss": 0.7997,
+      "step": 642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00017948143405889885,
+      "loss": 0.9374,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000179449423815621,
+      "loss": 0.8747,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017941741357234317,
+      "loss": 0.9518,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001793854033290653,
+      "loss": 0.7346,
+      "step": 646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017935339308578746,
+      "loss": 0.7847,
+      "step": 647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017932138284250962,
+      "loss": 0.8505,
+      "step": 648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017928937259923175,
+      "loss": 0.7663,
+      "step": 649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001792573623559539,
+      "loss": 0.7691,
+      "step": 650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017922535211267607,
+      "loss": 0.9292,
+      "step": 651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017919334186939823,
+      "loss": 0.8677,
+      "step": 652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017916133162612036,
+      "loss": 0.7725,
+      "step": 653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001791293213828425,
+      "loss": 0.8182,
+      "step": 654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017909731113956468,
+      "loss": 0.7939,
+      "step": 655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00017906530089628682,
+      "loss": 0.8192,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017903329065300898,
+      "loss": 0.773,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001790012804097311,
+      "loss": 0.8604,
+      "step": 658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001789692701664533,
+      "loss": 0.8956,
+      "step": 659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017893725992317543,
+      "loss": 0.8061,
+      "step": 660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017890524967989756,
+      "loss": 0.8825,
+      "step": 661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017887323943661972,
+      "loss": 0.8392,
+      "step": 662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017884122919334188,
+      "loss": 0.7899,
+      "step": 663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017880921895006404,
+      "loss": 0.7923,
+      "step": 664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017877720870678617,
+      "loss": 0.8045,
+      "step": 665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017874519846350833,
+      "loss": 0.8158,
+      "step": 666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001787131882202305,
+      "loss": 0.8124,
+      "step": 667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00017868117797695263,
+      "loss": 0.7424,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017864916773367479,
+      "loss": 0.8094,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017861715749039695,
+      "loss": 0.7541,
+      "step": 670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017858514724711908,
+      "loss": 0.775,
+      "step": 671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017855313700384124,
+      "loss": 0.8262,
+      "step": 672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001785211267605634,
+      "loss": 0.8062,
+      "step": 673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017848911651728556,
+      "loss": 0.8069,
+      "step": 674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001784571062740077,
+      "loss": 0.8423,
+      "step": 675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017842509603072982,
+      "loss": 0.7926,
+      "step": 676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000178393085787452,
+      "loss": 0.6658,
+      "step": 677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017836107554417414,
+      "loss": 0.7671,
+      "step": 678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001783290653008963,
+      "loss": 0.781,
+      "step": 679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00017829705505761843,
+      "loss": 0.8081,
+      "step": 680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001782650448143406,
+      "loss": 0.9887,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017823303457106275,
+      "loss": 0.9639,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001782010243277849,
+      "loss": 0.8063,
+      "step": 683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017816901408450705,
+      "loss": 0.827,
+      "step": 684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001781370038412292,
+      "loss": 0.8054,
+      "step": 685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017810499359795137,
+      "loss": 0.8743,
+      "step": 686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001780729833546735,
+      "loss": 0.8617,
+      "step": 687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017804097311139566,
+      "loss": 0.7558,
+      "step": 688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017800896286811782,
+      "loss": 0.8937,
+      "step": 689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017797695262483995,
+      "loss": 0.8778,
+      "step": 690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001779449423815621,
+      "loss": 0.8409,
+      "step": 691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00017791293213828427,
+      "loss": 0.7335,
+      "step": 692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001778809218950064,
+      "loss": 0.8858,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017784891165172856,
+      "loss": 0.8102,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001778169014084507,
+      "loss": 0.9256,
+      "step": 695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017778489116517288,
+      "loss": 0.8711,
+      "step": 696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017775288092189502,
+      "loss": 0.9271,
+      "step": 697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017772087067861715,
+      "loss": 0.8565,
+      "step": 698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001776888604353393,
+      "loss": 0.8492,
+      "step": 699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017765685019206147,
+      "loss": 0.8809,
+      "step": 700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017762483994878363,
+      "loss": 0.8059,
+      "step": 701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017759282970550576,
+      "loss": 0.8576,
+      "step": 702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017756081946222792,
+      "loss": 0.9853,
+      "step": 703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017752880921895008,
+      "loss": 0.787,
+      "step": 704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001774967989756722,
+      "loss": 0.8237,
+      "step": 705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00017746478873239437,
+      "loss": 0.8674,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017743277848911653,
+      "loss": 0.8336,
+      "step": 707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017740076824583866,
+      "loss": 0.8832,
+      "step": 708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017736875800256082,
+      "loss": 0.7301,
+      "step": 709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017733674775928298,
+      "loss": 0.7894,
+      "step": 710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017730473751600514,
+      "loss": 0.769,
+      "step": 711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017727272727272728,
+      "loss": 0.7709,
+      "step": 712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001772407170294494,
+      "loss": 0.9674,
+      "step": 713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001772087067861716,
+      "loss": 0.7916,
+      "step": 714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017717669654289373,
+      "loss": 0.7017,
+      "step": 715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001771446862996159,
+      "loss": 0.7493,
+      "step": 716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00017711267605633802,
+      "loss": 0.8,
+      "step": 717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001770806658130602,
+      "loss": 0.776,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017704865556978234,
+      "loss": 1.0176,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017701664532650447,
+      "loss": 0.7593,
+      "step": 720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017698463508322663,
+      "loss": 0.8618,
+      "step": 721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001769526248399488,
+      "loss": 0.7726,
+      "step": 722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017692061459667095,
+      "loss": 0.8271,
+      "step": 723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017688860435339309,
+      "loss": 0.8466,
+      "step": 724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017685659411011525,
+      "loss": 0.789,
+      "step": 725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001768245838668374,
+      "loss": 0.8271,
+      "step": 726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017679257362355954,
+      "loss": 0.7825,
+      "step": 727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001767605633802817,
+      "loss": 0.7793,
+      "step": 728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017672855313700386,
+      "loss": 0.8455,
+      "step": 729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000176696542893726,
+      "loss": 0.8552,
+      "step": 730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00017666453265044815,
+      "loss": 0.8997,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001766325224071703,
+      "loss": 0.8064,
+      "step": 732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017660051216389247,
+      "loss": 0.7565,
+      "step": 733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001765685019206146,
+      "loss": 0.9292,
+      "step": 734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017653649167733674,
+      "loss": 0.8543,
+      "step": 735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017650448143405892,
+      "loss": 0.6477,
+      "step": 736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017647247119078105,
+      "loss": 0.8989,
+      "step": 737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017644046094750321,
+      "loss": 0.8051,
+      "step": 738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017640845070422535,
+      "loss": 0.8427,
+      "step": 739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001763764404609475,
+      "loss": 0.8539,
+      "step": 740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017634443021766967,
+      "loss": 0.8112,
+      "step": 741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001763124199743918,
+      "loss": 0.8282,
+      "step": 742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00017628040973111396,
+      "loss": 0.7741,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017624839948783612,
+      "loss": 0.8012,
+      "step": 744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017621638924455828,
+      "loss": 0.8223,
+      "step": 745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001761843790012804,
+      "loss": 0.9167,
+      "step": 746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017615236875800257,
+      "loss": 0.8001,
+      "step": 747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017612035851472473,
+      "loss": 0.8137,
+      "step": 748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017608834827144686,
+      "loss": 0.8057,
+      "step": 749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017605633802816902,
+      "loss": 0.8505,
+      "step": 750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017602432778489118,
+      "loss": 0.8453,
+      "step": 751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017599231754161332,
+      "loss": 0.9007,
+      "step": 752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017596030729833548,
+      "loss": 0.734,
+      "step": 753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017592829705505764,
+      "loss": 0.8057,
+      "step": 754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001758962868117798,
+      "loss": 0.8516,
+      "step": 755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00017586427656850193,
+      "loss": 0.8686,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017583226632522406,
+      "loss": 0.812,
+      "step": 757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017580025608194625,
+      "loss": 0.8649,
+      "step": 758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017576824583866838,
+      "loss": 0.8142,
+      "step": 759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017573623559539054,
+      "loss": 0.861,
+      "step": 760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017570422535211267,
+      "loss": 0.8238,
+      "step": 761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017567221510883483,
+      "loss": 0.8481,
+      "step": 762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000175640204865557,
+      "loss": 0.7788,
+      "step": 763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017560819462227912,
+      "loss": 1.0263,
+      "step": 764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017557618437900128,
+      "loss": 0.9379,
+      "step": 765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017554417413572344,
+      "loss": 0.8304,
+      "step": 766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017551216389244558,
+      "loss": 0.8311,
+      "step": 767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00017548015364916774,
+      "loss": 0.8748,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001754481434058899,
+      "loss": 0.7966,
+      "step": 769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017541613316261206,
+      "loss": 0.8614,
+      "step": 770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001753841229193342,
+      "loss": 0.7993,
+      "step": 771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017535211267605635,
+      "loss": 0.8851,
+      "step": 772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001753201024327785,
+      "loss": 0.7351,
+      "step": 773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017528809218950064,
+      "loss": 0.7081,
+      "step": 774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001752560819462228,
+      "loss": 0.8082,
+      "step": 775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017522407170294493,
+      "loss": 0.7792,
+      "step": 776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017519206145966712,
+      "loss": 0.9288,
+      "step": 777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017516005121638925,
+      "loss": 0.7971,
+      "step": 778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017512804097311139,
+      "loss": 0.9193,
+      "step": 779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00017509603072983357,
+      "loss": 0.899,
+      "step": 780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001750640204865557,
+      "loss": 0.7891,
+      "step": 781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017503201024327787,
+      "loss": 0.8288,
+      "step": 782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000175,
+      "loss": 0.8687,
+      "step": 783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017496798975672216,
+      "loss": 0.8708,
+      "step": 784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017493597951344432,
+      "loss": 0.833,
+      "step": 785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017490396927016645,
+      "loss": 0.9007,
+      "step": 786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001748719590268886,
+      "loss": 0.7291,
+      "step": 787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017483994878361077,
+      "loss": 0.8407,
+      "step": 788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001748079385403329,
+      "loss": 0.7835,
+      "step": 789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017477592829705506,
+      "loss": 0.8138,
+      "step": 790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017474391805377722,
+      "loss": 0.6527,
+      "step": 791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017471190781049938,
+      "loss": 0.7952,
+      "step": 792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00017467989756722151,
+      "loss": 0.7749,
+      "step": 793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017464788732394365,
+      "loss": 0.7828,
+      "step": 794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017461587708066583,
+      "loss": 0.8516,
+      "step": 795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017458386683738797,
+      "loss": 0.9804,
+      "step": 796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017455185659411013,
+      "loss": 0.7898,
+      "step": 797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017451984635083226,
+      "loss": 0.7455,
+      "step": 798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017448783610755445,
+      "loss": 0.9052,
+      "step": 799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017445582586427658,
+      "loss": 0.9169,
+      "step": 800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001744238156209987,
+      "loss": 0.892,
+      "step": 801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017439180537772087,
+      "loss": 0.7793,
+      "step": 802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017435979513444303,
+      "loss": 0.8782,
+      "step": 803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001743277848911652,
+      "loss": 0.7318,
+      "step": 804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017429577464788732,
+      "loss": 0.8258,
+      "step": 805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00017426376440460948,
+      "loss": 0.8326,
+      "step": 806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017423175416133164,
+      "loss": 0.8562,
+      "step": 807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017419974391805378,
+      "loss": 0.8148,
+      "step": 808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017416773367477594,
+      "loss": 1.0542,
+      "step": 809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001741357234314981,
+      "loss": 0.7881,
+      "step": 810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017410371318822023,
+      "loss": 0.8953,
+      "step": 811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001740717029449424,
+      "loss": 0.7827,
+      "step": 812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017403969270166455,
+      "loss": 0.846,
+      "step": 813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001740076824583867,
+      "loss": 0.7435,
+      "step": 814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017397567221510884,
+      "loss": 0.7462,
+      "step": 815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017394366197183097,
+      "loss": 0.7516,
+      "step": 816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00017391165172855316,
+      "loss": 0.8499,
+      "step": 817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001738796414852753,
+      "loss": 0.8667,
+      "step": 818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017384763124199745,
+      "loss": 0.8456,
+      "step": 819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017381562099871959,
+      "loss": 0.6986,
+      "step": 820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017378361075544174,
+      "loss": 0.9566,
+      "step": 821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001737516005121639,
+      "loss": 0.8834,
+      "step": 822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017371959026888604,
+      "loss": 0.7587,
+      "step": 823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001736875800256082,
+      "loss": 0.8363,
+      "step": 824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017365556978233036,
+      "loss": 0.8066,
+      "step": 825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017362355953905252,
+      "loss": 0.8791,
+      "step": 826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017359154929577465,
+      "loss": 0.9008,
+      "step": 827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001735595390524968,
+      "loss": 0.8289,
+      "step": 828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017352752880921897,
+      "loss": 0.8363,
+      "step": 829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001734955185659411,
+      "loss": 0.8517,
+      "step": 830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00017346350832266326,
+      "loss": 0.8568,
+      "step": 831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017343149807938542,
+      "loss": 0.7404,
+      "step": 832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017339948783610755,
+      "loss": 0.8624,
+      "step": 833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017336747759282971,
+      "loss": 0.8341,
+      "step": 834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017333546734955187,
+      "loss": 0.7741,
+      "step": 835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017330345710627403,
+      "loss": 0.875,
+      "step": 836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017327144686299617,
+      "loss": 0.8368,
+      "step": 837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001732394366197183,
+      "loss": 0.7684,
+      "step": 838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017320742637644049,
+      "loss": 0.8006,
+      "step": 839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017317541613316262,
+      "loss": 0.891,
+      "step": 840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017314340588988478,
+      "loss": 0.8282,
+      "step": 841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001731113956466069,
+      "loss": 0.7962,
+      "step": 842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00017307938540332907,
+      "loss": 0.9345,
+      "step": 843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017304737516005123,
+      "loss": 0.8022,
+      "step": 844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017301536491677336,
+      "loss": 0.735,
+      "step": 845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017298335467349552,
+      "loss": 0.9587,
+      "step": 846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017295134443021768,
+      "loss": 0.8252,
+      "step": 847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017291933418693982,
+      "loss": 0.7618,
+      "step": 848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017288732394366197,
+      "loss": 0.8104,
+      "step": 849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017285531370038413,
+      "loss": 0.8187,
+      "step": 850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001728233034571063,
+      "loss": 0.8169,
+      "step": 851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017279129321382843,
+      "loss": 0.8646,
+      "step": 852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001727592829705506,
+      "loss": 0.8413,
+      "step": 853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017272727272727275,
+      "loss": 0.7767,
+      "step": 854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017269526248399488,
+      "loss": 0.7317,
+      "step": 855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00017266325224071704,
+      "loss": 0.7594,
+      "step": 856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001726312419974392,
+      "loss": 0.9051,
+      "step": 857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017259923175416136,
+      "loss": 0.7638,
+      "step": 858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001725672215108835,
+      "loss": 0.8661,
+      "step": 859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017253521126760562,
+      "loss": 0.8428,
+      "step": 860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001725032010243278,
+      "loss": 0.8634,
+      "step": 861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017247119078104994,
+      "loss": 0.8425,
+      "step": 862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001724391805377721,
+      "loss": 0.7275,
+      "step": 863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017240717029449424,
+      "loss": 0.9025,
+      "step": 864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001723751600512164,
+      "loss": 0.9377,
+      "step": 865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017234314980793856,
+      "loss": 0.8386,
+      "step": 866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001723111395646607,
+      "loss": 0.8123,
+      "step": 867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00017227912932138285,
+      "loss": 0.7319,
+      "step": 868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000172247119078105,
+      "loss": 0.7258,
+      "step": 869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017221510883482714,
+      "loss": 0.8995,
+      "step": 870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001721830985915493,
+      "loss": 0.7907,
+      "step": 871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017215108834827146,
+      "loss": 0.7364,
+      "step": 872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017211907810499362,
+      "loss": 0.7806,
+      "step": 873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017208706786171575,
+      "loss": 0.9809,
+      "step": 874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017205505761843789,
+      "loss": 0.8114,
+      "step": 875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017202304737516007,
+      "loss": 0.7692,
+      "step": 876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001719910371318822,
+      "loss": 0.9418,
+      "step": 877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017195902688860436,
+      "loss": 0.8379,
+      "step": 878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001719270166453265,
+      "loss": 0.8091,
+      "step": 879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017189500640204866,
+      "loss": 0.7287,
+      "step": 880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00017186299615877082,
+      "loss": 0.8372,
+      "step": 881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017183098591549295,
+      "loss": 0.8117,
+      "step": 882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017179897567221514,
+      "loss": 0.9157,
+      "step": 883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017176696542893727,
+      "loss": 0.9549,
+      "step": 884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017173495518565943,
+      "loss": 0.7739,
+      "step": 885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017170294494238156,
+      "loss": 0.7775,
+      "step": 886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017167093469910372,
+      "loss": 0.7471,
+      "step": 887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017163892445582588,
+      "loss": 0.7691,
+      "step": 888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017160691421254801,
+      "loss": 0.9197,
+      "step": 889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017157490396927017,
+      "loss": 0.9203,
+      "step": 890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017154289372599233,
+      "loss": 0.8488,
+      "step": 891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017151088348271447,
+      "loss": 0.7674,
+      "step": 892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00017147887323943663,
+      "loss": 0.8455,
+      "step": 893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017144686299615879,
+      "loss": 0.9019,
+      "step": 894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017141485275288095,
+      "loss": 0.8239,
+      "step": 895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017138284250960308,
+      "loss": 0.8681,
+      "step": 896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001713508322663252,
+      "loss": 0.7382,
+      "step": 897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001713188220230474,
+      "loss": 0.8118,
+      "step": 898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017128681177976953,
+      "loss": 0.9801,
+      "step": 899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001712548015364917,
+      "loss": 0.8616,
+      "step": 900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017122279129321382,
+      "loss": 0.8429,
+      "step": 901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017119078104993598,
+      "loss": 0.8818,
+      "step": 902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017115877080665814,
+      "loss": 0.8268,
+      "step": 903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017112676056338028,
+      "loss": 0.9014,
+      "step": 904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00017109475032010243,
+      "loss": 0.8034,
+      "step": 905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001710627400768246,
+      "loss": 0.8534,
+      "step": 906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017103072983354673,
+      "loss": 0.8832,
+      "step": 907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001709987195902689,
+      "loss": 0.7666,
+      "step": 908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017096670934699105,
+      "loss": 0.9292,
+      "step": 909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001709346991037132,
+      "loss": 0.8442,
+      "step": 910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017090268886043534,
+      "loss": 0.8922,
+      "step": 911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001708706786171575,
+      "loss": 0.9415,
+      "step": 912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017083866837387966,
+      "loss": 0.727,
+      "step": 913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001708066581306018,
+      "loss": 0.9026,
+      "step": 914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017077464788732395,
+      "loss": 0.8363,
+      "step": 915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001707426376440461,
+      "loss": 0.7485,
+      "step": 916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017071062740076827,
+      "loss": 0.8095,
+      "step": 917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001706786171574904,
+      "loss": 0.8737,
+      "step": 918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017064660691421254,
+      "loss": 0.8117,
+      "step": 919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017061459667093472,
+      "loss": 0.8381,
+      "step": 920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017058258642765686,
+      "loss": 0.8991,
+      "step": 921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017055057618437902,
+      "loss": 0.9253,
+      "step": 922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017051856594110115,
+      "loss": 0.9446,
+      "step": 923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001704865556978233,
+      "loss": 0.7891,
+      "step": 924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017045454545454547,
+      "loss": 0.9345,
+      "step": 925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001704225352112676,
+      "loss": 0.7159,
+      "step": 926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017039052496798976,
+      "loss": 0.8231,
+      "step": 927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017035851472471192,
+      "loss": 0.6187,
+      "step": 928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017032650448143405,
+      "loss": 0.8892,
+      "step": 929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001702944942381562,
+      "loss": 0.8652,
+      "step": 930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017026248399487837,
+      "loss": 0.6449,
+      "step": 931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017023047375160053,
+      "loss": 0.8845,
+      "step": 932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017019846350832266,
+      "loss": 0.7748,
+      "step": 933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017016645326504482,
+      "loss": 0.7577,
+      "step": 934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017013444302176698,
+      "loss": 0.6506,
+      "step": 935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017010243277848912,
+      "loss": 0.8713,
+      "step": 936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017007042253521128,
+      "loss": 1.0087,
+      "step": 937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00017003841229193344,
+      "loss": 0.7896,
+      "step": 938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001700064020486556,
+      "loss": 0.8297,
+      "step": 939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016997439180537773,
+      "loss": 0.9284,
+      "step": 940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016994238156209986,
+      "loss": 0.814,
+      "step": 941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016991037131882205,
+      "loss": 0.8792,
+      "step": 942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016987836107554418,
+      "loss": 0.8084,
+      "step": 943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016984635083226634,
+      "loss": 0.8307,
+      "step": 944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016981434058898847,
+      "loss": 0.7677,
+      "step": 945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016978233034571063,
+      "loss": 0.6914,
+      "step": 946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001697503201024328,
+      "loss": 0.9187,
+      "step": 947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016971830985915493,
+      "loss": 0.8614,
+      "step": 948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016968629961587709,
+      "loss": 0.7816,
+      "step": 949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016965428937259925,
+      "loss": 0.6937,
+      "step": 950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016962227912932138,
+      "loss": 0.8123,
+      "step": 951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016959026888604354,
+      "loss": 0.8092,
+      "step": 952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001695582586427657,
+      "loss": 0.8272,
+      "step": 953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016952624839948786,
+      "loss": 0.7629,
+      "step": 954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016949423815621,
+      "loss": 0.8971,
+      "step": 955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00016946222791293212,
+      "loss": 0.8397,
+      "step": 956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001694302176696543,
+      "loss": 0.9334,
+      "step": 957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016939820742637644,
+      "loss": 0.8623,
+      "step": 958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001693661971830986,
+      "loss": 0.8331,
+      "step": 959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016933418693982076,
+      "loss": 0.7947,
+      "step": 960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001693021766965429,
+      "loss": 0.8912,
+      "step": 961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016927016645326505,
+      "loss": 0.8452,
+      "step": 962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001692381562099872,
+      "loss": 0.937,
+      "step": 963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016920614596670937,
+      "loss": 0.8295,
+      "step": 964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001691741357234315,
+      "loss": 0.8396,
+      "step": 965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016914212548015367,
+      "loss": 0.8114,
+      "step": 966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001691101152368758,
+      "loss": 0.9572,
+      "step": 967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00016907810499359796,
+      "loss": 0.9657,
+      "step": 968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016904609475032012,
+      "loss": 0.8125,
+      "step": 969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016901408450704225,
+      "loss": 0.7858,
+      "step": 970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001689820742637644,
+      "loss": 0.7949,
+      "step": 971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016895006402048657,
+      "loss": 0.9038,
+      "step": 972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001689180537772087,
+      "loss": 0.8575,
+      "step": 973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016888604353393086,
+      "loss": 0.9311,
+      "step": 974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016885403329065302,
+      "loss": 1.0016,
+      "step": 975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016882202304737518,
+      "loss": 0.8346,
+      "step": 976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016879001280409732,
+      "loss": 0.7306,
+      "step": 977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016875800256081945,
+      "loss": 0.8549,
+      "step": 978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016872599231754164,
+      "loss": 0.7818,
+      "step": 979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016869398207426377,
+      "loss": 0.7568,
+      "step": 980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00016866197183098593,
+      "loss": 0.8288,
+      "step": 981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016862996158770806,
+      "loss": 0.8381,
+      "step": 982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016859795134443022,
+      "loss": 0.88,
+      "step": 983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016856594110115238,
+      "loss": 0.8837,
+      "step": 984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001685339308578745,
+      "loss": 0.8789,
+      "step": 985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001685019206145967,
+      "loss": 1.0149,
+      "step": 986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016846991037131883,
+      "loss": 0.8407,
+      "step": 987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016843790012804097,
+      "loss": 0.7562,
+      "step": 988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016840588988476312,
+      "loss": 0.8392,
+      "step": 989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016837387964148528,
+      "loss": 0.839,
+      "step": 990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016834186939820744,
+      "loss": 0.8957,
+      "step": 991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016830985915492958,
+      "loss": 0.8003,
+      "step": 992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00016827784891165174,
+      "loss": 0.8817,
+      "step": 993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001682458386683739,
+      "loss": 0.7596,
+      "step": 994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016821382842509603,
+      "loss": 0.8686,
+      "step": 995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001681818181818182,
+      "loss": 0.8168,
+      "step": 996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016814980793854035,
+      "loss": 0.8803,
+      "step": 997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001681177976952625,
+      "loss": 0.8371,
+      "step": 998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016808578745198464,
+      "loss": 0.7736,
+      "step": 999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016805377720870677,
+      "loss": 0.8171,
+      "step": 1000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016802176696542896,
+      "loss": 0.9794,
+      "step": 1001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001679897567221511,
+      "loss": 0.8088,
+      "step": 1002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016795774647887325,
+      "loss": 0.7904,
+      "step": 1003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016792573623559539,
+      "loss": 0.6922,
+      "step": 1004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00016789372599231755,
+      "loss": 0.894,
+      "step": 1005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001678617157490397,
+      "loss": 0.7121,
+      "step": 1006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016782970550576184,
+      "loss": 0.7592,
+      "step": 1007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.000167797695262484,
+      "loss": 0.8724,
+      "step": 1008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016776568501920616,
+      "loss": 0.7687,
+      "step": 1009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001677336747759283,
+      "loss": 0.8957,
+      "step": 1010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016770166453265045,
+      "loss": 0.883,
+      "step": 1011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001676696542893726,
+      "loss": 0.7848,
+      "step": 1012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016763764404609477,
+      "loss": 0.9225,
+      "step": 1013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001676056338028169,
+      "loss": 0.8078,
+      "step": 1014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016757362355953906,
+      "loss": 0.8168,
+      "step": 1015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016754161331626122,
+      "loss": 0.7901,
+      "step": 1016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016750960307298336,
+      "loss": 0.8682,
+      "step": 1017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00016747759282970551,
+      "loss": 0.973,
+      "step": 1018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016744558258642767,
+      "loss": 0.757,
+      "step": 1019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001674135723431498,
+      "loss": 0.871,
+      "step": 1020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016738156209987197,
+      "loss": 0.7839,
+      "step": 1021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001673495518565941,
+      "loss": 0.8795,
+      "step": 1022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001673175416133163,
+      "loss": 0.9069,
+      "step": 1023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016728553137003842,
+      "loss": 0.8072,
+      "step": 1024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016725352112676058,
+      "loss": 0.7672,
+      "step": 1025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001672215108834827,
+      "loss": 0.8948,
+      "step": 1026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016718950064020487,
+      "loss": 0.9456,
+      "step": 1027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016715749039692703,
+      "loss": 0.7235,
+      "step": 1028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016712548015364916,
+      "loss": 1.0129,
+      "step": 1029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016709346991037132,
+      "loss": 0.8455,
+      "step": 1030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00016706145966709348,
+      "loss": 0.8691,
+      "step": 1031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016702944942381562,
+      "loss": 0.8706,
+      "step": 1032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016699743918053778,
+      "loss": 0.9251,
+      "step": 1033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016696542893725994,
+      "loss": 0.739,
+      "step": 1034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001669334186939821,
+      "loss": 0.87,
+      "step": 1035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016690140845070423,
+      "loss": 0.7144,
+      "step": 1036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001668693982074264,
+      "loss": 0.7677,
+      "step": 1037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016683738796414855,
+      "loss": 0.8331,
+      "step": 1038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016680537772087068,
+      "loss": 0.983,
+      "step": 1039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016677336747759284,
+      "loss": 0.7403,
+      "step": 1040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000166741357234315,
+      "loss": 0.8796,
+      "step": 1041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00016670934699103713,
+      "loss": 0.7391,
+      "step": 1042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001666773367477593,
+      "loss": 0.8915,
+      "step": 1043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016664532650448143,
+      "loss": 0.7894,
+      "step": 1044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001666133162612036,
+      "loss": 0.8159,
+      "step": 1045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016658130601792574,
+      "loss": 0.7515,
+      "step": 1046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016654929577464788,
+      "loss": 0.7361,
+      "step": 1047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016651728553137004,
+      "loss": 0.7795,
+      "step": 1048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001664852752880922,
+      "loss": 0.7818,
+      "step": 1049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016645326504481436,
+      "loss": 0.7256,
+      "step": 1050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001664212548015365,
+      "loss": 0.79,
+      "step": 1051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016638924455825865,
+      "loss": 0.7785,
+      "step": 1052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001663572343149808,
+      "loss": 0.7881,
+      "step": 1053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016632522407170294,
+      "loss": 0.8496,
+      "step": 1054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001662932138284251,
+      "loss": 0.93,
+      "step": 1055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00016626120358514726,
+      "loss": 0.9037,
+      "step": 1056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016622919334186942,
+      "loss": 0.8355,
+      "step": 1057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016619718309859155,
+      "loss": 0.7729,
+      "step": 1058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001661651728553137,
+      "loss": 0.845,
+      "step": 1059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016613316261203587,
+      "loss": 0.8216,
+      "step": 1060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.000166101152368758,
+      "loss": 0.917,
+      "step": 1061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016606914212548017,
+      "loss": 0.8077,
+      "step": 1062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016603713188220233,
+      "loss": 0.8584,
+      "step": 1063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016600512163892446,
+      "loss": 0.765,
+      "step": 1064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016597311139564662,
+      "loss": 0.7088,
+      "step": 1065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016594110115236875,
+      "loss": 0.8962,
+      "step": 1066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016590909090909094,
+      "loss": 0.8149,
+      "step": 1067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00016587708066581307,
+      "loss": 0.8342,
+      "step": 1068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001658450704225352,
+      "loss": 0.9165,
+      "step": 1069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016581306017925736,
+      "loss": 0.8714,
+      "step": 1070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016578104993597952,
+      "loss": 0.9016,
+      "step": 1071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016574903969270168,
+      "loss": 0.9178,
+      "step": 1072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016571702944942382,
+      "loss": 0.8045,
+      "step": 1073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016568501920614597,
+      "loss": 0.8108,
+      "step": 1074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016565300896286813,
+      "loss": 0.7894,
+      "step": 1075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016562099871959027,
+      "loss": 0.7714,
+      "step": 1076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016558898847631243,
+      "loss": 0.7615,
+      "step": 1077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001655569782330346,
+      "loss": 0.9111,
+      "step": 1078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016552496798975675,
+      "loss": 0.6991,
+      "step": 1079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00016549295774647888,
+      "loss": 0.8983,
+      "step": 1080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000165460947503201,
+      "loss": 0.8262,
+      "step": 1081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001654289372599232,
+      "loss": 0.8302,
+      "step": 1082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016539692701664533,
+      "loss": 0.9348,
+      "step": 1083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001653649167733675,
+      "loss": 0.8503,
+      "step": 1084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016533290653008962,
+      "loss": 0.8291,
+      "step": 1085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016530089628681178,
+      "loss": 0.7829,
+      "step": 1086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016526888604353394,
+      "loss": 0.7694,
+      "step": 1087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016523687580025608,
+      "loss": 0.8052,
+      "step": 1088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016520486555697826,
+      "loss": 0.8735,
+      "step": 1089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001651728553137004,
+      "loss": 0.8246,
+      "step": 1090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016514084507042253,
+      "loss": 0.8225,
+      "step": 1091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001651088348271447,
+      "loss": 0.8438,
+      "step": 1092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00016507682458386685,
+      "loss": 0.7338,
+      "step": 1093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.000165044814340589,
+      "loss": 0.8122,
+      "step": 1094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016501280409731114,
+      "loss": 0.9026,
+      "step": 1095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001649807938540333,
+      "loss": 0.7971,
+      "step": 1096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016494878361075546,
+      "loss": 0.7791,
+      "step": 1097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001649167733674776,
+      "loss": 0.8555,
+      "step": 1098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016488476312419975,
+      "loss": 0.7929,
+      "step": 1099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001648527528809219,
+      "loss": 0.7231,
+      "step": 1100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016482074263764405,
+      "loss": 0.7696,
+      "step": 1101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001647887323943662,
+      "loss": 0.8284,
+      "step": 1102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016475672215108834,
+      "loss": 0.8297,
+      "step": 1103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016472471190781052,
+      "loss": 0.7617,
+      "step": 1104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016469270166453266,
+      "loss": 0.8579,
+      "step": 1105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00016466069142125482,
+      "loss": 0.8955,
+      "step": 1106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016462868117797695,
+      "loss": 0.8237,
+      "step": 1107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001645966709346991,
+      "loss": 0.8213,
+      "step": 1108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016456466069142127,
+      "loss": 0.81,
+      "step": 1109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001645326504481434,
+      "loss": 1.0071,
+      "step": 1110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016450064020486556,
+      "loss": 0.8321,
+      "step": 1111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016446862996158772,
+      "loss": 0.7857,
+      "step": 1112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016443661971830985,
+      "loss": 0.8592,
+      "step": 1113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016440460947503201,
+      "loss": 0.8426,
+      "step": 1114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016437259923175417,
+      "loss": 0.688,
+      "step": 1115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016434058898847633,
+      "loss": 0.7351,
+      "step": 1116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016430857874519847,
+      "loss": 0.8982,
+      "step": 1117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00016427656850192063,
+      "loss": 0.7535,
+      "step": 1118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016424455825864279,
+      "loss": 0.8489,
+      "step": 1119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016421254801536492,
+      "loss": 0.9313,
+      "step": 1120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016418053777208708,
+      "loss": 0.7996,
+      "step": 1121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016414852752880924,
+      "loss": 0.8349,
+      "step": 1122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016411651728553137,
+      "loss": 0.8766,
+      "step": 1123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016408450704225353,
+      "loss": 0.7425,
+      "step": 1124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016405249679897566,
+      "loss": 0.8088,
+      "step": 1125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016402048655569785,
+      "loss": 0.7724,
+      "step": 1126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016398847631241998,
+      "loss": 0.7568,
+      "step": 1127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016395646606914212,
+      "loss": 0.8736,
+      "step": 1128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016392445582586428,
+      "loss": 0.852,
+      "step": 1129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00016389244558258643,
+      "loss": 0.8514,
+      "step": 1130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001638604353393086,
+      "loss": 0.8352,
+      "step": 1131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016382842509603073,
+      "loss": 0.7726,
+      "step": 1132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001637964148527529,
+      "loss": 0.8638,
+      "step": 1133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016376440460947505,
+      "loss": 0.7749,
+      "step": 1134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016373239436619718,
+      "loss": 0.8643,
+      "step": 1135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016370038412291934,
+      "loss": 0.7193,
+      "step": 1136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001636683738796415,
+      "loss": 0.7939,
+      "step": 1137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016363636363636366,
+      "loss": 0.7662,
+      "step": 1138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001636043533930858,
+      "loss": 0.7581,
+      "step": 1139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016357234314980795,
+      "loss": 0.8214,
+      "step": 1140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001635403329065301,
+      "loss": 0.797,
+      "step": 1141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00016350832266325224,
+      "loss": 0.8902,
+      "step": 1142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001634763124199744,
+      "loss": 0.9522,
+      "step": 1143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016344430217669656,
+      "loss": 0.7986,
+      "step": 1144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001634122919334187,
+      "loss": 0.813,
+      "step": 1145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016338028169014086,
+      "loss": 0.7778,
+      "step": 1146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.000163348271446863,
+      "loss": 0.7929,
+      "step": 1147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016331626120358518,
+      "loss": 0.8318,
+      "step": 1148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001632842509603073,
+      "loss": 0.8964,
+      "step": 1149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016325224071702944,
+      "loss": 0.8236,
+      "step": 1150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001632202304737516,
+      "loss": 0.8395,
+      "step": 1151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016318822023047376,
+      "loss": 0.7612,
+      "step": 1152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016315620998719592,
+      "loss": 0.8671,
+      "step": 1153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016312419974391805,
+      "loss": 0.8454,
+      "step": 1154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001630921895006402,
+      "loss": 0.837,
+      "step": 1155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00016306017925736237,
+      "loss": 0.9328,
+      "step": 1156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001630281690140845,
+      "loss": 0.7665,
+      "step": 1157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016299615877080666,
+      "loss": 0.8216,
+      "step": 1158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016296414852752882,
+      "loss": 0.8768,
+      "step": 1159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016293213828425098,
+      "loss": 0.8369,
+      "step": 1160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016290012804097312,
+      "loss": 0.8003,
+      "step": 1161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016286811779769525,
+      "loss": 0.8642,
+      "step": 1162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016283610755441744,
+      "loss": 0.9401,
+      "step": 1163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016280409731113957,
+      "loss": 0.8097,
+      "step": 1164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016277208706786173,
+      "loss": 0.9433,
+      "step": 1165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001627400768245839,
+      "loss": 0.8477,
+      "step": 1166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016270806658130602,
+      "loss": 0.8546,
+      "step": 1167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00016267605633802818,
+      "loss": 0.7654,
+      "step": 1168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016264404609475031,
+      "loss": 0.8259,
+      "step": 1169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001626120358514725,
+      "loss": 0.8793,
+      "step": 1170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016258002560819463,
+      "loss": 0.8391,
+      "step": 1171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016254801536491677,
+      "loss": 0.8865,
+      "step": 1172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016251600512163893,
+      "loss": 0.7563,
+      "step": 1173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016248399487836109,
+      "loss": 0.8389,
+      "step": 1174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016245198463508325,
+      "loss": 0.7511,
+      "step": 1175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016241997439180538,
+      "loss": 0.8317,
+      "step": 1176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016238796414852754,
+      "loss": 0.7243,
+      "step": 1177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0001623559539052497,
+      "loss": 0.8274,
+      "step": 1178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016232394366197183,
+      "loss": 0.8048,
+      "step": 1179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.000162291933418694,
+      "loss": 0.8419,
+      "step": 1180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00016225992317541615,
+      "loss": 0.8228,
+      "step": 1181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016222791293213828,
+      "loss": 0.7862,
+      "step": 1182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016219590268886044,
+      "loss": 0.78,
+      "step": 1183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016216389244558258,
+      "loss": 0.9222,
+      "step": 1184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016213188220230476,
+      "loss": 1.0484,
+      "step": 1185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001620998719590269,
+      "loss": 0.8025,
+      "step": 1186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016206786171574903,
+      "loss": 0.808,
+      "step": 1187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001620358514724712,
+      "loss": 0.8624,
+      "step": 1188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016200384122919335,
+      "loss": 0.8888,
+      "step": 1189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001619718309859155,
+      "loss": 0.8384,
+      "step": 1190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016193982074263764,
+      "loss": 0.789,
+      "step": 1191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016190781049935983,
+      "loss": 0.841,
+      "step": 1192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00016187580025608196,
+      "loss": 0.895,
+      "step": 1193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001618437900128041,
+      "loss": 0.8373,
+      "step": 1194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016181177976952625,
+      "loss": 0.923,
+      "step": 1195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001617797695262484,
+      "loss": 0.936,
+      "step": 1196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016174775928297057,
+      "loss": 0.8757,
+      "step": 1197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001617157490396927,
+      "loss": 0.8439,
+      "step": 1198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016168373879641486,
+      "loss": 0.8421,
+      "step": 1199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016165172855313702,
+      "loss": 0.8203,
+      "step": 1200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016161971830985916,
+      "loss": 0.9084,
+      "step": 1201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016158770806658132,
+      "loss": 0.8856,
+      "step": 1202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016155569782330348,
+      "loss": 0.8032,
+      "step": 1203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001615236875800256,
+      "loss": 0.7867,
+      "step": 1204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00016149167733674777,
+      "loss": 0.8778,
+      "step": 1205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001614596670934699,
+      "loss": 0.7473,
+      "step": 1206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001614276568501921,
+      "loss": 0.7665,
+      "step": 1207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016139564660691422,
+      "loss": 0.8323,
+      "step": 1208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016136363636363635,
+      "loss": 0.9359,
+      "step": 1209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001613316261203585,
+      "loss": 0.9023,
+      "step": 1210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016129961587708067,
+      "loss": 0.6753,
+      "step": 1211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016126760563380283,
+      "loss": 0.9202,
+      "step": 1212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016123559539052497,
+      "loss": 0.7696,
+      "step": 1213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016120358514724713,
+      "loss": 0.8322,
+      "step": 1214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016117157490396928,
+      "loss": 0.7525,
+      "step": 1215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016113956466069142,
+      "loss": 0.7661,
+      "step": 1216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016110755441741358,
+      "loss": 0.7832,
+      "step": 1217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00016107554417413574,
+      "loss": 0.8472,
+      "step": 1218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001610435339308579,
+      "loss": 0.8848,
+      "step": 1219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016101152368758003,
+      "loss": 0.7783,
+      "step": 1220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001609795134443022,
+      "loss": 0.8594,
+      "step": 1221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016094750320102435,
+      "loss": 0.8967,
+      "step": 1222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016091549295774648,
+      "loss": 0.8321,
+      "step": 1223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016088348271446864,
+      "loss": 0.7883,
+      "step": 1224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001608514724711908,
+      "loss": 0.8831,
+      "step": 1225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016081946222791293,
+      "loss": 0.8175,
+      "step": 1226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001607874519846351,
+      "loss": 0.7739,
+      "step": 1227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016075544174135723,
+      "loss": 0.7986,
+      "step": 1228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001607234314980794,
+      "loss": 0.7464,
+      "step": 1229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016069142125480155,
+      "loss": 0.7166,
+      "step": 1230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00016065941101152368,
+      "loss": 0.8422,
+      "step": 1231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016062740076824584,
+      "loss": 0.7834,
+      "step": 1232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000160595390524968,
+      "loss": 0.7863,
+      "step": 1233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016056338028169016,
+      "loss": 0.8836,
+      "step": 1234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001605313700384123,
+      "loss": 0.8128,
+      "step": 1235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016049935979513445,
+      "loss": 0.7955,
+      "step": 1236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001604673495518566,
+      "loss": 0.9122,
+      "step": 1237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016043533930857874,
+      "loss": 0.7628,
+      "step": 1238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001604033290653009,
+      "loss": 0.7983,
+      "step": 1239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016037131882202306,
+      "loss": 0.7842,
+      "step": 1240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001603393085787452,
+      "loss": 0.8116,
+      "step": 1241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016030729833546736,
+      "loss": 0.9165,
+      "step": 1242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016027528809218951,
+      "loss": 0.7712,
+      "step": 1243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016024327784891167,
+      "loss": 0.8211,
+      "step": 1244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001602112676056338,
+      "loss": 0.8548,
+      "step": 1245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016017925736235597,
+      "loss": 0.7731,
+      "step": 1246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016014724711907813,
+      "loss": 0.7631,
+      "step": 1247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016011523687580026,
+      "loss": 0.9419,
+      "step": 1248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016008322663252242,
+      "loss": 0.6864,
+      "step": 1249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016005121638924455,
+      "loss": 0.8766,
+      "step": 1250
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016001920614596674,
+      "loss": 0.7305,
+      "step": 1251
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015998719590268887,
+      "loss": 0.821,
+      "step": 1252
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000159955185659411,
+      "loss": 0.7628,
+      "step": 1253
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015992317541613316,
+      "loss": 0.6833,
+      "step": 1254
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015989116517285532,
+      "loss": 0.8862,
+      "step": 1255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00015985915492957748,
+      "loss": 0.7829,
+      "step": 1256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015982714468629962,
+      "loss": 0.8492,
+      "step": 1257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015979513444302178,
+      "loss": 0.8174,
+      "step": 1258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015976312419974394,
+      "loss": 0.7324,
+      "step": 1259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015973111395646607,
+      "loss": 0.8772,
+      "step": 1260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015969910371318823,
+      "loss": 0.843,
+      "step": 1261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001596670934699104,
+      "loss": 0.793,
+      "step": 1262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015963508322663252,
+      "loss": 0.7718,
+      "step": 1263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015960307298335468,
+      "loss": 0.9135,
+      "step": 1264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001595710627400768,
+      "loss": 0.7182,
+      "step": 1265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000159539052496799,
+      "loss": 0.7075,
+      "step": 1266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015950704225352113,
+      "loss": 0.7862,
+      "step": 1267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00015947503201024327,
+      "loss": 0.8643,
+      "step": 1268
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015944302176696545,
+      "loss": 0.8241,
+      "step": 1269
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015941101152368759,
+      "loss": 0.7758,
+      "step": 1270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015937900128040974,
+      "loss": 0.7544,
+      "step": 1271
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015934699103713188,
+      "loss": 0.8678,
+      "step": 1272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015931498079385406,
+      "loss": 0.7558,
+      "step": 1273
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001592829705505762,
+      "loss": 0.7393,
+      "step": 1274
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015925096030729833,
+      "loss": 1.0046,
+      "step": 1275
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001592189500640205,
+      "loss": 0.6813,
+      "step": 1276
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015918693982074265,
+      "loss": 0.9063,
+      "step": 1277
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001591549295774648,
+      "loss": 0.773,
+      "step": 1278
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015912291933418694,
+      "loss": 0.8058,
+      "step": 1279
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001590909090909091,
+      "loss": 0.8138,
+      "step": 1280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00015905889884763126,
+      "loss": 0.8457,
+      "step": 1281
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001590268886043534,
+      "loss": 0.7412,
+      "step": 1282
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015899487836107555,
+      "loss": 0.7379,
+      "step": 1283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015896286811779771,
+      "loss": 0.79,
+      "step": 1284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015893085787451985,
+      "loss": 0.7496,
+      "step": 1285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.000158898847631242,
+      "loss": 0.7759,
+      "step": 1286
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015886683738796414,
+      "loss": 0.7461,
+      "step": 1287
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015883482714468633,
+      "loss": 0.8152,
+      "step": 1288
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015880281690140846,
+      "loss": 0.6619,
+      "step": 1289
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001587708066581306,
+      "loss": 0.6811,
+      "step": 1290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015873879641485275,
+      "loss": 0.7638,
+      "step": 1291
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001587067861715749,
+      "loss": 0.8279,
+      "step": 1292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00015867477592829707,
+      "loss": 0.7919,
+      "step": 1293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001586427656850192,
+      "loss": 0.8066,
+      "step": 1294
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015861075544174136,
+      "loss": 0.7825,
+      "step": 1295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015857874519846352,
+      "loss": 0.8071,
+      "step": 1296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015854673495518566,
+      "loss": 0.7476,
+      "step": 1297
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015851472471190782,
+      "loss": 0.8105,
+      "step": 1298
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015848271446862997,
+      "loss": 0.8574,
+      "step": 1299
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015845070422535213,
+      "loss": 0.6954,
+      "step": 1300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015841869398207427,
+      "loss": 0.6404,
+      "step": 1301
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015838668373879643,
+      "loss": 0.8067,
+      "step": 1302
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001583546734955186,
+      "loss": 0.7663,
+      "step": 1303
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015832266325224072,
+      "loss": 0.7391,
+      "step": 1304
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015829065300896288,
+      "loss": 0.7501,
+      "step": 1305
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00015825864276568504,
+      "loss": 0.73,
+      "step": 1306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015822663252240717,
+      "loss": 0.8249,
+      "step": 1307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015819462227912933,
+      "loss": 0.7858,
+      "step": 1308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015816261203585146,
+      "loss": 0.7624,
+      "step": 1309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015813060179257365,
+      "loss": 0.7692,
+      "step": 1310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015809859154929578,
+      "loss": 0.7569,
+      "step": 1311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015806658130601792,
+      "loss": 0.8599,
+      "step": 1312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015803457106274008,
+      "loss": 0.78,
+      "step": 1313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015800256081946224,
+      "loss": 0.7294,
+      "step": 1314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001579705505761844,
+      "loss": 0.7562,
+      "step": 1315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015793854033290653,
+      "loss": 0.7559,
+      "step": 1316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0001579065300896287,
+      "loss": 0.8064,
+      "step": 1317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00015787451984635085,
+      "loss": 0.8606,
+      "step": 1318
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015784250960307298,
+      "loss": 0.8431,
+      "step": 1319
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015781049935979514,
+      "loss": 0.8695,
+      "step": 1320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001577784891165173,
+      "loss": 0.7834,
+      "step": 1321
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015774647887323943,
+      "loss": 0.8134,
+      "step": 1322
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001577144686299616,
+      "loss": 0.7756,
+      "step": 1323
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015768245838668375,
+      "loss": 0.8375,
+      "step": 1324
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001576504481434059,
+      "loss": 0.8352,
+      "step": 1325
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015761843790012805,
+      "loss": 0.8797,
+      "step": 1326
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001575864276568502,
+      "loss": 0.8347,
+      "step": 1327
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015755441741357236,
+      "loss": 0.8626,
+      "step": 1328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001575224071702945,
+      "loss": 0.7528,
+      "step": 1329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00015749039692701666,
+      "loss": 0.807,
+      "step": 1330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001574583866837388,
+      "loss": 0.7309,
+      "step": 1331
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015742637644046098,
+      "loss": 0.7916,
+      "step": 1332
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001573943661971831,
+      "loss": 0.7863,
+      "step": 1333
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015736235595390524,
+      "loss": 0.7182,
+      "step": 1334
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001573303457106274,
+      "loss": 0.7021,
+      "step": 1335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015729833546734956,
+      "loss": 0.8308,
+      "step": 1336
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015726632522407172,
+      "loss": 0.8533,
+      "step": 1337
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015723431498079385,
+      "loss": 0.7562,
+      "step": 1338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015720230473751601,
+      "loss": 0.7798,
+      "step": 1339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015717029449423817,
+      "loss": 0.7456,
+      "step": 1340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0001571382842509603,
+      "loss": 0.7887,
+      "step": 1341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015710627400768247,
+      "loss": 0.8712,
+      "step": 1342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00015707426376440463,
+      "loss": 0.7809,
+      "step": 1343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015704225352112676,
+      "loss": 0.7527,
+      "step": 1344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015701024327784892,
+      "loss": 0.8252,
+      "step": 1345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015697823303457108,
+      "loss": 0.7615,
+      "step": 1346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015694622279129324,
+      "loss": 0.7895,
+      "step": 1347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015691421254801537,
+      "loss": 0.7395,
+      "step": 1348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001568822023047375,
+      "loss": 0.7583,
+      "step": 1349
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001568501920614597,
+      "loss": 0.7255,
+      "step": 1350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015681818181818182,
+      "loss": 0.7934,
+      "step": 1351
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015678617157490398,
+      "loss": 0.8444,
+      "step": 1352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015675416133162612,
+      "loss": 0.7627,
+      "step": 1353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015672215108834828,
+      "loss": 0.6533,
+      "step": 1354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015669014084507043,
+      "loss": 0.8006,
+      "step": 1355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00015665813060179257,
+      "loss": 0.7667,
+      "step": 1356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015662612035851473,
+      "loss": 0.7063,
+      "step": 1357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001565941101152369,
+      "loss": 0.8588,
+      "step": 1358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015656209987195905,
+      "loss": 0.8057,
+      "step": 1359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015653008962868118,
+      "loss": 0.8413,
+      "step": 1360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015649807938540334,
+      "loss": 0.7312,
+      "step": 1361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001564660691421255,
+      "loss": 0.7998,
+      "step": 1362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015643405889884763,
+      "loss": 0.9268,
+      "step": 1363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001564020486555698,
+      "loss": 0.9466,
+      "step": 1364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015637003841229195,
+      "loss": 0.7932,
+      "step": 1365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015633802816901408,
+      "loss": 0.744,
+      "step": 1366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015630601792573624,
+      "loss": 0.7963,
+      "step": 1367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00015627400768245838,
+      "loss": 0.799,
+      "step": 1368
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015624199743918056,
+      "loss": 0.7829,
+      "step": 1369
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001562099871959027,
+      "loss": 0.7367,
+      "step": 1370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015617797695262483,
+      "loss": 0.7996,
+      "step": 1371
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015614596670934702,
+      "loss": 0.7828,
+      "step": 1372
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015611395646606915,
+      "loss": 0.7755,
+      "step": 1373
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001560819462227913,
+      "loss": 0.7365,
+      "step": 1374
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015604993597951344,
+      "loss": 0.7415,
+      "step": 1375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001560179257362356,
+      "loss": 0.7704,
+      "step": 1376
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015598591549295776,
+      "loss": 0.6882,
+      "step": 1377
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001559539052496799,
+      "loss": 0.7461,
+      "step": 1378
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015592189500640205,
+      "loss": 0.8627,
+      "step": 1379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001558898847631242,
+      "loss": 0.8533,
+      "step": 1380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015585787451984635,
+      "loss": 0.81,
+      "step": 1381
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001558258642765685,
+      "loss": 0.7401,
+      "step": 1382
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015579385403329066,
+      "loss": 0.8605,
+      "step": 1383
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015576184379001282,
+      "loss": 0.8051,
+      "step": 1384
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015572983354673496,
+      "loss": 0.8358,
+      "step": 1385
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015569782330345712,
+      "loss": 0.7555,
+      "step": 1386
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015566581306017928,
+      "loss": 0.7716,
+      "step": 1387
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001556338028169014,
+      "loss": 0.7215,
+      "step": 1388
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015560179257362357,
+      "loss": 0.7867,
+      "step": 1389
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001555697823303457,
+      "loss": 0.7081,
+      "step": 1390
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001555377720870679,
+      "loss": 0.8732,
+      "step": 1391
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015550576184379002,
+      "loss": 0.7357,
+      "step": 1392
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015547375160051215,
+      "loss": 0.9173,
+      "step": 1393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015544174135723431,
+      "loss": 0.8296,
+      "step": 1394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015540973111395647,
+      "loss": 0.7953,
+      "step": 1395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015537772087067863,
+      "loss": 0.7846,
+      "step": 1396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015534571062740077,
+      "loss": 0.7175,
+      "step": 1397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015531370038412293,
+      "loss": 0.7458,
+      "step": 1398
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015528169014084509,
+      "loss": 0.6837,
+      "step": 1399
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015524967989756722,
+      "loss": 0.8058,
+      "step": 1400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015521766965428938,
+      "loss": 0.8511,
+      "step": 1401
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015518565941101154,
+      "loss": 1.0781,
+      "step": 1402
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015515364916773367,
+      "loss": 0.7767,
+      "step": 1403
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015512163892445583,
+      "loss": 0.8578,
+      "step": 1404
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.000155089628681178,
+      "loss": 0.8774,
+      "step": 1405
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015505761843790015,
+      "loss": 0.796,
+      "step": 1406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015502560819462228,
+      "loss": 0.8432,
+      "step": 1407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015499359795134442,
+      "loss": 0.7676,
+      "step": 1408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001549615877080666,
+      "loss": 0.7811,
+      "step": 1409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015492957746478874,
+      "loss": 0.821,
+      "step": 1410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001548975672215109,
+      "loss": 0.8126,
+      "step": 1411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015486555697823303,
+      "loss": 0.8039,
+      "step": 1412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015483354673495521,
+      "loss": 0.7433,
+      "step": 1413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015480153649167735,
+      "loss": 0.7994,
+      "step": 1414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015476952624839948,
+      "loss": 0.8593,
+      "step": 1415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015473751600512164,
+      "loss": 0.888,
+      "step": 1416
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001547055057618438,
+      "loss": 0.7323,
+      "step": 1417
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015467349551856596,
+      "loss": 0.7995,
+      "step": 1418
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001546414852752881,
+      "loss": 0.8355,
+      "step": 1419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015460947503201025,
+      "loss": 0.8602,
+      "step": 1420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001545774647887324,
+      "loss": 0.8036,
+      "step": 1421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015454545454545454,
+      "loss": 0.8361,
+      "step": 1422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001545134443021767,
+      "loss": 0.6903,
+      "step": 1423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015448143405889886,
+      "loss": 0.7859,
+      "step": 1424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.000154449423815621,
+      "loss": 0.678,
+      "step": 1425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015441741357234316,
+      "loss": 0.8705,
+      "step": 1426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015438540332906532,
+      "loss": 0.8014,
+      "step": 1427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015435339308578748,
+      "loss": 0.8278,
+      "step": 1428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001543213828425096,
+      "loss": 0.7633,
+      "step": 1429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015428937259923174,
+      "loss": 0.8391,
+      "step": 1430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015425736235595393,
+      "loss": 0.9004,
+      "step": 1431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015422535211267606,
+      "loss": 0.7682,
+      "step": 1432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015419334186939822,
+      "loss": 0.8185,
+      "step": 1433
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015416133162612035,
+      "loss": 0.6972,
+      "step": 1434
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001541293213828425,
+      "loss": 0.7351,
+      "step": 1435
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015409731113956467,
+      "loss": 0.7794,
+      "step": 1436
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001540653008962868,
+      "loss": 0.8494,
+      "step": 1437
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015403329065300897,
+      "loss": 0.895,
+      "step": 1438
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015400128040973113,
+      "loss": 0.7281,
+      "step": 1439
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015396927016645328,
+      "loss": 0.7994,
+      "step": 1440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015393725992317542,
+      "loss": 0.7465,
+      "step": 1441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015390524967989758,
+      "loss": 0.7384,
+      "step": 1442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015387323943661974,
+      "loss": 0.9285,
+      "step": 1443
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015384122919334187,
+      "loss": 0.7307,
+      "step": 1444
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015380921895006403,
+      "loss": 0.8531,
+      "step": 1445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001537772087067862,
+      "loss": 0.7822,
+      "step": 1446
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015374519846350832,
+      "loss": 0.7575,
+      "step": 1447
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015371318822023048,
+      "loss": 0.6788,
+      "step": 1448
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015368117797695264,
+      "loss": 0.7673,
+      "step": 1449
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001536491677336748,
+      "loss": 0.7529,
+      "step": 1450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015361715749039693,
+      "loss": 0.8443,
+      "step": 1451
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015358514724711907,
+      "loss": 0.8187,
+      "step": 1452
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015355313700384125,
+      "loss": 0.8934,
+      "step": 1453
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015352112676056339,
+      "loss": 0.7839,
+      "step": 1454
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015348911651728555,
+      "loss": 0.7308,
+      "step": 1455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015345710627400768,
+      "loss": 0.6911,
+      "step": 1456
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015342509603072984,
+      "loss": 0.7407,
+      "step": 1457
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.000153393085787452,
+      "loss": 0.8135,
+      "step": 1458
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015336107554417413,
+      "loss": 0.7502,
+      "step": 1459
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001533290653008963,
+      "loss": 0.7934,
+      "step": 1460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015329705505761845,
+      "loss": 0.8031,
+      "step": 1461
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015326504481434058,
+      "loss": 0.8114,
+      "step": 1462
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015323303457106274,
+      "loss": 0.7653,
+      "step": 1463
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001532010243277849,
+      "loss": 0.8415,
+      "step": 1464
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015316901408450706,
+      "loss": 0.8392,
+      "step": 1465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001531370038412292,
+      "loss": 0.8775,
+      "step": 1466
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015310499359795136,
+      "loss": 0.7011,
+      "step": 1467
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015307298335467351,
+      "loss": 0.8593,
+      "step": 1468
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015304097311139565,
+      "loss": 0.864,
+      "step": 1469
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001530089628681178,
+      "loss": 0.7902,
+      "step": 1470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015297695262483994,
+      "loss": 0.9209,
+      "step": 1471
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015294494238156213,
+      "loss": 0.8251,
+      "step": 1472
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015291293213828426,
+      "loss": 0.7643,
+      "step": 1473
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001528809218950064,
+      "loss": 0.7794,
+      "step": 1474
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015284891165172858,
+      "loss": 0.8476,
+      "step": 1475
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001528169014084507,
+      "loss": 0.8143,
+      "step": 1476
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015278489116517287,
+      "loss": 0.8373,
+      "step": 1477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.000152752880921895,
+      "loss": 0.7823,
+      "step": 1478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015272087067861716,
+      "loss": 0.6537,
+      "step": 1479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015268886043533932,
+      "loss": 0.7732,
+      "step": 1480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015265685019206146,
+      "loss": 0.7946,
+      "step": 1481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015262483994878362,
+      "loss": 0.8311,
+      "step": 1482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015259282970550578,
+      "loss": 0.9314,
+      "step": 1483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001525608194622279,
+      "loss": 0.8991,
+      "step": 1484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015252880921895007,
+      "loss": 0.764,
+      "step": 1485
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015249679897567223,
+      "loss": 0.7203,
+      "step": 1486
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001524647887323944,
+      "loss": 0.8256,
+      "step": 1487
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015243277848911652,
+      "loss": 0.6777,
+      "step": 1488
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015240076824583865,
+      "loss": 0.8374,
+      "step": 1489
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015236875800256084,
+      "loss": 0.7811,
+      "step": 1490
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015233674775928297,
+      "loss": 0.795,
+      "step": 1491
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015230473751600513,
+      "loss": 0.7515,
+      "step": 1492
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015227272727272727,
+      "loss": 0.8313,
+      "step": 1493
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015224071702944943,
+      "loss": 0.8,
+      "step": 1494
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015220870678617159,
+      "loss": 0.6592,
+      "step": 1495
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015217669654289372,
+      "loss": 0.8699,
+      "step": 1496
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015214468629961588,
+      "loss": 0.9075,
+      "step": 1497
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015211267605633804,
+      "loss": 0.7307,
+      "step": 1498
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001520806658130602,
+      "loss": 0.7924,
+      "step": 1499
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015204865556978233,
+      "loss": 0.6903,
+      "step": 1500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001520166453265045,
+      "loss": 0.767,
+      "step": 1501
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015198463508322665,
+      "loss": 0.9658,
+      "step": 1502
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015195262483994878,
+      "loss": 0.8506,
+      "step": 1503
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015192061459667094,
+      "loss": 0.7243,
+      "step": 1504
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001518886043533931,
+      "loss": 0.7994,
+      "step": 1505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015185659411011523,
+      "loss": 0.8432,
+      "step": 1506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001518245838668374,
+      "loss": 0.7705,
+      "step": 1507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015179257362355955,
+      "loss": 0.8179,
+      "step": 1508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015176056338028171,
+      "loss": 0.8934,
+      "step": 1509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015172855313700385,
+      "loss": 0.8337,
+      "step": 1510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015169654289372598,
+      "loss": 0.7848,
+      "step": 1511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015166453265044817,
+      "loss": 0.7159,
+      "step": 1512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001516325224071703,
+      "loss": 0.8245,
+      "step": 1513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015160051216389246,
+      "loss": 0.7876,
+      "step": 1514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001515685019206146,
+      "loss": 0.7292,
+      "step": 1515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015153649167733675,
+      "loss": 0.88,
+      "step": 1516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001515044814340589,
+      "loss": 0.7606,
+      "step": 1517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00015147247119078104,
+      "loss": 0.8915,
+      "step": 1518
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001514404609475032,
+      "loss": 0.7974,
+      "step": 1519
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015140845070422536,
+      "loss": 0.8591,
+      "step": 1520
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001513764404609475,
+      "loss": 0.76,
+      "step": 1521
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015134443021766966,
+      "loss": 0.6582,
+      "step": 1522
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015131241997439182,
+      "loss": 0.7636,
+      "step": 1523
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015128040973111397,
+      "loss": 0.7579,
+      "step": 1524
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001512483994878361,
+      "loss": 0.7617,
+      "step": 1525
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015121638924455827,
+      "loss": 0.8255,
+      "step": 1526
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015118437900128043,
+      "loss": 0.9101,
+      "step": 1527
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015115236875800256,
+      "loss": 0.8265,
+      "step": 1528
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015112035851472472,
+      "loss": 0.8122,
+      "step": 1529
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015108834827144688,
+      "loss": 0.785,
+      "step": 1530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00015105633802816904,
+      "loss": 0.881,
+      "step": 1531
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015102432778489117,
+      "loss": 0.8246,
+      "step": 1532
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001509923175416133,
+      "loss": 0.8312,
+      "step": 1533
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001509603072983355,
+      "loss": 0.8054,
+      "step": 1534
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015092829705505762,
+      "loss": 0.8018,
+      "step": 1535
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015089628681177978,
+      "loss": 0.8094,
+      "step": 1536
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015086427656850192,
+      "loss": 0.7646,
+      "step": 1537
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015083226632522408,
+      "loss": 0.7819,
+      "step": 1538
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015080025608194624,
+      "loss": 0.8075,
+      "step": 1539
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015076824583866837,
+      "loss": 0.6932,
+      "step": 1540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015073623559539053,
+      "loss": 0.7649,
+      "step": 1541
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001507042253521127,
+      "loss": 0.7814,
+      "step": 1542
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00015067221510883482,
+      "loss": 0.7637,
+      "step": 1543
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015064020486555698,
+      "loss": 0.9169,
+      "step": 1544
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015060819462227914,
+      "loss": 0.8295,
+      "step": 1545
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001505761843790013,
+      "loss": 0.8248,
+      "step": 1546
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015054417413572343,
+      "loss": 0.748,
+      "step": 1547
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015051216389244557,
+      "loss": 0.7917,
+      "step": 1548
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015048015364916775,
+      "loss": 0.803,
+      "step": 1549
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015044814340588989,
+      "loss": 0.7679,
+      "step": 1550
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015041613316261205,
+      "loss": 0.7322,
+      "step": 1551
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001503841229193342,
+      "loss": 0.8185,
+      "step": 1552
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015035211267605636,
+      "loss": 0.8821,
+      "step": 1553
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001503201024327785,
+      "loss": 0.7855,
+      "step": 1554
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015028809218950063,
+      "loss": 0.7525,
+      "step": 1555
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00015025608194622282,
+      "loss": 0.7629,
+      "step": 1556
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015022407170294495,
+      "loss": 0.7666,
+      "step": 1557
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001501920614596671,
+      "loss": 0.706,
+      "step": 1558
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015016005121638924,
+      "loss": 0.7744,
+      "step": 1559
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001501280409731114,
+      "loss": 0.8378,
+      "step": 1560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015009603072983356,
+      "loss": 0.884,
+      "step": 1561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001500640204865557,
+      "loss": 0.9005,
+      "step": 1562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015003201024327785,
+      "loss": 0.7622,
+      "step": 1563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.7539,
+      "step": 1564
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014996798975672215,
+      "loss": 0.8841,
+      "step": 1565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001499359795134443,
+      "loss": 0.7795,
+      "step": 1566
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014990396927016647,
+      "loss": 0.7759,
+      "step": 1567
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014987195902688863,
+      "loss": 0.8808,
+      "step": 1568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014983994878361076,
+      "loss": 0.8022,
+      "step": 1569
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001498079385403329,
+      "loss": 0.8546,
+      "step": 1570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014977592829705508,
+      "loss": 0.8223,
+      "step": 1571
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001497439180537772,
+      "loss": 0.8803,
+      "step": 1572
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014971190781049937,
+      "loss": 0.836,
+      "step": 1573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001496798975672215,
+      "loss": 0.8962,
+      "step": 1574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014964788732394366,
+      "loss": 0.8101,
+      "step": 1575
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014961587708066582,
+      "loss": 0.8347,
+      "step": 1576
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014958386683738796,
+      "loss": 0.9249,
+      "step": 1577
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014955185659411014,
+      "loss": 0.746,
+      "step": 1578
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014951984635083228,
+      "loss": 0.7876,
+      "step": 1579
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014948783610755443,
+      "loss": 0.8489,
+      "step": 1580
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014945582586427657,
+      "loss": 0.7888,
+      "step": 1581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014942381562099873,
+      "loss": 0.7828,
+      "step": 1582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001493918053777209,
+      "loss": 0.6971,
+      "step": 1583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014935979513444302,
+      "loss": 0.8581,
+      "step": 1584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014932778489116518,
+      "loss": 0.7731,
+      "step": 1585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014929577464788734,
+      "loss": 0.8862,
+      "step": 1586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014926376440460947,
+      "loss": 0.7576,
+      "step": 1587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014923175416133163,
+      "loss": 0.8686,
+      "step": 1588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001491997439180538,
+      "loss": 0.8595,
+      "step": 1589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014916773367477595,
+      "loss": 0.8491,
+      "step": 1590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014913572343149808,
+      "loss": 0.6411,
+      "step": 1591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014910371318822022,
+      "loss": 0.9346,
+      "step": 1592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001490717029449424,
+      "loss": 0.7693,
+      "step": 1593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014903969270166454,
+      "loss": 0.7544,
+      "step": 1594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001490076824583867,
+      "loss": 0.6368,
+      "step": 1595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014897567221510883,
+      "loss": 0.7607,
+      "step": 1596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.000148943661971831,
+      "loss": 0.7344,
+      "step": 1597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014891165172855315,
+      "loss": 0.8042,
+      "step": 1598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014887964148527528,
+      "loss": 0.8853,
+      "step": 1599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014884763124199744,
+      "loss": 0.7796,
+      "step": 1600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001488156209987196,
+      "loss": 0.8594,
+      "step": 1601
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014878361075544173,
+      "loss": 0.8939,
+      "step": 1602
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001487516005121639,
+      "loss": 0.744,
+      "step": 1603
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014871959026888605,
+      "loss": 0.8537,
+      "step": 1604
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001486875800256082,
+      "loss": 0.7636,
+      "step": 1605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014865556978233035,
+      "loss": 0.8777,
+      "step": 1606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001486235595390525,
+      "loss": 0.7943,
+      "step": 1607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014859154929577467,
+      "loss": 0.7552,
+      "step": 1608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001485595390524968,
+      "loss": 0.8262,
+      "step": 1609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014852752880921896,
+      "loss": 0.7661,
+      "step": 1610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014849551856594112,
+      "loss": 0.7442,
+      "step": 1611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014846350832266328,
+      "loss": 0.9476,
+      "step": 1612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001484314980793854,
+      "loss": 0.7575,
+      "step": 1613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014839948783610754,
+      "loss": 0.8325,
+      "step": 1614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014836747759282973,
+      "loss": 0.7925,
+      "step": 1615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014833546734955186,
+      "loss": 0.8862,
+      "step": 1616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014830345710627402,
+      "loss": 0.7349,
+      "step": 1617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014827144686299615,
+      "loss": 0.736,
+      "step": 1618
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014823943661971831,
+      "loss": 0.8037,
+      "step": 1619
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014820742637644047,
+      "loss": 0.8439,
+      "step": 1620
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0001481754161331626,
+      "loss": 0.7942,
+      "step": 1621
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014814340588988477,
+      "loss": 0.8167,
+      "step": 1622
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014811139564660693,
+      "loss": 0.8849,
+      "step": 1623
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014807938540332906,
+      "loss": 0.8071,
+      "step": 1624
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014804737516005122,
+      "loss": 0.8058,
+      "step": 1625
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014801536491677338,
+      "loss": 0.7679,
+      "step": 1626
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014798335467349554,
+      "loss": 0.8025,
+      "step": 1627
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014795134443021767,
+      "loss": 0.7375,
+      "step": 1628
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014791933418693983,
+      "loss": 0.9299,
+      "step": 1629
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000147887323943662,
+      "loss": 0.7056,
+      "step": 1630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014785531370038412,
+      "loss": 0.8508,
+      "step": 1631
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014782330345710628,
+      "loss": 0.7971,
+      "step": 1632
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014779129321382844,
+      "loss": 0.7834,
+      "step": 1633
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001477592829705506,
+      "loss": 0.7162,
+      "step": 1634
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014772727272727274,
+      "loss": 0.8369,
+      "step": 1635
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014769526248399487,
+      "loss": 0.8172,
+      "step": 1636
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014766325224071705,
+      "loss": 0.804,
+      "step": 1637
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001476312419974392,
+      "loss": 0.8805,
+      "step": 1638
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014759923175416135,
+      "loss": 0.8724,
+      "step": 1639
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014756722151088348,
+      "loss": 0.9489,
+      "step": 1640
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014753521126760564,
+      "loss": 0.8291,
+      "step": 1641
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001475032010243278,
+      "loss": 0.799,
+      "step": 1642
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00014747119078104993,
+      "loss": 0.7886,
+      "step": 1643
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001474391805377721,
+      "loss": 0.7437,
+      "step": 1644
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014740717029449425,
+      "loss": 0.7589,
+      "step": 1645
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014737516005121638,
+      "loss": 0.7261,
+      "step": 1646
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014734314980793854,
+      "loss": 0.7568,
+      "step": 1647
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001473111395646607,
+      "loss": 0.7779,
+      "step": 1648
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014727912932138286,
+      "loss": 0.8424,
+      "step": 1649
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000147247119078105,
+      "loss": 0.9575,
+      "step": 1650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014721510883482713,
+      "loss": 0.8215,
+      "step": 1651
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014718309859154932,
+      "loss": 0.8809,
+      "step": 1652
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014715108834827145,
+      "loss": 0.9328,
+      "step": 1653
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001471190781049936,
+      "loss": 0.824,
+      "step": 1654
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00014708706786171577,
+      "loss": 0.8128,
+      "step": 1655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001470550576184379,
+      "loss": 0.8321,
+      "step": 1656
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014702304737516006,
+      "loss": 0.7925,
+      "step": 1657
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001469910371318822,
+      "loss": 0.7691,
+      "step": 1658
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014695902688860438,
+      "loss": 0.7419,
+      "step": 1659
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001469270166453265,
+      "loss": 0.7686,
+      "step": 1660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014689500640204865,
+      "loss": 0.715,
+      "step": 1661
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001468629961587708,
+      "loss": 0.7874,
+      "step": 1662
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014683098591549297,
+      "loss": 0.8192,
+      "step": 1663
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014679897567221513,
+      "loss": 0.6889,
+      "step": 1664
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014676696542893726,
+      "loss": 0.8251,
+      "step": 1665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014673495518565942,
+      "loss": 0.7909,
+      "step": 1666
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00014670294494238158,
+      "loss": 0.7737,
+      "step": 1667
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001466709346991037,
+      "loss": 0.8291,
+      "step": 1668
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014663892445582587,
+      "loss": 0.8179,
+      "step": 1669
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014660691421254803,
+      "loss": 0.7561,
+      "step": 1670
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001465749039692702,
+      "loss": 0.8656,
+      "step": 1671
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014654289372599232,
+      "loss": 0.8404,
+      "step": 1672
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014651088348271445,
+      "loss": 0.796,
+      "step": 1673
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014647887323943664,
+      "loss": 0.7077,
+      "step": 1674
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014644686299615877,
+      "loss": 0.874,
+      "step": 1675
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014641485275288093,
+      "loss": 0.718,
+      "step": 1676
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014638284250960307,
+      "loss": 0.7866,
+      "step": 1677
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014635083226632523,
+      "loss": 0.8582,
+      "step": 1678
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014631882202304739,
+      "loss": 0.7491,
+      "step": 1679
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00014628681177976952,
+      "loss": 0.7662,
+      "step": 1680
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001462548015364917,
+      "loss": 0.8673,
+      "step": 1681
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014622279129321384,
+      "loss": 0.8003,
+      "step": 1682
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014619078104993597,
+      "loss": 0.7525,
+      "step": 1683
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014615877080665813,
+      "loss": 0.8586,
+      "step": 1684
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001461267605633803,
+      "loss": 0.7801,
+      "step": 1685
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014609475032010245,
+      "loss": 0.8113,
+      "step": 1686
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014606274007682458,
+      "loss": 0.8013,
+      "step": 1687
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014603072983354674,
+      "loss": 0.8769,
+      "step": 1688
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001459987195902689,
+      "loss": 0.7501,
+      "step": 1689
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014596670934699104,
+      "loss": 0.7747,
+      "step": 1690
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001459346991037132,
+      "loss": 0.7341,
+      "step": 1691
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014590268886043536,
+      "loss": 0.7857,
+      "step": 1692
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00014587067861715751,
+      "loss": 0.9432,
+      "step": 1693
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014583866837387965,
+      "loss": 0.7982,
+      "step": 1694
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014580665813060178,
+      "loss": 0.8055,
+      "step": 1695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014577464788732397,
+      "loss": 0.8128,
+      "step": 1696
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001457426376440461,
+      "loss": 0.7193,
+      "step": 1697
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014571062740076826,
+      "loss": 0.7919,
+      "step": 1698
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001456786171574904,
+      "loss": 0.7772,
+      "step": 1699
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014564660691421255,
+      "loss": 0.8278,
+      "step": 1700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001456145966709347,
+      "loss": 0.7938,
+      "step": 1701
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014558258642765684,
+      "loss": 0.8413,
+      "step": 1702
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.000145550576184379,
+      "loss": 0.8201,
+      "step": 1703
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014551856594110116,
+      "loss": 0.8155,
+      "step": 1704
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001454865556978233,
+      "loss": 0.7686,
+      "step": 1705
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00014545454545454546,
+      "loss": 0.7208,
+      "step": 1706
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014542253521126762,
+      "loss": 0.8199,
+      "step": 1707
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014539052496798978,
+      "loss": 0.7633,
+      "step": 1708
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001453585147247119,
+      "loss": 0.8063,
+      "step": 1709
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014532650448143407,
+      "loss": 0.8905,
+      "step": 1710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014529449423815623,
+      "loss": 0.8149,
+      "step": 1711
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014526248399487836,
+      "loss": 0.7763,
+      "step": 1712
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014523047375160052,
+      "loss": 0.8105,
+      "step": 1713
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014519846350832268,
+      "loss": 0.9777,
+      "step": 1714
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001451664532650448,
+      "loss": 0.9211,
+      "step": 1715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00014513444302176697,
+      "loss": 0.8483,
+      "step": 1716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001451024327784891,
+      "loss": 0.7226,
+      "step": 1717
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001450704225352113,
+      "loss": 0.8005,
+      "step": 1718
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014503841229193343,
+      "loss": 0.7437,
+      "step": 1719
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014500640204865559,
+      "loss": 0.8249,
+      "step": 1720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014497439180537772,
+      "loss": 0.6258,
+      "step": 1721
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014494238156209988,
+      "loss": 0.7421,
+      "step": 1722
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014491037131882204,
+      "loss": 0.7896,
+      "step": 1723
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014487836107554417,
+      "loss": 0.6566,
+      "step": 1724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014484635083226633,
+      "loss": 0.7335,
+      "step": 1725
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001448143405889885,
+      "loss": 0.735,
+      "step": 1726
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014478233034571062,
+      "loss": 0.9428,
+      "step": 1727
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014475032010243278,
+      "loss": 0.7685,
+      "step": 1728
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014471830985915494,
+      "loss": 0.8517,
+      "step": 1729
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001446862996158771,
+      "loss": 0.6876,
+      "step": 1730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00014465428937259923,
+      "loss": 0.8987,
+      "step": 1731
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001446222791293214,
+      "loss": 0.7934,
+      "step": 1732
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014459026888604355,
+      "loss": 0.8222,
+      "step": 1733
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001445582586427657,
+      "loss": 0.8085,
+      "step": 1734
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014452624839948785,
+      "loss": 0.8252,
+      "step": 1735
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014449423815621,
+      "loss": 0.7478,
+      "step": 1736
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014446222791293214,
+      "loss": 0.7776,
+      "step": 1737
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001444302176696543,
+      "loss": 0.8924,
+      "step": 1738
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014439820742637643,
+      "loss": 0.912,
+      "step": 1739
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014436619718309862,
+      "loss": 0.7564,
+      "step": 1740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014433418693982075,
+      "loss": 0.7776,
+      "step": 1741
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014430217669654288,
+      "loss": 0.7541,
+      "step": 1742
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00014427016645326504,
+      "loss": 0.8976,
+      "step": 1743
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001442381562099872,
+      "loss": 0.8021,
+      "step": 1744
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014420614596670936,
+      "loss": 0.7639,
+      "step": 1745
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001441741357234315,
+      "loss": 0.7838,
+      "step": 1746
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014414212548015366,
+      "loss": 0.9058,
+      "step": 1747
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014411011523687582,
+      "loss": 0.8116,
+      "step": 1748
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014407810499359795,
+      "loss": 0.8207,
+      "step": 1749
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001440460947503201,
+      "loss": 0.869,
+      "step": 1750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014401408450704227,
+      "loss": 0.7934,
+      "step": 1751
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014398207426376443,
+      "loss": 0.741,
+      "step": 1752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014395006402048656,
+      "loss": 0.6828,
+      "step": 1753
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001439180537772087,
+      "loss": 0.7805,
+      "step": 1754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00014388604353393088,
+      "loss": 1.0021,
+      "step": 1755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.000143854033290653,
+      "loss": 0.7014,
+      "step": 1756
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014382202304737517,
+      "loss": 0.8346,
+      "step": 1757
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014379001280409733,
+      "loss": 0.8753,
+      "step": 1758
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014375800256081946,
+      "loss": 0.742,
+      "step": 1759
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014372599231754162,
+      "loss": 0.7709,
+      "step": 1760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014369398207426376,
+      "loss": 0.6313,
+      "step": 1761
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014366197183098594,
+      "loss": 0.7081,
+      "step": 1762
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014362996158770808,
+      "loss": 0.8308,
+      "step": 1763
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001435979513444302,
+      "loss": 0.781,
+      "step": 1764
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014356594110115237,
+      "loss": 0.8683,
+      "step": 1765
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014353393085787453,
+      "loss": 0.7069,
+      "step": 1766
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001435019206145967,
+      "loss": 0.7619,
+      "step": 1767
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014346991037131882,
+      "loss": 0.7792,
+      "step": 1768
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014343790012804098,
+      "loss": 0.8741,
+      "step": 1769
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014340588988476314,
+      "loss": 0.6966,
+      "step": 1770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014337387964148527,
+      "loss": 0.8611,
+      "step": 1771
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014334186939820743,
+      "loss": 0.8527,
+      "step": 1772
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001433098591549296,
+      "loss": 0.9097,
+      "step": 1773
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014327784891165175,
+      "loss": 0.7858,
+      "step": 1774
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014324583866837389,
+      "loss": 1.0002,
+      "step": 1775
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014321382842509602,
+      "loss": 0.9138,
+      "step": 1776
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001431818181818182,
+      "loss": 0.8807,
+      "step": 1777
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014314980793854034,
+      "loss": 0.922,
+      "step": 1778
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001431177976952625,
+      "loss": 0.7861,
+      "step": 1779
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014308578745198463,
+      "loss": 0.7813,
+      "step": 1780
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001430537772087068,
+      "loss": 0.8319,
+      "step": 1781
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014302176696542895,
+      "loss": 0.7703,
+      "step": 1782
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014298975672215108,
+      "loss": 0.7735,
+      "step": 1783
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014295774647887324,
+      "loss": 0.786,
+      "step": 1784
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001429257362355954,
+      "loss": 0.8819,
+      "step": 1785
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014289372599231753,
+      "loss": 0.8143,
+      "step": 1786
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001428617157490397,
+      "loss": 0.8072,
+      "step": 1787
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014282970550576185,
+      "loss": 0.7692,
+      "step": 1788
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014279769526248401,
+      "loss": 0.7239,
+      "step": 1789
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014276568501920615,
+      "loss": 0.7584,
+      "step": 1790
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001427336747759283,
+      "loss": 0.7888,
+      "step": 1791
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014270166453265047,
+      "loss": 0.8133,
+      "step": 1792
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001426696542893726,
+      "loss": 0.8262,
+      "step": 1793
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014263764404609476,
+      "loss": 0.6968,
+      "step": 1794
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014260563380281692,
+      "loss": 0.7521,
+      "step": 1795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014257362355953905,
+      "loss": 0.6863,
+      "step": 1796
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001425416133162612,
+      "loss": 0.7868,
+      "step": 1797
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014250960307298334,
+      "loss": 0.7872,
+      "step": 1798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014247759282970553,
+      "loss": 0.8016,
+      "step": 1799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014244558258642766,
+      "loss": 0.8514,
+      "step": 1800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001424135723431498,
+      "loss": 0.8743,
+      "step": 1801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014238156209987196,
+      "loss": 0.7418,
+      "step": 1802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014234955185659412,
+      "loss": 0.854,
+      "step": 1803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014231754161331628,
+      "loss": 0.7746,
+      "step": 1804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.0001422855313700384,
+      "loss": 0.7321,
+      "step": 1805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014225352112676057,
+      "loss": 0.7725,
+      "step": 1806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014222151088348273,
+      "loss": 0.9379,
+      "step": 1807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014218950064020486,
+      "loss": 0.7166,
+      "step": 1808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014215749039692702,
+      "loss": 0.7212,
+      "step": 1809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014212548015364918,
+      "loss": 0.8487,
+      "step": 1810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014209346991037134,
+      "loss": 0.9052,
+      "step": 1811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014206145966709347,
+      "loss": 0.7333,
+      "step": 1812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014202944942381563,
+      "loss": 0.7515,
+      "step": 1813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001419974391805378,
+      "loss": 0.7751,
+      "step": 1814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014196542893725992,
+      "loss": 0.7935,
+      "step": 1815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014193341869398208,
+      "loss": 0.7858,
+      "step": 1816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014190140845070424,
+      "loss": 0.7857,
+      "step": 1817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00014186939820742638,
+      "loss": 0.7668,
+      "step": 1818
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014183738796414854,
+      "loss": 0.8822,
+      "step": 1819
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014180537772087067,
+      "loss": 0.6709,
+      "step": 1820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014177336747759286,
+      "loss": 0.7848,
+      "step": 1821
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000141741357234315,
+      "loss": 0.6744,
+      "step": 1822
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014170934699103712,
+      "loss": 0.8619,
+      "step": 1823
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014167733674775928,
+      "loss": 0.7224,
+      "step": 1824
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014164532650448144,
+      "loss": 0.7624,
+      "step": 1825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001416133162612036,
+      "loss": 0.8271,
+      "step": 1826
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014158130601792573,
+      "loss": 0.9406,
+      "step": 1827
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001415492957746479,
+      "loss": 0.8056,
+      "step": 1828
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014151728553137005,
+      "loss": 0.8948,
+      "step": 1829
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014148527528809219,
+      "loss": 0.7166,
+      "step": 1830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014145326504481435,
+      "loss": 0.7185,
+      "step": 1831
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001414212548015365,
+      "loss": 0.8773,
+      "step": 1832
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014138924455825867,
+      "loss": 0.891,
+      "step": 1833
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001413572343149808,
+      "loss": 0.7602,
+      "step": 1834
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014132522407170296,
+      "loss": 0.726,
+      "step": 1835
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014129321382842512,
+      "loss": 0.6867,
+      "step": 1836
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014126120358514725,
+      "loss": 0.7673,
+      "step": 1837
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001412291933418694,
+      "loss": 0.8232,
+      "step": 1838
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014119718309859157,
+      "loss": 0.8625,
+      "step": 1839
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001411651728553137,
+      "loss": 0.7699,
+      "step": 1840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014113316261203586,
+      "loss": 0.9115,
+      "step": 1841
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000141101152368758,
+      "loss": 0.6825,
+      "step": 1842
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014106914212548018,
+      "loss": 0.8604,
+      "step": 1843
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014103713188220231,
+      "loss": 0.8227,
+      "step": 1844
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014100512163892445,
+      "loss": 0.8292,
+      "step": 1845
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001409731113956466,
+      "loss": 0.8363,
+      "step": 1846
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014094110115236877,
+      "loss": 0.8236,
+      "step": 1847
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014090909090909093,
+      "loss": 0.7956,
+      "step": 1848
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014087708066581306,
+      "loss": 0.708,
+      "step": 1849
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014084507042253522,
+      "loss": 0.8345,
+      "step": 1850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014081306017925738,
+      "loss": 0.8184,
+      "step": 1851
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001407810499359795,
+      "loss": 0.7525,
+      "step": 1852
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014074903969270167,
+      "loss": 0.7258,
+      "step": 1853
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014071702944942383,
+      "loss": 0.7488,
+      "step": 1854
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014068501920614596,
+      "loss": 0.9129,
+      "step": 1855
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014065300896286812,
+      "loss": 0.9012,
+      "step": 1856
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014062099871959026,
+      "loss": 0.8658,
+      "step": 1857
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014058898847631244,
+      "loss": 0.8366,
+      "step": 1858
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014055697823303458,
+      "loss": 0.9142,
+      "step": 1859
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014052496798975674,
+      "loss": 0.7863,
+      "step": 1860
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014049295774647887,
+      "loss": 0.7065,
+      "step": 1861
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014046094750320103,
+      "loss": 0.7804,
+      "step": 1862
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001404289372599232,
+      "loss": 0.823,
+      "step": 1863
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014039692701664532,
+      "loss": 0.8132,
+      "step": 1864
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001403649167733675,
+      "loss": 0.8052,
+      "step": 1865
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014033290653008964,
+      "loss": 0.9199,
+      "step": 1866
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014030089628681177,
+      "loss": 0.8074,
+      "step": 1867
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00014026888604353393,
+      "loss": 0.8003,
+      "step": 1868
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001402368758002561,
+      "loss": 0.7303,
+      "step": 1869
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014020486555697825,
+      "loss": 0.847,
+      "step": 1870
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014017285531370038,
+      "loss": 0.8361,
+      "step": 1871
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014014084507042254,
+      "loss": 0.8121,
+      "step": 1872
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001401088348271447,
+      "loss": 0.8575,
+      "step": 1873
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014007682458386684,
+      "loss": 0.7784,
+      "step": 1874
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000140044814340589,
+      "loss": 0.7621,
+      "step": 1875
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014001280409731116,
+      "loss": 0.754,
+      "step": 1876
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001399807938540333,
+      "loss": 0.8696,
+      "step": 1877
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013994878361075545,
+      "loss": 0.7023,
+      "step": 1878
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013991677336747758,
+      "loss": 0.7708,
+      "step": 1879
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00013988476312419977,
+      "loss": 0.8324,
+      "step": 1880
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001398527528809219,
+      "loss": 0.7259,
+      "step": 1881
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013982074263764403,
+      "loss": 0.8394,
+      "step": 1882
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001397887323943662,
+      "loss": 0.7512,
+      "step": 1883
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013975672215108835,
+      "loss": 0.8402,
+      "step": 1884
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001397247119078105,
+      "loss": 0.8124,
+      "step": 1885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013969270166453265,
+      "loss": 0.7244,
+      "step": 1886
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001396606914212548,
+      "loss": 0.8578,
+      "step": 1887
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013962868117797697,
+      "loss": 0.8294,
+      "step": 1888
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001395966709346991,
+      "loss": 0.7087,
+      "step": 1889
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013956466069142126,
+      "loss": 0.831,
+      "step": 1890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013953265044814342,
+      "loss": 0.8128,
+      "step": 1891
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00013950064020486558,
+      "loss": 0.726,
+      "step": 1892
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001394686299615877,
+      "loss": 0.7587,
+      "step": 1893
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013943661971830987,
+      "loss": 0.8507,
+      "step": 1894
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013940460947503203,
+      "loss": 0.679,
+      "step": 1895
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013937259923175416,
+      "loss": 0.8174,
+      "step": 1896
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013934058898847632,
+      "loss": 0.8223,
+      "step": 1897
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013930857874519848,
+      "loss": 0.752,
+      "step": 1898
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013927656850192061,
+      "loss": 0.8452,
+      "step": 1899
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013924455825864277,
+      "loss": 0.8213,
+      "step": 1900
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001392125480153649,
+      "loss": 0.7748,
+      "step": 1901
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001391805377720871,
+      "loss": 0.8694,
+      "step": 1902
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013914852752880923,
+      "loss": 0.7582,
+      "step": 1903
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013911651728553136,
+      "loss": 0.8288,
+      "step": 1904
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013908450704225352,
+      "loss": 0.7937,
+      "step": 1905
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00013905249679897568,
+      "loss": 0.7749,
+      "step": 1906
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013902048655569784,
+      "loss": 0.7771,
+      "step": 1907
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013898847631241997,
+      "loss": 0.8423,
+      "step": 1908
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013895646606914213,
+      "loss": 0.9121,
+      "step": 1909
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001389244558258643,
+      "loss": 0.7859,
+      "step": 1910
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013889244558258642,
+      "loss": 0.9535,
+      "step": 1911
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013886043533930858,
+      "loss": 0.6877,
+      "step": 1912
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013882842509603074,
+      "loss": 0.9029,
+      "step": 1913
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001387964148527529,
+      "loss": 0.8638,
+      "step": 1914
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013876440460947504,
+      "loss": 0.7333,
+      "step": 1915
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001387323943661972,
+      "loss": 0.7361,
+      "step": 1916
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00013870038412291936,
+      "loss": 0.8098,
+      "step": 1917
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0001386683738796415,
+      "loss": 0.9054,
+      "step": 1918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013863636363636365,
+      "loss": 0.8737,
+      "step": 1919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001386043533930858,
+      "loss": 0.9087,
+      "step": 1920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013857234314980794,
+      "loss": 0.6999,
+      "step": 1921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001385403329065301,
+      "loss": 0.6442,
+      "step": 1922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013850832266325223,
+      "loss": 0.6952,
+      "step": 1923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013847631241997442,
+      "loss": 0.9183,
+      "step": 1924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013844430217669655,
+      "loss": 0.694,
+      "step": 1925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013841229193341868,
+      "loss": 0.7194,
+      "step": 1926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013838028169014084,
+      "loss": 0.8059,
+      "step": 1927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.000138348271446863,
+      "loss": 0.7858,
+      "step": 1928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013831626120358516,
+      "loss": 0.8135,
+      "step": 1929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001382842509603073,
+      "loss": 0.8308,
+      "step": 1930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00013825224071702946,
+      "loss": 0.8387,
+      "step": 1931
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013822023047375162,
+      "loss": 0.7318,
+      "step": 1932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013818822023047375,
+      "loss": 0.7419,
+      "step": 1933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001381562099871959,
+      "loss": 0.8114,
+      "step": 1934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013812419974391807,
+      "loss": 0.6677,
+      "step": 1935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001380921895006402,
+      "loss": 0.7079,
+      "step": 1936
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013806017925736236,
+      "loss": 0.892,
+      "step": 1937
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013802816901408452,
+      "loss": 0.7486,
+      "step": 1938
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013799615877080668,
+      "loss": 0.6975,
+      "step": 1939
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001379641485275288,
+      "loss": 0.6812,
+      "step": 1940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013793213828425097,
+      "loss": 0.8656,
+      "step": 1941
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013790012804097313,
+      "loss": 0.7954,
+      "step": 1942
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00013786811779769527,
+      "loss": 0.8582,
+      "step": 1943
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013783610755441743,
+      "loss": 0.9114,
+      "step": 1944
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013780409731113956,
+      "loss": 0.8069,
+      "step": 1945
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013777208706786174,
+      "loss": 0.9102,
+      "step": 1946
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013774007682458388,
+      "loss": 0.7862,
+      "step": 1947
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.000137708066581306,
+      "loss": 0.9053,
+      "step": 1948
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013767605633802817,
+      "loss": 0.6623,
+      "step": 1949
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013764404609475033,
+      "loss": 0.8222,
+      "step": 1950
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001376120358514725,
+      "loss": 0.8463,
+      "step": 1951
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013758002560819462,
+      "loss": 0.8529,
+      "step": 1952
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013754801536491678,
+      "loss": 0.7915,
+      "step": 1953
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013751600512163894,
+      "loss": 0.7528,
+      "step": 1954
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013748399487836107,
+      "loss": 0.8282,
+      "step": 1955
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00013745198463508323,
+      "loss": 0.7087,
+      "step": 1956
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001374199743918054,
+      "loss": 0.8415,
+      "step": 1957
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013738796414852753,
+      "loss": 0.8534,
+      "step": 1958
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001373559539052497,
+      "loss": 0.8408,
+      "step": 1959
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013732394366197182,
+      "loss": 0.815,
+      "step": 1960
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.000137291933418694,
+      "loss": 0.8512,
+      "step": 1961
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013725992317541614,
+      "loss": 0.9043,
+      "step": 1962
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013722791293213827,
+      "loss": 0.8152,
+      "step": 1963
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013719590268886043,
+      "loss": 0.7459,
+      "step": 1964
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001371638924455826,
+      "loss": 0.8359,
+      "step": 1965
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013713188220230475,
+      "loss": 0.6972,
+      "step": 1966
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013709987195902688,
+      "loss": 0.9094,
+      "step": 1967
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00013706786171574904,
+      "loss": 0.7326,
+      "step": 1968
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001370358514724712,
+      "loss": 0.7704,
+      "step": 1969
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013700384122919334,
+      "loss": 0.8843,
+      "step": 1970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001369718309859155,
+      "loss": 0.8769,
+      "step": 1971
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013693982074263766,
+      "loss": 0.7074,
+      "step": 1972
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013690781049935982,
+      "loss": 0.6944,
+      "step": 1973
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013687580025608195,
+      "loss": 0.8209,
+      "step": 1974
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001368437900128041,
+      "loss": 0.7799,
+      "step": 1975
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013681177976952627,
+      "loss": 0.7663,
+      "step": 1976
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001367797695262484,
+      "loss": 0.8629,
+      "step": 1977
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013674775928297056,
+      "loss": 0.7364,
+      "step": 1978
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013671574903969272,
+      "loss": 0.699,
+      "step": 1979
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00013668373879641485,
+      "loss": 0.7545,
+      "step": 1980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.000136651728553137,
+      "loss": 0.7783,
+      "step": 1981
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013661971830985914,
+      "loss": 0.8319,
+      "step": 1982
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013658770806658133,
+      "loss": 0.7703,
+      "step": 1983
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013655569782330346,
+      "loss": 0.9258,
+      "step": 1984
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001365236875800256,
+      "loss": 0.8609,
+      "step": 1985
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013649167733674776,
+      "loss": 0.8083,
+      "step": 1986
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013645966709346992,
+      "loss": 0.8553,
+      "step": 1987
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013642765685019208,
+      "loss": 0.7923,
+      "step": 1988
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001363956466069142,
+      "loss": 0.823,
+      "step": 1989
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013636363636363637,
+      "loss": 0.7516,
+      "step": 1990
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013633162612035853,
+      "loss": 0.7102,
+      "step": 1991
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013629961587708066,
+      "loss": 0.7908,
+      "step": 1992
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00013626760563380282,
+      "loss": 0.7762,
+      "step": 1993
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013623559539052498,
+      "loss": 0.8503,
+      "step": 1994
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013620358514724711,
+      "loss": 0.7719,
+      "step": 1995
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013617157490396927,
+      "loss": 0.7436,
+      "step": 1996
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013613956466069143,
+      "loss": 0.7638,
+      "step": 1997
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001361075544174136,
+      "loss": 0.7241,
+      "step": 1998
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013607554417413573,
+      "loss": 0.7975,
+      "step": 1999
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013604353393085789,
+      "loss": 0.8595,
+      "step": 2000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013601152368758005,
+      "loss": 0.8379,
+      "step": 2001
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013597951344430218,
+      "loss": 0.9449,
+      "step": 2002
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013594750320102434,
+      "loss": 0.6544,
+      "step": 2003
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013591549295774647,
+      "loss": 0.791,
+      "step": 2004
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00013588348271446866,
+      "loss": 0.9656,
+      "step": 2005
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001358514724711908,
+      "loss": 0.7185,
+      "step": 2006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013581946222791292,
+      "loss": 0.8398,
+      "step": 2007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013578745198463508,
+      "loss": 0.8204,
+      "step": 2008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013575544174135724,
+      "loss": 0.8501,
+      "step": 2009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001357234314980794,
+      "loss": 0.7279,
+      "step": 2010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013569142125480153,
+      "loss": 0.6601,
+      "step": 2011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001356594110115237,
+      "loss": 0.9364,
+      "step": 2012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013562740076824585,
+      "loss": 0.7515,
+      "step": 2013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.000135595390524968,
+      "loss": 0.7803,
+      "step": 2014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013556338028169015,
+      "loss": 0.8684,
+      "step": 2015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001355313700384123,
+      "loss": 0.8614,
+      "step": 2016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00013549935979513444,
+      "loss": 0.7589,
+      "step": 2017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001354673495518566,
+      "loss": 0.8162,
+      "step": 2018
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013543533930857876,
+      "loss": 0.7767,
+      "step": 2019
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013540332906530092,
+      "loss": 0.7627,
+      "step": 2020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013537131882202305,
+      "loss": 0.9271,
+      "step": 2021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013533930857874518,
+      "loss": 0.8937,
+      "step": 2022
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013530729833546737,
+      "loss": 0.7259,
+      "step": 2023
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001352752880921895,
+      "loss": 0.8774,
+      "step": 2024
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013524327784891166,
+      "loss": 0.9098,
+      "step": 2025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001352112676056338,
+      "loss": 0.7975,
+      "step": 2026
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013517925736235598,
+      "loss": 0.8001,
+      "step": 2027
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013514724711907812,
+      "loss": 0.8083,
+      "step": 2028
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013511523687580025,
+      "loss": 0.8397,
+      "step": 2029
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001350832266325224,
+      "loss": 0.8172,
+      "step": 2030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00013505121638924457,
+      "loss": 0.9286,
+      "step": 2031
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013501920614596673,
+      "loss": 0.8163,
+      "step": 2032
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013498719590268886,
+      "loss": 0.8016,
+      "step": 2033
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013495518565941102,
+      "loss": 0.7472,
+      "step": 2034
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013492317541613318,
+      "loss": 0.8486,
+      "step": 2035
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001348911651728553,
+      "loss": 0.7617,
+      "step": 2036
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013485915492957747,
+      "loss": 0.7298,
+      "step": 2037
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013482714468629963,
+      "loss": 0.8364,
+      "step": 2038
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013479513444302176,
+      "loss": 0.8083,
+      "step": 2039
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013476312419974392,
+      "loss": 0.7807,
+      "step": 2040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013473111395646606,
+      "loss": 0.8154,
+      "step": 2041
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013469910371318824,
+      "loss": 0.9473,
+      "step": 2042
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00013466709346991038,
+      "loss": 0.8027,
+      "step": 2043
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001346350832266325,
+      "loss": 0.9493,
+      "step": 2044
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001346030729833547,
+      "loss": 0.9331,
+      "step": 2045
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013457106274007683,
+      "loss": 0.7347,
+      "step": 2046
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.000134539052496799,
+      "loss": 0.8928,
+      "step": 2047
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013450704225352112,
+      "loss": 0.9055,
+      "step": 2048
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013447503201024328,
+      "loss": 0.7776,
+      "step": 2049
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013444302176696544,
+      "loss": 0.9204,
+      "step": 2050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013441101152368757,
+      "loss": 0.7964,
+      "step": 2051
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013437900128040973,
+      "loss": 0.7855,
+      "step": 2052
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001343469910371319,
+      "loss": 0.8286,
+      "step": 2053
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013431498079385405,
+      "loss": 0.8406,
+      "step": 2054
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013428297055057619,
+      "loss": 0.7564,
+      "step": 2055
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00013425096030729835,
+      "loss": 0.8321,
+      "step": 2056
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001342189500640205,
+      "loss": 0.7934,
+      "step": 2057
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013418693982074264,
+      "loss": 0.7673,
+      "step": 2058
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001341549295774648,
+      "loss": 0.6509,
+      "step": 2059
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013412291933418696,
+      "loss": 0.7446,
+      "step": 2060
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001340909090909091,
+      "loss": 0.7802,
+      "step": 2061
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013405889884763125,
+      "loss": 0.673,
+      "step": 2062
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013402688860435338,
+      "loss": 0.7185,
+      "step": 2063
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013399487836107557,
+      "loss": 0.7423,
+      "step": 2064
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001339628681177977,
+      "loss": 0.7384,
+      "step": 2065
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013393085787451983,
+      "loss": 0.7325,
+      "step": 2066
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.000133898847631242,
+      "loss": 0.7808,
+      "step": 2067
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00013386683738796415,
+      "loss": 0.8697,
+      "step": 2068
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013383482714468631,
+      "loss": 0.7318,
+      "step": 2069
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013380281690140845,
+      "loss": 0.8843,
+      "step": 2070
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001337708066581306,
+      "loss": 0.8745,
+      "step": 2071
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013373879641485277,
+      "loss": 0.8322,
+      "step": 2072
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001337067861715749,
+      "loss": 0.8294,
+      "step": 2073
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013367477592829706,
+      "loss": 0.8241,
+      "step": 2074
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013364276568501922,
+      "loss": 0.7071,
+      "step": 2075
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013361075544174135,
+      "loss": 0.7174,
+      "step": 2076
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001335787451984635,
+      "loss": 0.8647,
+      "step": 2077
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013354673495518567,
+      "loss": 0.7391,
+      "step": 2078
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013351472471190783,
+      "loss": 0.8611,
+      "step": 2079
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013348271446862996,
+      "loss": 0.7397,
+      "step": 2080
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00013345070422535212,
+      "loss": 0.9228,
+      "step": 2081
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013341869398207428,
+      "loss": 0.7624,
+      "step": 2082
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013338668373879642,
+      "loss": 0.8612,
+      "step": 2083
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013335467349551858,
+      "loss": 0.7967,
+      "step": 2084
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001333226632522407,
+      "loss": 0.6998,
+      "step": 2085
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0001332906530089629,
+      "loss": 0.7196,
+      "step": 2086
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013325864276568503,
+      "loss": 0.8389,
+      "step": 2087
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013322663252240716,
+      "loss": 0.8659,
+      "step": 2088
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013319462227912932,
+      "loss": 0.7824,
+      "step": 2089
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013316261203585148,
+      "loss": 0.8423,
+      "step": 2090
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013313060179257364,
+      "loss": 0.8072,
+      "step": 2091
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013309859154929577,
+      "loss": 0.688,
+      "step": 2092
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00013306658130601793,
+      "loss": 0.816,
+      "step": 2093
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001330345710627401,
+      "loss": 0.8681,
+      "step": 2094
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013300256081946222,
+      "loss": 0.7747,
+      "step": 2095
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013297055057618438,
+      "loss": 0.7212,
+      "step": 2096
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013293854033290654,
+      "loss": 0.7292,
+      "step": 2097
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013290653008962868,
+      "loss": 0.7687,
+      "step": 2098
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013287451984635084,
+      "loss": 0.8876,
+      "step": 2099
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.000132842509603073,
+      "loss": 0.7323,
+      "step": 2100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013281049935979516,
+      "loss": 0.7519,
+      "step": 2101
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001327784891165173,
+      "loss": 0.8192,
+      "step": 2102
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013274647887323942,
+      "loss": 0.7152,
+      "step": 2103
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001327144686299616,
+      "loss": 0.8495,
+      "step": 2104
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00013268245838668374,
+      "loss": 0.7164,
+      "step": 2105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0001326504481434059,
+      "loss": 0.7314,
+      "step": 2106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013261843790012803,
+      "loss": 1.0317,
+      "step": 2107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013258642765685022,
+      "loss": 0.7477,
+      "step": 2108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013255441741357235,
+      "loss": 0.6991,
+      "step": 2109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013252240717029449,
+      "loss": 0.7741,
+      "step": 2110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013249039692701665,
+      "loss": 0.7659,
+      "step": 2111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001324583866837388,
+      "loss": 0.8777,
+      "step": 2112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013242637644046097,
+      "loss": 0.7962,
+      "step": 2113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001323943661971831,
+      "loss": 0.6705,
+      "step": 2114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013236235595390526,
+      "loss": 0.8803,
+      "step": 2115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013233034571062742,
+      "loss": 0.7764,
+      "step": 2116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00013229833546734955,
+      "loss": 0.7597,
+      "step": 2117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.0001322663252240717,
+      "loss": 0.8411,
+      "step": 2118
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013223431498079387,
+      "loss": 0.7529,
+      "step": 2119
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.000132202304737516,
+      "loss": 0.7424,
+      "step": 2120
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013217029449423816,
+      "loss": 0.8081,
+      "step": 2121
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013213828425096032,
+      "loss": 0.8407,
+      "step": 2122
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013210627400768248,
+      "loss": 0.8361,
+      "step": 2123
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013207426376440461,
+      "loss": 0.7547,
+      "step": 2124
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013204225352112675,
+      "loss": 0.7599,
+      "step": 2125
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013201024327784893,
+      "loss": 0.8283,
+      "step": 2126
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013197823303457107,
+      "loss": 0.8298,
+      "step": 2127
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013194622279129323,
+      "loss": 0.8707,
+      "step": 2128
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013191421254801536,
+      "loss": 0.7072,
+      "step": 2129
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013188220230473752,
+      "loss": 0.8107,
+      "step": 2130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00013185019206145968,
+      "loss": 0.7704,
+      "step": 2131
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001318181818181818,
+      "loss": 0.8061,
+      "step": 2132
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013178617157490397,
+      "loss": 0.7352,
+      "step": 2133
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013175416133162613,
+      "loss": 0.8896,
+      "step": 2134
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013172215108834826,
+      "loss": 0.7622,
+      "step": 2135
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013169014084507042,
+      "loss": 0.8314,
+      "step": 2136
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013165813060179258,
+      "loss": 0.8009,
+      "step": 2137
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013162612035851474,
+      "loss": 0.7935,
+      "step": 2138
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013159411011523688,
+      "loss": 0.8486,
+      "step": 2139
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013156209987195904,
+      "loss": 0.7717,
+      "step": 2140
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001315300896286812,
+      "loss": 0.727,
+      "step": 2141
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00013149807938540333,
+      "loss": 0.774,
+      "step": 2142
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001314660691421255,
+      "loss": 0.7797,
+      "step": 2143
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013143405889884762,
+      "loss": 0.905,
+      "step": 2144
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001314020486555698,
+      "loss": 0.8966,
+      "step": 2145
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013137003841229194,
+      "loss": 0.7789,
+      "step": 2146
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013133802816901407,
+      "loss": 0.7626,
+      "step": 2147
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013130601792573626,
+      "loss": 0.7022,
+      "step": 2148
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001312740076824584,
+      "loss": 0.8213,
+      "step": 2149
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013124199743918055,
+      "loss": 0.6726,
+      "step": 2150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013120998719590268,
+      "loss": 0.8322,
+      "step": 2151
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013117797695262484,
+      "loss": 0.8236,
+      "step": 2152
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000131145966709347,
+      "loss": 0.9294,
+      "step": 2153
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013111395646606914,
+      "loss": 0.7224,
+      "step": 2154
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001310819462227913,
+      "loss": 0.7275,
+      "step": 2155
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00013104993597951346,
+      "loss": 0.7875,
+      "step": 2156
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001310179257362356,
+      "loss": 0.7809,
+      "step": 2157
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013098591549295775,
+      "loss": 0.7978,
+      "step": 2158
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001309539052496799,
+      "loss": 0.8105,
+      "step": 2159
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013092189500640207,
+      "loss": 0.8092,
+      "step": 2160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001308898847631242,
+      "loss": 0.7096,
+      "step": 2161
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013085787451984633,
+      "loss": 0.7163,
+      "step": 2162
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013082586427656852,
+      "loss": 0.7948,
+      "step": 2163
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013079385403329065,
+      "loss": 0.8346,
+      "step": 2164
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001307618437900128,
+      "loss": 0.8237,
+      "step": 2165
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013072983354673495,
+      "loss": 0.8551,
+      "step": 2166
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013069782330345713,
+      "loss": 0.8693,
+      "step": 2167
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00013066581306017927,
+      "loss": 0.811,
+      "step": 2168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001306338028169014,
+      "loss": 0.6716,
+      "step": 2169
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013060179257362356,
+      "loss": 0.7152,
+      "step": 2170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013056978233034572,
+      "loss": 0.7828,
+      "step": 2171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013053777208706788,
+      "loss": 0.832,
+      "step": 2172
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013050576184379,
+      "loss": 0.8243,
+      "step": 2173
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013047375160051217,
+      "loss": 0.7984,
+      "step": 2174
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013044174135723433,
+      "loss": 0.9256,
+      "step": 2175
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013040973111395646,
+      "loss": 0.7593,
+      "step": 2176
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013037772087067862,
+      "loss": 0.6879,
+      "step": 2177
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013034571062740078,
+      "loss": 0.68,
+      "step": 2178
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013031370038412291,
+      "loss": 0.9258,
+      "step": 2179
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013028169014084507,
+      "loss": 0.8539,
+      "step": 2180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00013024967989756723,
+      "loss": 0.8874,
+      "step": 2181
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001302176696542894,
+      "loss": 0.865,
+      "step": 2182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013018565941101153,
+      "loss": 0.8276,
+      "step": 2183
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013015364916773366,
+      "loss": 0.79,
+      "step": 2184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013012163892445585,
+      "loss": 0.8691,
+      "step": 2185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013008962868117798,
+      "loss": 0.8191,
+      "step": 2186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013005761843790014,
+      "loss": 0.8892,
+      "step": 2187
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00013002560819462227,
+      "loss": 0.9245,
+      "step": 2188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012999359795134443,
+      "loss": 0.7298,
+      "step": 2189
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001299615877080666,
+      "loss": 0.7226,
+      "step": 2190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012992957746478872,
+      "loss": 0.8049,
+      "step": 2191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012989756722151088,
+      "loss": 0.7365,
+      "step": 2192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00012986555697823304,
+      "loss": 0.8101,
+      "step": 2193
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001298335467349552,
+      "loss": 0.9158,
+      "step": 2194
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012980153649167734,
+      "loss": 0.7195,
+      "step": 2195
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001297695262483995,
+      "loss": 0.7619,
+      "step": 2196
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012973751600512166,
+      "loss": 0.7061,
+      "step": 2197
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001297055057618438,
+      "loss": 0.7237,
+      "step": 2198
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012967349551856595,
+      "loss": 0.8493,
+      "step": 2199
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001296414852752881,
+      "loss": 0.7741,
+      "step": 2200
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012960947503201024,
+      "loss": 0.7358,
+      "step": 2201
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001295774647887324,
+      "loss": 0.8846,
+      "step": 2202
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012954545454545456,
+      "loss": 0.664,
+      "step": 2203
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012951344430217672,
+      "loss": 0.8512,
+      "step": 2204
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012948143405889885,
+      "loss": 0.7398,
+      "step": 2205
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00012944942381562099,
+      "loss": 0.888,
+      "step": 2206
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012941741357234317,
+      "loss": 0.7014,
+      "step": 2207
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001293854033290653,
+      "loss": 0.7836,
+      "step": 2208
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012935339308578746,
+      "loss": 0.7568,
+      "step": 2209
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001293213828425096,
+      "loss": 0.776,
+      "step": 2210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012928937259923176,
+      "loss": 0.761,
+      "step": 2211
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012925736235595392,
+      "loss": 0.8871,
+      "step": 2212
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012922535211267605,
+      "loss": 0.7431,
+      "step": 2213
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001291933418693982,
+      "loss": 0.8047,
+      "step": 2214
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012916133162612037,
+      "loss": 0.7397,
+      "step": 2215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001291293213828425,
+      "loss": 0.714,
+      "step": 2216
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012909731113956466,
+      "loss": 0.7932,
+      "step": 2217
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00012906530089628682,
+      "loss": 0.7079,
+      "step": 2218
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012903329065300898,
+      "loss": 0.7138,
+      "step": 2219
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012900128040973111,
+      "loss": 0.8064,
+      "step": 2220
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012896927016645327,
+      "loss": 0.9156,
+      "step": 2221
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012893725992317543,
+      "loss": 0.8304,
+      "step": 2222
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012890524967989757,
+      "loss": 0.6974,
+      "step": 2223
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012887323943661973,
+      "loss": 0.7663,
+      "step": 2224
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012884122919334189,
+      "loss": 0.8244,
+      "step": 2225
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012880921895006405,
+      "loss": 0.8414,
+      "step": 2226
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012877720870678618,
+      "loss": 0.7259,
+      "step": 2227
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001287451984635083,
+      "loss": 0.7257,
+      "step": 2228
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001287131882202305,
+      "loss": 0.7268,
+      "step": 2229
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00012868117797695263,
+      "loss": 0.76,
+      "step": 2230
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001286491677336748,
+      "loss": 0.8708,
+      "step": 2231
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012861715749039692,
+      "loss": 0.6455,
+      "step": 2232
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012858514724711908,
+      "loss": 0.7873,
+      "step": 2233
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012855313700384124,
+      "loss": 0.7654,
+      "step": 2234
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012852112676056337,
+      "loss": 0.8234,
+      "step": 2235
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012848911651728553,
+      "loss": 0.8136,
+      "step": 2236
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001284571062740077,
+      "loss": 0.7151,
+      "step": 2237
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012842509603072983,
+      "loss": 0.8137,
+      "step": 2238
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.000128393085787452,
+      "loss": 0.82,
+      "step": 2239
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012836107554417415,
+      "loss": 0.7529,
+      "step": 2240
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.0001283290653008963,
+      "loss": 0.8577,
+      "step": 2241
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012829705505761844,
+      "loss": 0.8473,
+      "step": 2242
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00012826504481434057,
+      "loss": 0.7634,
+      "step": 2243
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012823303457106276,
+      "loss": 0.6804,
+      "step": 2244
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001282010243277849,
+      "loss": 0.7933,
+      "step": 2245
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012816901408450705,
+      "loss": 0.7828,
+      "step": 2246
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012813700384122918,
+      "loss": 0.6968,
+      "step": 2247
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012810499359795137,
+      "loss": 0.9029,
+      "step": 2248
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001280729833546735,
+      "loss": 0.8473,
+      "step": 2249
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012804097311139564,
+      "loss": 0.8141,
+      "step": 2250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012800896286811782,
+      "loss": 0.8096,
+      "step": 2251
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012797695262483996,
+      "loss": 0.8011,
+      "step": 2252
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012794494238156212,
+      "loss": 0.7916,
+      "step": 2253
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012791293213828425,
+      "loss": 0.8641,
+      "step": 2254
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001278809218950064,
+      "loss": 0.8875,
+      "step": 2255
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00012784891165172857,
+      "loss": 0.8044,
+      "step": 2256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001278169014084507,
+      "loss": 0.745,
+      "step": 2257
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012778489116517286,
+      "loss": 0.7276,
+      "step": 2258
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012775288092189502,
+      "loss": 0.7136,
+      "step": 2259
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012772087067861715,
+      "loss": 0.7546,
+      "step": 2260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001276888604353393,
+      "loss": 0.7768,
+      "step": 2261
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012765685019206147,
+      "loss": 0.8184,
+      "step": 2262
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012762483994878363,
+      "loss": 0.8803,
+      "step": 2263
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012759282970550576,
+      "loss": 0.6996,
+      "step": 2264
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001275608194622279,
+      "loss": 0.8094,
+      "step": 2265
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012752880921895008,
+      "loss": 0.8643,
+      "step": 2266
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012749679897567222,
+      "loss": 0.8093,
+      "step": 2267
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00012746478873239438,
+      "loss": 0.7798,
+      "step": 2268
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001274327784891165,
+      "loss": 0.805,
+      "step": 2269
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012740076824583867,
+      "loss": 0.8661,
+      "step": 2270
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012736875800256083,
+      "loss": 0.7384,
+      "step": 2271
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012733674775928296,
+      "loss": 0.7012,
+      "step": 2272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012730473751600512,
+      "loss": 0.821,
+      "step": 2273
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012727272727272728,
+      "loss": 0.7364,
+      "step": 2274
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012724071702944941,
+      "loss": 0.7597,
+      "step": 2275
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012720870678617157,
+      "loss": 0.8765,
+      "step": 2276
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012717669654289373,
+      "loss": 0.8899,
+      "step": 2277
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0001271446862996159,
+      "loss": 0.9863,
+      "step": 2278
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012711267605633803,
+      "loss": 0.7546,
+      "step": 2279
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012708066581306019,
+      "loss": 0.8696,
+      "step": 2280
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00012704865556978235,
+      "loss": 0.8557,
+      "step": 2281
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012701664532650448,
+      "loss": 0.6489,
+      "step": 2282
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012698463508322664,
+      "loss": 0.7624,
+      "step": 2283
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001269526248399488,
+      "loss": 0.7215,
+      "step": 2284
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012692061459667096,
+      "loss": 0.7462,
+      "step": 2285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001268886043533931,
+      "loss": 0.741,
+      "step": 2286
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012685659411011522,
+      "loss": 0.8578,
+      "step": 2287
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001268245838668374,
+      "loss": 0.7616,
+      "step": 2288
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012679257362355954,
+      "loss": 0.7315,
+      "step": 2289
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001267605633802817,
+      "loss": 0.7028,
+      "step": 2290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012672855313700384,
+      "loss": 0.867,
+      "step": 2291
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.000126696542893726,
+      "loss": 0.7759,
+      "step": 2292
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00012666453265044815,
+      "loss": 0.7721,
+      "step": 2293
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001266325224071703,
+      "loss": 0.851,
+      "step": 2294
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012660051216389245,
+      "loss": 0.8215,
+      "step": 2295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001265685019206146,
+      "loss": 0.8142,
+      "step": 2296
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012653649167733674,
+      "loss": 0.8249,
+      "step": 2297
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001265044814340589,
+      "loss": 0.8673,
+      "step": 2298
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012647247119078106,
+      "loss": 0.7959,
+      "step": 2299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012644046094750322,
+      "loss": 0.7484,
+      "step": 2300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012640845070422535,
+      "loss": 0.7854,
+      "step": 2301
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001263764404609475,
+      "loss": 0.8113,
+      "step": 2302
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012634443021766967,
+      "loss": 0.858,
+      "step": 2303
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001263124199743918,
+      "loss": 0.7071,
+      "step": 2304
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012628040973111396,
+      "loss": 0.7705,
+      "step": 2305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00012624839948783612,
+      "loss": 0.7816,
+      "step": 2306
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012621638924455828,
+      "loss": 0.8441,
+      "step": 2307
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012618437900128042,
+      "loss": 0.7452,
+      "step": 2308
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012615236875800255,
+      "loss": 0.6828,
+      "step": 2309
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012612035851472474,
+      "loss": 0.7328,
+      "step": 2310
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012608834827144687,
+      "loss": 0.8251,
+      "step": 2311
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012605633802816903,
+      "loss": 0.8271,
+      "step": 2312
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012602432778489116,
+      "loss": 0.9255,
+      "step": 2313
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012599231754161332,
+      "loss": 0.8365,
+      "step": 2314
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012596030729833548,
+      "loss": 0.8354,
+      "step": 2315
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001259282970550576,
+      "loss": 0.8308,
+      "step": 2316
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012589628681177977,
+      "loss": 0.7735,
+      "step": 2317
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00012586427656850193,
+      "loss": 0.8198,
+      "step": 2318
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012583226632522407,
+      "loss": 0.8061,
+      "step": 2319
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012580025608194622,
+      "loss": 0.7423,
+      "step": 2320
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012576824583866838,
+      "loss": 0.8387,
+      "step": 2321
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012573623559539054,
+      "loss": 0.8104,
+      "step": 2322
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012570422535211268,
+      "loss": 0.6895,
+      "step": 2323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001256722151088348,
+      "loss": 0.7788,
+      "step": 2324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.000125640204865557,
+      "loss": 0.6953,
+      "step": 2325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012560819462227913,
+      "loss": 0.8365,
+      "step": 2326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001255761843790013,
+      "loss": 0.6794,
+      "step": 2327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012554417413572345,
+      "loss": 0.7183,
+      "step": 2328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012551216389244558,
+      "loss": 0.7555,
+      "step": 2329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012548015364916774,
+      "loss": 0.6211,
+      "step": 2330
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00012544814340588987,
+      "loss": 0.7012,
+      "step": 2331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012541613316261206,
+      "loss": 0.8457,
+      "step": 2332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001253841229193342,
+      "loss": 0.7994,
+      "step": 2333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012535211267605635,
+      "loss": 0.8869,
+      "step": 2334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012532010243277849,
+      "loss": 0.7663,
+      "step": 2335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012528809218950065,
+      "loss": 0.7474,
+      "step": 2336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001252560819462228,
+      "loss": 0.7913,
+      "step": 2337
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012522407170294494,
+      "loss": 0.716,
+      "step": 2338
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001251920614596671,
+      "loss": 0.7733,
+      "step": 2339
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012516005121638926,
+      "loss": 0.8193,
+      "step": 2340
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001251280409731114,
+      "loss": 0.7387,
+      "step": 2341
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00012509603072983355,
+      "loss": 0.7716,
+      "step": 2342
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001250640204865557,
+      "loss": 0.8527,
+      "step": 2343
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012503201024327787,
+      "loss": 0.8887,
+      "step": 2344
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000125,
+      "loss": 0.8372,
+      "step": 2345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012496798975672214,
+      "loss": 0.7878,
+      "step": 2346
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012493597951344432,
+      "loss": 0.8135,
+      "step": 2347
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012490396927016645,
+      "loss": 0.7553,
+      "step": 2348
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012487195902688861,
+      "loss": 0.7788,
+      "step": 2349
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012483994878361075,
+      "loss": 0.702,
+      "step": 2350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001248079385403329,
+      "loss": 0.7777,
+      "step": 2351
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012477592829705507,
+      "loss": 0.7782,
+      "step": 2352
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001247439180537772,
+      "loss": 0.7064,
+      "step": 2353
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001247119078104994,
+      "loss": 0.87,
+      "step": 2354
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012467989756722152,
+      "loss": 0.7461,
+      "step": 2355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00012464788732394365,
+      "loss": 0.7626,
+      "step": 2356
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001246158770806658,
+      "loss": 0.8239,
+      "step": 2357
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012458386683738797,
+      "loss": 0.8218,
+      "step": 2358
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012455185659411013,
+      "loss": 0.8193,
+      "step": 2359
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012451984635083226,
+      "loss": 0.7553,
+      "step": 2360
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012448783610755442,
+      "loss": 0.8027,
+      "step": 2361
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012445582586427658,
+      "loss": 0.742,
+      "step": 2362
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012442381562099872,
+      "loss": 0.7981,
+      "step": 2363
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012439180537772088,
+      "loss": 0.7316,
+      "step": 2364
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012435979513444304,
+      "loss": 0.855,
+      "step": 2365
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001243277848911652,
+      "loss": 0.8466,
+      "step": 2366
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012429577464788733,
+      "loss": 0.8062,
+      "step": 2367
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00012426376440460946,
+      "loss": 0.828,
+      "step": 2368
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012423175416133165,
+      "loss": 0.8328,
+      "step": 2369
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012419974391805378,
+      "loss": 0.8228,
+      "step": 2370
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012416773367477594,
+      "loss": 0.7368,
+      "step": 2371
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012413572343149807,
+      "loss": 0.6311,
+      "step": 2372
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012410371318822023,
+      "loss": 0.8687,
+      "step": 2373
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001240717029449424,
+      "loss": 0.7835,
+      "step": 2374
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012403969270166453,
+      "loss": 0.7978,
+      "step": 2375
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012400768245838668,
+      "loss": 0.7309,
+      "step": 2376
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012397567221510884,
+      "loss": 0.912,
+      "step": 2377
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012394366197183098,
+      "loss": 0.8043,
+      "step": 2378
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012391165172855314,
+      "loss": 0.7248,
+      "step": 2379
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001238796414852753,
+      "loss": 0.7984,
+      "step": 2380
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00012384763124199746,
+      "loss": 0.6796,
+      "step": 2381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001238156209987196,
+      "loss": 0.7677,
+      "step": 2382
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012378361075544175,
+      "loss": 0.6908,
+      "step": 2383
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001237516005121639,
+      "loss": 0.7271,
+      "step": 2384
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012371959026888604,
+      "loss": 0.8619,
+      "step": 2385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001236875800256082,
+      "loss": 0.8978,
+      "step": 2386
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012365556978233036,
+      "loss": 0.8246,
+      "step": 2387
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012362355953905252,
+      "loss": 0.77,
+      "step": 2388
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012359154929577465,
+      "loss": 0.8321,
+      "step": 2389
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012355953905249679,
+      "loss": 0.8505,
+      "step": 2390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012352752880921897,
+      "loss": 0.8258,
+      "step": 2391
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.0001234955185659411,
+      "loss": 0.7932,
+      "step": 2392
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00012346350832266327,
+      "loss": 0.8022,
+      "step": 2393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001234314980793854,
+      "loss": 0.8167,
+      "step": 2394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012339948783610756,
+      "loss": 0.9885,
+      "step": 2395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012336747759282972,
+      "loss": 0.8107,
+      "step": 2396
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012333546734955185,
+      "loss": 0.834,
+      "step": 2397
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.000123303457106274,
+      "loss": 0.8665,
+      "step": 2398
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012327144686299617,
+      "loss": 0.8698,
+      "step": 2399
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0001232394366197183,
+      "loss": 0.6976,
+      "step": 2400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012320742637644046,
+      "loss": 0.6956,
+      "step": 2401
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012317541613316262,
+      "loss": 0.7654,
+      "step": 2402
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012314340588988478,
+      "loss": 0.79,
+      "step": 2403
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012311139564660691,
+      "loss": 0.7098,
+      "step": 2404
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012307938540332907,
+      "loss": 0.6704,
+      "step": 2405
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00012304737516005123,
+      "loss": 0.7923,
+      "step": 2406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012301536491677337,
+      "loss": 0.7044,
+      "step": 2407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012298335467349553,
+      "loss": 0.8322,
+      "step": 2408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001229513444302177,
+      "loss": 0.7624,
+      "step": 2409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012291933418693982,
+      "loss": 0.8066,
+      "step": 2410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012288732394366198,
+      "loss": 0.7718,
+      "step": 2411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001228553137003841,
+      "loss": 0.7995,
+      "step": 2412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001228233034571063,
+      "loss": 0.8573,
+      "step": 2413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012279129321382843,
+      "loss": 0.8743,
+      "step": 2414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.0001227592829705506,
+      "loss": 0.8752,
+      "step": 2415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012272727272727272,
+      "loss": 0.7974,
+      "step": 2416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012269526248399488,
+      "loss": 0.7705,
+      "step": 2417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00012266325224071704,
+      "loss": 0.9263,
+      "step": 2418
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012263124199743918,
+      "loss": 0.8065,
+      "step": 2419
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012259923175416134,
+      "loss": 0.7873,
+      "step": 2420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001225672215108835,
+      "loss": 0.8183,
+      "step": 2421
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012253521126760563,
+      "loss": 0.7773,
+      "step": 2422
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001225032010243278,
+      "loss": 0.8035,
+      "step": 2423
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012247119078104995,
+      "loss": 0.8513,
+      "step": 2424
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001224391805377721,
+      "loss": 0.8112,
+      "step": 2425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012240717029449424,
+      "loss": 0.6508,
+      "step": 2426
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012237516005121637,
+      "loss": 0.7103,
+      "step": 2427
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012234314980793856,
+      "loss": 0.7747,
+      "step": 2428
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001223111395646607,
+      "loss": 0.8929,
+      "step": 2429
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00012227912932138285,
+      "loss": 0.7981,
+      "step": 2430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.000122247119078105,
+      "loss": 0.8632,
+      "step": 2431
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012221510883482714,
+      "loss": 0.9111,
+      "step": 2432
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001221830985915493,
+      "loss": 0.7826,
+      "step": 2433
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012215108834827144,
+      "loss": 0.8707,
+      "step": 2434
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012211907810499362,
+      "loss": 0.7968,
+      "step": 2435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012208706786171576,
+      "loss": 0.7949,
+      "step": 2436
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012205505761843789,
+      "loss": 0.8064,
+      "step": 2437
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012202304737516006,
+      "loss": 0.7231,
+      "step": 2438
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001219910371318822,
+      "loss": 0.7804,
+      "step": 2439
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012195902688860437,
+      "loss": 0.7764,
+      "step": 2440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0001219270166453265,
+      "loss": 0.6856,
+      "step": 2441
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012189500640204865,
+      "loss": 0.8854,
+      "step": 2442
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00012186299615877082,
+      "loss": 0.8723,
+      "step": 2443
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012183098591549295,
+      "loss": 0.7711,
+      "step": 2444
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012179897567221513,
+      "loss": 0.7159,
+      "step": 2445
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012176696542893726,
+      "loss": 0.7352,
+      "step": 2446
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012173495518565943,
+      "loss": 0.672,
+      "step": 2447
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012170294494238157,
+      "loss": 0.7927,
+      "step": 2448
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012167093469910371,
+      "loss": 0.763,
+      "step": 2449
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012163892445582587,
+      "loss": 0.7552,
+      "step": 2450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012160691421254802,
+      "loss": 0.7458,
+      "step": 2451
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012157490396927018,
+      "loss": 0.8155,
+      "step": 2452
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012154289372599232,
+      "loss": 0.8065,
+      "step": 2453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012151088348271447,
+      "loss": 0.8267,
+      "step": 2454
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012147887323943663,
+      "loss": 0.6709,
+      "step": 2455
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00012144686299615878,
+      "loss": 0.8355,
+      "step": 2456
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012141485275288094,
+      "loss": 0.7979,
+      "step": 2457
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012138284250960308,
+      "loss": 0.809,
+      "step": 2458
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012135083226632522,
+      "loss": 0.6368,
+      "step": 2459
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012131882202304739,
+      "loss": 0.9048,
+      "step": 2460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012128681177976952,
+      "loss": 0.7964,
+      "step": 2461
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001212548015364917,
+      "loss": 0.7142,
+      "step": 2462
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012122279129321383,
+      "loss": 0.8145,
+      "step": 2463
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012119078104993597,
+      "loss": 0.8478,
+      "step": 2464
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012115877080665813,
+      "loss": 0.7651,
+      "step": 2465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012112676056338028,
+      "loss": 0.7467,
+      "step": 2466
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012109475032010244,
+      "loss": 0.6697,
+      "step": 2467
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00012106274007682459,
+      "loss": 0.7564,
+      "step": 2468
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012103072983354673,
+      "loss": 0.8616,
+      "step": 2469
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012099871959026889,
+      "loss": 0.7635,
+      "step": 2470
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012096670934699104,
+      "loss": 0.8381,
+      "step": 2471
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001209346991037132,
+      "loss": 0.9119,
+      "step": 2472
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012090268886043534,
+      "loss": 0.8192,
+      "step": 2473
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001208706786171575,
+      "loss": 0.6943,
+      "step": 2474
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012083866837387965,
+      "loss": 0.6984,
+      "step": 2475
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001208066581306018,
+      "loss": 0.7814,
+      "step": 2476
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012077464788732396,
+      "loss": 0.7123,
+      "step": 2477
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001207426376440461,
+      "loss": 0.9142,
+      "step": 2478
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012071062740076826,
+      "loss": 0.7761,
+      "step": 2479
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012067861715749041,
+      "loss": 0.9016,
+      "step": 2480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00012064660691421254,
+      "loss": 0.744,
+      "step": 2481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012061459667093471,
+      "loss": 0.8666,
+      "step": 2482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012058258642765685,
+      "loss": 0.7122,
+      "step": 2483
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012055057618437902,
+      "loss": 0.7377,
+      "step": 2484
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012051856594110115,
+      "loss": 0.8891,
+      "step": 2485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001204865556978233,
+      "loss": 0.7556,
+      "step": 2486
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012045454545454546,
+      "loss": 0.885,
+      "step": 2487
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001204225352112676,
+      "loss": 0.6879,
+      "step": 2488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012039052496798976,
+      "loss": 0.8877,
+      "step": 2489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012035851472471191,
+      "loss": 0.7082,
+      "step": 2490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012032650448143406,
+      "loss": 0.7311,
+      "step": 2491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012029449423815622,
+      "loss": 0.7437,
+      "step": 2492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00012026248399487836,
+      "loss": 0.8843,
+      "step": 2493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012023047375160052,
+      "loss": 0.8721,
+      "step": 2494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012019846350832267,
+      "loss": 0.8123,
+      "step": 2495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012016645326504482,
+      "loss": 0.7354,
+      "step": 2496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012013444302176698,
+      "loss": 0.8254,
+      "step": 2497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012010243277848912,
+      "loss": 0.7907,
+      "step": 2498
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012007042253521128,
+      "loss": 0.8544,
+      "step": 2499
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012003841229193343,
+      "loss": 0.805,
+      "step": 2500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00012000640204865559,
+      "loss": 0.7642,
+      "step": 2501
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011997439180537773,
+      "loss": 0.7418,
+      "step": 2502
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011994238156209987,
+      "loss": 0.8139,
+      "step": 2503
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011991037131882204,
+      "loss": 0.7301,
+      "step": 2504
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011987836107554417,
+      "loss": 0.8319,
+      "step": 2505
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00011984635083226635,
+      "loss": 0.8436,
+      "step": 2506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011981434058898848,
+      "loss": 0.7254,
+      "step": 2507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011978233034571062,
+      "loss": 0.7714,
+      "step": 2508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011975032010243278,
+      "loss": 0.7144,
+      "step": 2509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011971830985915493,
+      "loss": 0.7382,
+      "step": 2510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011968629961587709,
+      "loss": 0.7934,
+      "step": 2511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011965428937259924,
+      "loss": 0.7392,
+      "step": 2512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011962227912932138,
+      "loss": 0.6926,
+      "step": 2513
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011959026888604354,
+      "loss": 0.8156,
+      "step": 2514
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011955825864276569,
+      "loss": 0.8945,
+      "step": 2515
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011952624839948785,
+      "loss": 0.6613,
+      "step": 2516
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011949423815621,
+      "loss": 0.7856,
+      "step": 2517
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00011946222791293213,
+      "loss": 0.7174,
+      "step": 2518
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001194302176696543,
+      "loss": 0.7856,
+      "step": 2519
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011939820742637645,
+      "loss": 0.7686,
+      "step": 2520
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011936619718309861,
+      "loss": 0.7643,
+      "step": 2521
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011933418693982075,
+      "loss": 0.6553,
+      "step": 2522
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011930217669654289,
+      "loss": 0.6704,
+      "step": 2523
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011927016645326506,
+      "loss": 0.6359,
+      "step": 2524
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011923815620998719,
+      "loss": 0.67,
+      "step": 2525
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011920614596670937,
+      "loss": 0.8417,
+      "step": 2526
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001191741357234315,
+      "loss": 0.7306,
+      "step": 2527
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011914212548015367,
+      "loss": 0.7274,
+      "step": 2528
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001191101152368758,
+      "loss": 0.6903,
+      "step": 2529
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011907810499359795,
+      "loss": 0.7844,
+      "step": 2530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00011904609475032011,
+      "loss": 0.7448,
+      "step": 2531
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011901408450704226,
+      "loss": 0.7006,
+      "step": 2532
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011898207426376442,
+      "loss": 0.6626,
+      "step": 2533
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011895006402048656,
+      "loss": 0.6921,
+      "step": 2534
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011891805377720871,
+      "loss": 0.7757,
+      "step": 2535
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011888604353393087,
+      "loss": 0.7851,
+      "step": 2536
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011885403329065301,
+      "loss": 0.6929,
+      "step": 2537
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011882202304737517,
+      "loss": 0.7555,
+      "step": 2538
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011879001280409732,
+      "loss": 0.6841,
+      "step": 2539
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011875800256081945,
+      "loss": 0.7737,
+      "step": 2540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011872599231754163,
+      "loss": 0.7018,
+      "step": 2541
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011869398207426376,
+      "loss": 0.682,
+      "step": 2542
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00011866197183098593,
+      "loss": 0.725,
+      "step": 2543
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011862996158770807,
+      "loss": 0.6858,
+      "step": 2544
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011859795134443021,
+      "loss": 0.755,
+      "step": 2545
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011856594110115238,
+      "loss": 0.7388,
+      "step": 2546
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011853393085787452,
+      "loss": 0.8237,
+      "step": 2547
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011850192061459669,
+      "loss": 0.7838,
+      "step": 2548
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011846991037131882,
+      "loss": 0.6704,
+      "step": 2549
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011843790012804097,
+      "loss": 0.6804,
+      "step": 2550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011840588988476313,
+      "loss": 0.7985,
+      "step": 2551
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011837387964148528,
+      "loss": 0.6407,
+      "step": 2552
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011834186939820744,
+      "loss": 0.7917,
+      "step": 2553
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011830985915492958,
+      "loss": 0.6359,
+      "step": 2554
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011827784891165174,
+      "loss": 0.6689,
+      "step": 2555
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00011824583866837389,
+      "loss": 0.6379,
+      "step": 2556
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011821382842509603,
+      "loss": 0.6704,
+      "step": 2557
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001181818181818182,
+      "loss": 0.776,
+      "step": 2558
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011814980793854034,
+      "loss": 0.7535,
+      "step": 2559
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001181177976952625,
+      "loss": 0.7159,
+      "step": 2560
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011808578745198465,
+      "loss": 0.7401,
+      "step": 2561
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011805377720870678,
+      "loss": 0.8352,
+      "step": 2562
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011802176696542895,
+      "loss": 0.8276,
+      "step": 2563
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011798975672215108,
+      "loss": 0.872,
+      "step": 2564
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011795774647887326,
+      "loss": 0.6557,
+      "step": 2565
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011792573623559539,
+      "loss": 0.698,
+      "step": 2566
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00011789372599231754,
+      "loss": 0.7534,
+      "step": 2567
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0001178617157490397,
+      "loss": 0.8154,
+      "step": 2568
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011782970550576184,
+      "loss": 0.7103,
+      "step": 2569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.000117797695262484,
+      "loss": 0.7232,
+      "step": 2570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011776568501920615,
+      "loss": 0.7581,
+      "step": 2571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001177336747759283,
+      "loss": 0.8078,
+      "step": 2572
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011770166453265045,
+      "loss": 0.7282,
+      "step": 2573
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001176696542893726,
+      "loss": 0.7558,
+      "step": 2574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011763764404609476,
+      "loss": 0.7168,
+      "step": 2575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011760563380281691,
+      "loss": 0.6692,
+      "step": 2576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011757362355953905,
+      "loss": 0.8424,
+      "step": 2577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011754161331626121,
+      "loss": 0.8837,
+      "step": 2578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011750960307298336,
+      "loss": 0.9305,
+      "step": 2579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011747759282970552,
+      "loss": 0.7255,
+      "step": 2580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00011744558258642767,
+      "loss": 0.7444,
+      "step": 2581
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001174135723431498,
+      "loss": 0.7215,
+      "step": 2582
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011738156209987197,
+      "loss": 0.6114,
+      "step": 2583
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.0001173495518565941,
+      "loss": 0.8114,
+      "step": 2584
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011731754161331628,
+      "loss": 0.7906,
+      "step": 2585
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011728553137003841,
+      "loss": 0.7395,
+      "step": 2586
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011725352112676058,
+      "loss": 0.912,
+      "step": 2587
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011722151088348272,
+      "loss": 0.6466,
+      "step": 2588
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011718950064020486,
+      "loss": 0.92,
+      "step": 2589
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011715749039692702,
+      "loss": 0.6754,
+      "step": 2590
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011712548015364917,
+      "loss": 0.7114,
+      "step": 2591
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011709346991037133,
+      "loss": 0.7,
+      "step": 2592
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00011706145966709347,
+      "loss": 0.7569,
+      "step": 2593
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011702944942381562,
+      "loss": 0.841,
+      "step": 2594
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011699743918053778,
+      "loss": 0.7138,
+      "step": 2595
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011696542893725993,
+      "loss": 0.828,
+      "step": 2596
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011693341869398209,
+      "loss": 0.6626,
+      "step": 2597
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011690140845070423,
+      "loss": 0.8019,
+      "step": 2598
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011686939820742638,
+      "loss": 0.6483,
+      "step": 2599
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011683738796414854,
+      "loss": 0.7735,
+      "step": 2600
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011680537772087068,
+      "loss": 0.6717,
+      "step": 2601
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011677336747759284,
+      "loss": 0.7058,
+      "step": 2602
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011674135723431499,
+      "loss": 0.7001,
+      "step": 2603
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011670934699103712,
+      "loss": 0.783,
+      "step": 2604
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001166773367477593,
+      "loss": 0.6655,
+      "step": 2605
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00011664532650448143,
+      "loss": 0.8735,
+      "step": 2606
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001166133162612036,
+      "loss": 0.749,
+      "step": 2607
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011658130601792574,
+      "loss": 0.7387,
+      "step": 2608
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011654929577464788,
+      "loss": 0.6799,
+      "step": 2609
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011651728553137004,
+      "loss": 0.7341,
+      "step": 2610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011648527528809219,
+      "loss": 0.7244,
+      "step": 2611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011645326504481435,
+      "loss": 0.8251,
+      "step": 2612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001164212548015365,
+      "loss": 0.7844,
+      "step": 2613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011638924455825865,
+      "loss": 0.7674,
+      "step": 2614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001163572343149808,
+      "loss": 0.7316,
+      "step": 2615
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011632522407170295,
+      "loss": 0.7028,
+      "step": 2616
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001162932138284251,
+      "loss": 0.758,
+      "step": 2617
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00011626120358514725,
+      "loss": 0.8201,
+      "step": 2618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011622919334186941,
+      "loss": 0.8022,
+      "step": 2619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011619718309859156,
+      "loss": 0.7581,
+      "step": 2620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011616517285531369,
+      "loss": 0.648,
+      "step": 2621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011613316261203586,
+      "loss": 0.5945,
+      "step": 2622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011610115236875801,
+      "loss": 0.6554,
+      "step": 2623
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011606914212548017,
+      "loss": 0.7788,
+      "step": 2624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011603713188220232,
+      "loss": 0.7396,
+      "step": 2625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011600512163892445,
+      "loss": 0.7621,
+      "step": 2626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011597311139564662,
+      "loss": 0.7778,
+      "step": 2627
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011594110115236876,
+      "loss": 0.7829,
+      "step": 2628
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011590909090909093,
+      "loss": 0.8175,
+      "step": 2629
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011587708066581306,
+      "loss": 0.7608,
+      "step": 2630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00011584507042253521,
+      "loss": 0.7722,
+      "step": 2631
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011581306017925737,
+      "loss": 0.7389,
+      "step": 2632
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011578104993597951,
+      "loss": 0.6949,
+      "step": 2633
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011574903969270167,
+      "loss": 0.7604,
+      "step": 2634
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011571702944942382,
+      "loss": 0.7373,
+      "step": 2635
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011568501920614597,
+      "loss": 0.7898,
+      "step": 2636
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011565300896286813,
+      "loss": 0.7864,
+      "step": 2637
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011562099871959027,
+      "loss": 0.6213,
+      "step": 2638
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011558898847631243,
+      "loss": 0.7132,
+      "step": 2639
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011555697823303458,
+      "loss": 0.7971,
+      "step": 2640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011552496798975674,
+      "loss": 0.694,
+      "step": 2641
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011549295774647888,
+      "loss": 0.7641,
+      "step": 2642
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00011546094750320102,
+      "loss": 0.8138,
+      "step": 2643
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011542893725992319,
+      "loss": 0.6394,
+      "step": 2644
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011539692701664532,
+      "loss": 0.7759,
+      "step": 2645
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001153649167733675,
+      "loss": 0.7264,
+      "step": 2646
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011533290653008963,
+      "loss": 0.7146,
+      "step": 2647
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011530089628681177,
+      "loss": 0.8451,
+      "step": 2648
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011526888604353395,
+      "loss": 0.8832,
+      "step": 2649
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011523687580025608,
+      "loss": 0.7071,
+      "step": 2650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011520486555697825,
+      "loss": 0.8829,
+      "step": 2651
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011517285531370039,
+      "loss": 0.7301,
+      "step": 2652
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011514084507042253,
+      "loss": 0.7227,
+      "step": 2653
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011510883482714469,
+      "loss": 0.7201,
+      "step": 2654
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00011507682458386684,
+      "loss": 0.7937,
+      "step": 2655
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.000115044814340589,
+      "loss": 0.8179,
+      "step": 2656
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011501280409731114,
+      "loss": 0.77,
+      "step": 2657
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011498079385403329,
+      "loss": 0.6928,
+      "step": 2658
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011494878361075545,
+      "loss": 0.6454,
+      "step": 2659
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001149167733674776,
+      "loss": 0.7725,
+      "step": 2660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011488476312419976,
+      "loss": 0.9504,
+      "step": 2661
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001148527528809219,
+      "loss": 0.719,
+      "step": 2662
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011482074263764404,
+      "loss": 0.7262,
+      "step": 2663
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011478873239436621,
+      "loss": 0.7349,
+      "step": 2664
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011475672215108834,
+      "loss": 0.63,
+      "step": 2665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011472471190781052,
+      "loss": 0.7517,
+      "step": 2666
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011469270166453265,
+      "loss": 0.6737,
+      "step": 2667
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00011466069142125482,
+      "loss": 0.806,
+      "step": 2668
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011462868117797695,
+      "loss": 0.7152,
+      "step": 2669
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001145966709346991,
+      "loss": 0.6558,
+      "step": 2670
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011456466069142126,
+      "loss": 0.8174,
+      "step": 2671
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001145326504481434,
+      "loss": 0.6523,
+      "step": 2672
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011450064020486557,
+      "loss": 0.8625,
+      "step": 2673
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011446862996158771,
+      "loss": 0.7599,
+      "step": 2674
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011443661971830986,
+      "loss": 0.8176,
+      "step": 2675
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011440460947503202,
+      "loss": 0.807,
+      "step": 2676
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011437259923175416,
+      "loss": 0.7015,
+      "step": 2677
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011434058898847632,
+      "loss": 0.7423,
+      "step": 2678
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011430857874519847,
+      "loss": 0.7265,
+      "step": 2679
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011427656850192062,
+      "loss": 0.7397,
+      "step": 2680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00011424455825864278,
+      "loss": 0.6523,
+      "step": 2681
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011421254801536492,
+      "loss": 0.8245,
+      "step": 2682
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011418053777208708,
+      "loss": 0.8496,
+      "step": 2683
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011414852752880923,
+      "loss": 0.6092,
+      "step": 2684
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011411651728553136,
+      "loss": 0.7027,
+      "step": 2685
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011408450704225353,
+      "loss": 0.6607,
+      "step": 2686
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011405249679897567,
+      "loss": 0.7962,
+      "step": 2687
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011402048655569784,
+      "loss": 0.7765,
+      "step": 2688
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011398847631241997,
+      "loss": 0.7252,
+      "step": 2689
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011395646606914212,
+      "loss": 0.7404,
+      "step": 2690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011392445582586428,
+      "loss": 0.7645,
+      "step": 2691
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011389244558258643,
+      "loss": 0.8634,
+      "step": 2692
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00011386043533930859,
+      "loss": 0.7868,
+      "step": 2693
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011382842509603073,
+      "loss": 0.8595,
+      "step": 2694
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011379641485275289,
+      "loss": 0.724,
+      "step": 2695
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011376440460947504,
+      "loss": 0.7439,
+      "step": 2696
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011373239436619718,
+      "loss": 0.766,
+      "step": 2697
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011370038412291934,
+      "loss": 0.7569,
+      "step": 2698
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011366837387964149,
+      "loss": 0.8244,
+      "step": 2699
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011363636363636365,
+      "loss": 0.7504,
+      "step": 2700
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001136043533930858,
+      "loss": 0.7657,
+      "step": 2701
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011357234314980794,
+      "loss": 0.7275,
+      "step": 2702
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0001135403329065301,
+      "loss": 0.6684,
+      "step": 2703
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011350832266325225,
+      "loss": 0.7726,
+      "step": 2704
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011347631241997441,
+      "loss": 0.7801,
+      "step": 2705
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00011344430217669655,
+      "loss": 0.6749,
+      "step": 2706
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011341229193341869,
+      "loss": 0.7266,
+      "step": 2707
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011338028169014086,
+      "loss": 0.8835,
+      "step": 2708
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011334827144686299,
+      "loss": 0.6929,
+      "step": 2709
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011331626120358517,
+      "loss": 0.7982,
+      "step": 2710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001132842509603073,
+      "loss": 0.7641,
+      "step": 2711
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011325224071702945,
+      "loss": 0.7761,
+      "step": 2712
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001132202304737516,
+      "loss": 0.7583,
+      "step": 2713
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011318822023047375,
+      "loss": 0.6922,
+      "step": 2714
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011315620998719591,
+      "loss": 0.6959,
+      "step": 2715
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011312419974391806,
+      "loss": 0.7338,
+      "step": 2716
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001130921895006402,
+      "loss": 0.8185,
+      "step": 2717
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00011306017925736236,
+      "loss": 0.7502,
+      "step": 2718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011302816901408451,
+      "loss": 0.5995,
+      "step": 2719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011299615877080667,
+      "loss": 0.6897,
+      "step": 2720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011296414852752882,
+      "loss": 0.7661,
+      "step": 2721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011293213828425098,
+      "loss": 0.7784,
+      "step": 2722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011290012804097312,
+      "loss": 0.7971,
+      "step": 2723
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011286811779769525,
+      "loss": 0.8032,
+      "step": 2724
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011283610755441743,
+      "loss": 0.8649,
+      "step": 2725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011280409731113957,
+      "loss": 0.6978,
+      "step": 2726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011277208706786173,
+      "loss": 0.7986,
+      "step": 2727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011274007682458388,
+      "loss": 0.6867,
+      "step": 2728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011270806658130601,
+      "loss": 0.7625,
+      "step": 2729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011267605633802819,
+      "loss": 0.8267,
+      "step": 2730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00011264404609475032,
+      "loss": 0.7203,
+      "step": 2731
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011261203585147249,
+      "loss": 0.689,
+      "step": 2732
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011258002560819462,
+      "loss": 0.7133,
+      "step": 2733
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011254801536491677,
+      "loss": 0.6844,
+      "step": 2734
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011251600512163893,
+      "loss": 0.8168,
+      "step": 2735
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011248399487836108,
+      "loss": 0.6101,
+      "step": 2736
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011245198463508324,
+      "loss": 0.8262,
+      "step": 2737
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011241997439180538,
+      "loss": 0.8676,
+      "step": 2738
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011238796414852753,
+      "loss": 0.792,
+      "step": 2739
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011235595390524969,
+      "loss": 0.6827,
+      "step": 2740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011232394366197184,
+      "loss": 0.79,
+      "step": 2741
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.000112291933418694,
+      "loss": 0.8492,
+      "step": 2742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00011225992317541614,
+      "loss": 0.6781,
+      "step": 2743
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011222791293213827,
+      "loss": 0.7206,
+      "step": 2744
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011219590268886045,
+      "loss": 0.7191,
+      "step": 2745
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011216389244558258,
+      "loss": 0.7237,
+      "step": 2746
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011213188220230475,
+      "loss": 0.6823,
+      "step": 2747
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011209987195902689,
+      "loss": 0.8338,
+      "step": 2748
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011206786171574903,
+      "loss": 0.6784,
+      "step": 2749
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011203585147247119,
+      "loss": 0.7582,
+      "step": 2750
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011200384122919334,
+      "loss": 0.7048,
+      "step": 2751
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001119718309859155,
+      "loss": 0.7567,
+      "step": 2752
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011193982074263764,
+      "loss": 0.8061,
+      "step": 2753
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011190781049935982,
+      "loss": 0.8665,
+      "step": 2754
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00011187580025608195,
+      "loss": 0.7462,
+      "step": 2755
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001118437900128041,
+      "loss": 0.6521,
+      "step": 2756
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011181177976952626,
+      "loss": 0.7702,
+      "step": 2757
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001117797695262484,
+      "loss": 0.6603,
+      "step": 2758
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011174775928297056,
+      "loss": 0.655,
+      "step": 2759
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011171574903969271,
+      "loss": 0.6808,
+      "step": 2760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011168373879641485,
+      "loss": 0.7912,
+      "step": 2761
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011165172855313701,
+      "loss": 0.7383,
+      "step": 2762
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011161971830985916,
+      "loss": 0.7879,
+      "step": 2763
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011158770806658132,
+      "loss": 0.846,
+      "step": 2764
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011155569782330347,
+      "loss": 0.754,
+      "step": 2765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001115236875800256,
+      "loss": 0.7713,
+      "step": 2766
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00011149167733674777,
+      "loss": 0.8566,
+      "step": 2767
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001114596670934699,
+      "loss": 0.7089,
+      "step": 2768
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011142765685019208,
+      "loss": 0.7885,
+      "step": 2769
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011139564660691421,
+      "loss": 0.7838,
+      "step": 2770
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011136363636363636,
+      "loss": 0.7092,
+      "step": 2771
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011133162612035852,
+      "loss": 0.8319,
+      "step": 2772
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011129961587708066,
+      "loss": 0.7356,
+      "step": 2773
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011126760563380282,
+      "loss": 0.7034,
+      "step": 2774
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011123559539052497,
+      "loss": 0.8437,
+      "step": 2775
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011120358514724712,
+      "loss": 0.8333,
+      "step": 2776
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011117157490396928,
+      "loss": 0.8018,
+      "step": 2777
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011113956466069142,
+      "loss": 0.7066,
+      "step": 2778
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011110755441741358,
+      "loss": 0.6987,
+      "step": 2779
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011107554417413573,
+      "loss": 0.7384,
+      "step": 2780
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00011104353393085789,
+      "loss": 0.8401,
+      "step": 2781
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011101152368758003,
+      "loss": 0.7865,
+      "step": 2782
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011097951344430218,
+      "loss": 0.8528,
+      "step": 2783
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011094750320102434,
+      "loss": 0.822,
+      "step": 2784
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011091549295774649,
+      "loss": 0.7093,
+      "step": 2785
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011088348271446865,
+      "loss": 0.7562,
+      "step": 2786
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011085147247119079,
+      "loss": 0.7744,
+      "step": 2787
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011081946222791292,
+      "loss": 0.7343,
+      "step": 2788
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001107874519846351,
+      "loss": 0.7553,
+      "step": 2789
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011075544174135723,
+      "loss": 0.6779,
+      "step": 2790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001107234314980794,
+      "loss": 0.7096,
+      "step": 2791
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011069142125480154,
+      "loss": 0.8799,
+      "step": 2792
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00011065941101152368,
+      "loss": 0.6849,
+      "step": 2793
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011062740076824584,
+      "loss": 0.6549,
+      "step": 2794
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011059539052496799,
+      "loss": 0.7539,
+      "step": 2795
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011056338028169015,
+      "loss": 0.7529,
+      "step": 2796
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001105313700384123,
+      "loss": 0.8081,
+      "step": 2797
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011049935979513444,
+      "loss": 0.6778,
+      "step": 2798
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001104673495518566,
+      "loss": 0.8088,
+      "step": 2799
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011043533930857875,
+      "loss": 0.7295,
+      "step": 2800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011040332906530091,
+      "loss": 0.726,
+      "step": 2801
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011037131882202305,
+      "loss": 0.7224,
+      "step": 2802
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001103393085787452,
+      "loss": 0.7661,
+      "step": 2803
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011030729833546736,
+      "loss": 0.8402,
+      "step": 2804
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001102752880921895,
+      "loss": 0.6793,
+      "step": 2805
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00011024327784891167,
+      "loss": 0.8112,
+      "step": 2806
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011021126760563381,
+      "loss": 0.7924,
+      "step": 2807
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011017925736235597,
+      "loss": 0.7062,
+      "step": 2808
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011014724711907812,
+      "loss": 0.8203,
+      "step": 2809
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011011523687580025,
+      "loss": 0.6892,
+      "step": 2810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011008322663252242,
+      "loss": 0.7386,
+      "step": 2811
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011005121638924456,
+      "loss": 0.7778,
+      "step": 2812
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00011001920614596673,
+      "loss": 0.7696,
+      "step": 2813
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00010998719590268886,
+      "loss": 0.7173,
+      "step": 2814
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00010995518565941101,
+      "loss": 0.7153,
+      "step": 2815
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00010992317541613317,
+      "loss": 0.8022,
+      "step": 2816
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00010989116517285531,
+      "loss": 0.8339,
+      "step": 2817
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00010985915492957747,
+      "loss": 0.6598,
+      "step": 2818
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010982714468629962,
+      "loss": 0.812,
+      "step": 2819
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010979513444302177,
+      "loss": 0.7727,
+      "step": 2820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010976312419974393,
+      "loss": 0.7132,
+      "step": 2821
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010973111395646607,
+      "loss": 0.6565,
+      "step": 2822
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010969910371318823,
+      "loss": 0.7577,
+      "step": 2823
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010966709346991038,
+      "loss": 0.7975,
+      "step": 2824
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010963508322663251,
+      "loss": 0.9308,
+      "step": 2825
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010960307298335468,
+      "loss": 0.7287,
+      "step": 2826
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010957106274007682,
+      "loss": 0.635,
+      "step": 2827
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010953905249679899,
+      "loss": 0.7248,
+      "step": 2828
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010950704225352114,
+      "loss": 0.6549,
+      "step": 2829
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010947503201024327,
+      "loss": 0.661,
+      "step": 2830
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00010944302176696544,
+      "loss": 0.822,
+      "step": 2831
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010941101152368758,
+      "loss": 0.7901,
+      "step": 2832
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010937900128040975,
+      "loss": 0.7717,
+      "step": 2833
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010934699103713188,
+      "loss": 0.7677,
+      "step": 2834
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010931498079385406,
+      "loss": 0.7423,
+      "step": 2835
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010928297055057619,
+      "loss": 0.7465,
+      "step": 2836
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010925096030729833,
+      "loss": 0.7759,
+      "step": 2837
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001092189500640205,
+      "loss": 0.8715,
+      "step": 2838
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010918693982074264,
+      "loss": 0.9301,
+      "step": 2839
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001091549295774648,
+      "loss": 0.7621,
+      "step": 2840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010912291933418695,
+      "loss": 0.6702,
+      "step": 2841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010909090909090909,
+      "loss": 0.7293,
+      "step": 2842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00010905889884763125,
+      "loss": 0.7899,
+      "step": 2843
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001090268886043534,
+      "loss": 0.7037,
+      "step": 2844
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010899487836107556,
+      "loss": 0.8262,
+      "step": 2845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001089628681177977,
+      "loss": 0.7469,
+      "step": 2846
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010893085787451984,
+      "loss": 0.7508,
+      "step": 2847
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010889884763124201,
+      "loss": 0.8034,
+      "step": 2848
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010886683738796414,
+      "loss": 0.7848,
+      "step": 2849
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010883482714468632,
+      "loss": 0.7512,
+      "step": 2850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010880281690140845,
+      "loss": 0.8466,
+      "step": 2851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001087708066581306,
+      "loss": 0.7281,
+      "step": 2852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010873879641485276,
+      "loss": 0.7975,
+      "step": 2853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001087067861715749,
+      "loss": 0.8252,
+      "step": 2854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010867477592829706,
+      "loss": 0.6473,
+      "step": 2855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00010864276568501921,
+      "loss": 0.7509,
+      "step": 2856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010861075544174135,
+      "loss": 0.7057,
+      "step": 2857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010857874519846351,
+      "loss": 0.6553,
+      "step": 2858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010854673495518566,
+      "loss": 0.7575,
+      "step": 2859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010851472471190782,
+      "loss": 0.6393,
+      "step": 2860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010848271446862997,
+      "loss": 0.6733,
+      "step": 2861
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010845070422535213,
+      "loss": 0.6764,
+      "step": 2862
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010841869398207427,
+      "loss": 0.7881,
+      "step": 2863
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010838668373879642,
+      "loss": 0.6654,
+      "step": 2864
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010835467349551858,
+      "loss": 0.7564,
+      "step": 2865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010832266325224072,
+      "loss": 0.8032,
+      "step": 2866
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010829065300896288,
+      "loss": 0.7194,
+      "step": 2867
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00010825864276568503,
+      "loss": 0.8221,
+      "step": 2868
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010822663252240716,
+      "loss": 0.7619,
+      "step": 2869
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010819462227912934,
+      "loss": 0.7463,
+      "step": 2870
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010816261203585147,
+      "loss": 0.7136,
+      "step": 2871
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010813060179257364,
+      "loss": 0.6622,
+      "step": 2872
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010809859154929577,
+      "loss": 0.9716,
+      "step": 2873
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010806658130601792,
+      "loss": 0.7546,
+      "step": 2874
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010803457106274008,
+      "loss": 0.8198,
+      "step": 2875
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010800256081946223,
+      "loss": 0.8527,
+      "step": 2876
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010797055057618439,
+      "loss": 0.7693,
+      "step": 2877
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010793854033290653,
+      "loss": 0.7703,
+      "step": 2878
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010790653008962868,
+      "loss": 0.7257,
+      "step": 2879
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010787451984635084,
+      "loss": 0.6584,
+      "step": 2880
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00010784250960307299,
+      "loss": 0.7969,
+      "step": 2881
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010781049935979515,
+      "loss": 0.686,
+      "step": 2882
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010777848911651729,
+      "loss": 0.789,
+      "step": 2883
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010774647887323944,
+      "loss": 0.7876,
+      "step": 2884
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001077144686299616,
+      "loss": 0.7464,
+      "step": 2885
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010768245838668374,
+      "loss": 0.8391,
+      "step": 2886
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001076504481434059,
+      "loss": 0.7214,
+      "step": 2887
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010761843790012805,
+      "loss": 0.7708,
+      "step": 2888
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010758642765685021,
+      "loss": 0.7889,
+      "step": 2889
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010755441741357236,
+      "loss": 0.6609,
+      "step": 2890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010752240717029449,
+      "loss": 0.7613,
+      "step": 2891
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00010749039692701666,
+      "loss": 0.7319,
+      "step": 2892
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.0001074583866837388,
+      "loss": 0.6815,
+      "step": 2893
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010742637644046097,
+      "loss": 0.8365,
+      "step": 2894
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001073943661971831,
+      "loss": 0.6658,
+      "step": 2895
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010736235595390525,
+      "loss": 0.7065,
+      "step": 2896
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0001073303457106274,
+      "loss": 0.7503,
+      "step": 2897
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010729833546734955,
+      "loss": 0.7549,
+      "step": 2898
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010726632522407171,
+      "loss": 0.8223,
+      "step": 2899
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010723431498079386,
+      "loss": 0.7683,
+      "step": 2900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.000107202304737516,
+      "loss": 0.7744,
+      "step": 2901
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010717029449423816,
+      "loss": 0.7275,
+      "step": 2902
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010713828425096031,
+      "loss": 0.8255,
+      "step": 2903
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010710627400768247,
+      "loss": 0.7826,
+      "step": 2904
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010707426376440462,
+      "loss": 0.7893,
+      "step": 2905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00010704225352112676,
+      "loss": 0.7067,
+      "step": 2906
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010701024327784892,
+      "loss": 0.7584,
+      "step": 2907
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010697823303457107,
+      "loss": 0.7899,
+      "step": 2908
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010694622279129323,
+      "loss": 0.7133,
+      "step": 2909
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010691421254801538,
+      "loss": 0.663,
+      "step": 2910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010688220230473751,
+      "loss": 0.755,
+      "step": 2911
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010685019206145968,
+      "loss": 0.7052,
+      "step": 2912
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010681818181818181,
+      "loss": 0.6,
+      "step": 2913
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010678617157490399,
+      "loss": 0.8111,
+      "step": 2914
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010675416133162612,
+      "loss": 0.7169,
+      "step": 2915
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010672215108834827,
+      "loss": 0.8247,
+      "step": 2916
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010669014084507043,
+      "loss": 0.7434,
+      "step": 2917
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00010665813060179257,
+      "loss": 0.8262,
+      "step": 2918
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010662612035851473,
+      "loss": 0.7125,
+      "step": 2919
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010659411011523688,
+      "loss": 0.7373,
+      "step": 2920
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010656209987195904,
+      "loss": 0.6206,
+      "step": 2921
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010653008962868118,
+      "loss": 0.8749,
+      "step": 2922
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010649807938540333,
+      "loss": 0.8677,
+      "step": 2923
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010646606914212549,
+      "loss": 0.8175,
+      "step": 2924
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010643405889884764,
+      "loss": 0.8014,
+      "step": 2925
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001064020486555698,
+      "loss": 0.7386,
+      "step": 2926
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010637003841229194,
+      "loss": 0.6876,
+      "step": 2927
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010633802816901408,
+      "loss": 0.6543,
+      "step": 2928
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010630601792573625,
+      "loss": 0.7554,
+      "step": 2929
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010627400768245838,
+      "loss": 0.8674,
+      "step": 2930
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00010624199743918055,
+      "loss": 0.7695,
+      "step": 2931
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010620998719590269,
+      "loss": 0.6313,
+      "step": 2932
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010617797695262483,
+      "loss": 0.6583,
+      "step": 2933
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.000106145966709347,
+      "loss": 0.8419,
+      "step": 2934
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010611395646606914,
+      "loss": 0.6434,
+      "step": 2935
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010608194622279131,
+      "loss": 0.6814,
+      "step": 2936
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010604993597951345,
+      "loss": 0.7242,
+      "step": 2937
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010601792573623559,
+      "loss": 0.7061,
+      "step": 2938
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010598591549295775,
+      "loss": 0.6439,
+      "step": 2939
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001059539052496799,
+      "loss": 0.7235,
+      "step": 2940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010592189500640206,
+      "loss": 0.7892,
+      "step": 2941
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001058898847631242,
+      "loss": 0.7795,
+      "step": 2942
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00010585787451984635,
+      "loss": 0.7974,
+      "step": 2943
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010582586427656851,
+      "loss": 0.7119,
+      "step": 2944
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010579385403329066,
+      "loss": 0.7623,
+      "step": 2945
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010576184379001282,
+      "loss": 0.8284,
+      "step": 2946
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010572983354673496,
+      "loss": 0.6824,
+      "step": 2947
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010569782330345712,
+      "loss": 0.8561,
+      "step": 2948
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010566581306017927,
+      "loss": 0.7821,
+      "step": 2949
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001056338028169014,
+      "loss": 0.585,
+      "step": 2950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010560179257362357,
+      "loss": 0.7845,
+      "step": 2951
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001055697823303457,
+      "loss": 0.6917,
+      "step": 2952
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010553777208706788,
+      "loss": 0.8157,
+      "step": 2953
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010550576184379001,
+      "loss": 0.8446,
+      "step": 2954
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010547375160051216,
+      "loss": 0.8268,
+      "step": 2955
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00010544174135723432,
+      "loss": 0.7445,
+      "step": 2956
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010540973111395646,
+      "loss": 0.7484,
+      "step": 2957
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010537772087067862,
+      "loss": 0.9661,
+      "step": 2958
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010534571062740077,
+      "loss": 0.7076,
+      "step": 2959
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010531370038412292,
+      "loss": 0.7548,
+      "step": 2960
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010528169014084508,
+      "loss": 0.7675,
+      "step": 2961
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010524967989756722,
+      "loss": 0.7624,
+      "step": 2962
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010521766965428938,
+      "loss": 0.7903,
+      "step": 2963
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010518565941101153,
+      "loss": 0.8342,
+      "step": 2964
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010515364916773368,
+      "loss": 0.8473,
+      "step": 2965
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010512163892445584,
+      "loss": 0.7703,
+      "step": 2966
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010508962868117798,
+      "loss": 0.7901,
+      "step": 2967
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00010505761843790014,
+      "loss": 0.7052,
+      "step": 2968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010502560819462229,
+      "loss": 0.587,
+      "step": 2969
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010499359795134442,
+      "loss": 0.7496,
+      "step": 2970
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001049615877080666,
+      "loss": 0.7893,
+      "step": 2971
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010492957746478873,
+      "loss": 0.8012,
+      "step": 2972
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001048975672215109,
+      "loss": 0.8808,
+      "step": 2973
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010486555697823303,
+      "loss": 0.7379,
+      "step": 2974
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001048335467349552,
+      "loss": 0.7547,
+      "step": 2975
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010480153649167734,
+      "loss": 0.7066,
+      "step": 2976
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010476952624839948,
+      "loss": 0.5656,
+      "step": 2977
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010473751600512164,
+      "loss": 0.7861,
+      "step": 2978
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010470550576184379,
+      "loss": 0.804,
+      "step": 2979
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00010467349551856595,
+      "loss": 0.805,
+      "step": 2980
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001046414852752881,
+      "loss": 0.7605,
+      "step": 2981
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010460947503201024,
+      "loss": 0.6171,
+      "step": 2982
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001045774647887324,
+      "loss": 0.8036,
+      "step": 2983
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010454545454545455,
+      "loss": 0.7473,
+      "step": 2984
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010451344430217671,
+      "loss": 0.7969,
+      "step": 2985
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010448143405889885,
+      "loss": 0.6046,
+      "step": 2986
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.000104449423815621,
+      "loss": 0.7979,
+      "step": 2987
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010441741357234316,
+      "loss": 0.7891,
+      "step": 2988
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010438540332906531,
+      "loss": 0.7196,
+      "step": 2989
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010435339308578747,
+      "loss": 0.7584,
+      "step": 2990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010432138284250961,
+      "loss": 0.7585,
+      "step": 2991
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010428937259923175,
+      "loss": 0.8006,
+      "step": 2992
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00010425736235595392,
+      "loss": 0.7747,
+      "step": 2993
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010422535211267605,
+      "loss": 0.7553,
+      "step": 2994
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010419334186939822,
+      "loss": 0.6685,
+      "step": 2995
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010416133162612036,
+      "loss": 0.7339,
+      "step": 2996
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001041293213828425,
+      "loss": 0.624,
+      "step": 2997
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010409731113956466,
+      "loss": 0.838,
+      "step": 2998
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010406530089628681,
+      "loss": 0.7045,
+      "step": 2999
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010403329065300897,
+      "loss": 0.6755,
+      "step": 3000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010400128040973112,
+      "loss": 0.7125,
+      "step": 3001
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010396927016645328,
+      "loss": 0.8702,
+      "step": 3002
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010393725992317542,
+      "loss": 0.7519,
+      "step": 3003
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010390524967989757,
+      "loss": 0.7437,
+      "step": 3004
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010387323943661973,
+      "loss": 0.8127,
+      "step": 3005
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00010384122919334187,
+      "loss": 0.8399,
+      "step": 3006
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010380921895006403,
+      "loss": 0.6314,
+      "step": 3007
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010377720870678618,
+      "loss": 0.6732,
+      "step": 3008
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010374519846350831,
+      "loss": 0.7058,
+      "step": 3009
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010371318822023049,
+      "loss": 0.6973,
+      "step": 3010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010368117797695263,
+      "loss": 0.8507,
+      "step": 3011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010364916773367479,
+      "loss": 0.9196,
+      "step": 3012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010361715749039694,
+      "loss": 0.7347,
+      "step": 3013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010358514724711907,
+      "loss": 0.8507,
+      "step": 3014
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010355313700384124,
+      "loss": 0.6584,
+      "step": 3015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010352112676056338,
+      "loss": 0.7154,
+      "step": 3016
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010348911651728555,
+      "loss": 0.7608,
+      "step": 3017
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00010345710627400768,
+      "loss": 0.7605,
+      "step": 3018
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010342509603072983,
+      "loss": 0.7796,
+      "step": 3019
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010339308578745199,
+      "loss": 0.8512,
+      "step": 3020
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010336107554417414,
+      "loss": 0.7987,
+      "step": 3021
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001033290653008963,
+      "loss": 0.722,
+      "step": 3022
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010329705505761844,
+      "loss": 0.8755,
+      "step": 3023
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010326504481434059,
+      "loss": 0.6803,
+      "step": 3024
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010323303457106275,
+      "loss": 0.7591,
+      "step": 3025
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001032010243277849,
+      "loss": 0.6921,
+      "step": 3026
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010316901408450705,
+      "loss": 0.7855,
+      "step": 3027
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001031370038412292,
+      "loss": 0.8757,
+      "step": 3028
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010310499359795136,
+      "loss": 0.72,
+      "step": 3029
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0001030729833546735,
+      "loss": 0.6634,
+      "step": 3030
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00010304097311139564,
+      "loss": 0.744,
+      "step": 3031
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010300896286811781,
+      "loss": 0.7928,
+      "step": 3032
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010297695262483994,
+      "loss": 0.7854,
+      "step": 3033
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010294494238156212,
+      "loss": 0.8709,
+      "step": 3034
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010291293213828425,
+      "loss": 0.8246,
+      "step": 3035
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001028809218950064,
+      "loss": 0.8054,
+      "step": 3036
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010284891165172857,
+      "loss": 0.7955,
+      "step": 3037
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0001028169014084507,
+      "loss": 0.6639,
+      "step": 3038
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010278489116517288,
+      "loss": 0.6431,
+      "step": 3039
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010275288092189501,
+      "loss": 0.7188,
+      "step": 3040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010272087067861715,
+      "loss": 0.7602,
+      "step": 3041
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010268886043533931,
+      "loss": 0.6552,
+      "step": 3042
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00010265685019206146,
+      "loss": 0.76,
+      "step": 3043
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010262483994878362,
+      "loss": 0.7439,
+      "step": 3044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010259282970550577,
+      "loss": 0.7147,
+      "step": 3045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010256081946222791,
+      "loss": 0.743,
+      "step": 3046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010252880921895007,
+      "loss": 0.6316,
+      "step": 3047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010249679897567222,
+      "loss": 0.8192,
+      "step": 3048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010246478873239438,
+      "loss": 0.6575,
+      "step": 3049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010243277848911653,
+      "loss": 0.7623,
+      "step": 3050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010240076824583866,
+      "loss": 0.8161,
+      "step": 3051
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010236875800256083,
+      "loss": 0.7611,
+      "step": 3052
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010233674775928296,
+      "loss": 0.7585,
+      "step": 3053
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010230473751600514,
+      "loss": 0.6443,
+      "step": 3054
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010227272727272727,
+      "loss": 0.6584,
+      "step": 3055
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00010224071702944942,
+      "loss": 0.6959,
+      "step": 3056
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010220870678617158,
+      "loss": 0.7398,
+      "step": 3057
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010217669654289372,
+      "loss": 0.766,
+      "step": 3058
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010214468629961588,
+      "loss": 0.7159,
+      "step": 3059
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010211267605633803,
+      "loss": 0.737,
+      "step": 3060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010208066581306019,
+      "loss": 0.7171,
+      "step": 3061
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010204865556978233,
+      "loss": 0.7649,
+      "step": 3062
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010201664532650448,
+      "loss": 0.799,
+      "step": 3063
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010198463508322664,
+      "loss": 0.8145,
+      "step": 3064
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010195262483994879,
+      "loss": 0.8177,
+      "step": 3065
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010192061459667095,
+      "loss": 0.7402,
+      "step": 3066
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010188860435339309,
+      "loss": 0.6594,
+      "step": 3067
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00010185659411011524,
+      "loss": 0.764,
+      "step": 3068
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001018245838668374,
+      "loss": 0.7704,
+      "step": 3069
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010179257362355954,
+      "loss": 0.8791,
+      "step": 3070
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001017605633802817,
+      "loss": 0.7022,
+      "step": 3071
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010172855313700385,
+      "loss": 0.6411,
+      "step": 3072
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010169654289372598,
+      "loss": 0.8373,
+      "step": 3073
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010166453265044816,
+      "loss": 0.8367,
+      "step": 3074
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010163252240717029,
+      "loss": 0.7559,
+      "step": 3075
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010160051216389246,
+      "loss": 0.6553,
+      "step": 3076
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001015685019206146,
+      "loss": 0.8219,
+      "step": 3077
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010153649167733674,
+      "loss": 0.7503,
+      "step": 3078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001015044814340589,
+      "loss": 0.797,
+      "step": 3079
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010147247119078105,
+      "loss": 0.6492,
+      "step": 3080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00010144046094750321,
+      "loss": 0.9552,
+      "step": 3081
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010140845070422535,
+      "loss": 0.6169,
+      "step": 3082
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001013764404609475,
+      "loss": 0.7344,
+      "step": 3083
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010134443021766966,
+      "loss": 0.6645,
+      "step": 3084
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001013124199743918,
+      "loss": 0.9632,
+      "step": 3085
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010128040973111397,
+      "loss": 0.7757,
+      "step": 3086
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010124839948783611,
+      "loss": 0.77,
+      "step": 3087
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010121638924455827,
+      "loss": 0.6727,
+      "step": 3088
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010118437900128042,
+      "loss": 0.5957,
+      "step": 3089
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010115236875800256,
+      "loss": 0.683,
+      "step": 3090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010112035851472472,
+      "loss": 0.7904,
+      "step": 3091
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010108834827144687,
+      "loss": 0.6627,
+      "step": 3092
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00010105633802816903,
+      "loss": 0.7328,
+      "step": 3093
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010102432778489118,
+      "loss": 0.8373,
+      "step": 3094
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010099231754161331,
+      "loss": 0.7284,
+      "step": 3095
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010096030729833548,
+      "loss": 0.7506,
+      "step": 3096
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010092829705505761,
+      "loss": 0.8394,
+      "step": 3097
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010089628681177979,
+      "loss": 0.7253,
+      "step": 3098
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010086427656850192,
+      "loss": 0.8019,
+      "step": 3099
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010083226632522407,
+      "loss": 0.864,
+      "step": 3100
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010080025608194623,
+      "loss": 0.827,
+      "step": 3101
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010076824583866837,
+      "loss": 0.6554,
+      "step": 3102
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010073623559539053,
+      "loss": 0.7624,
+      "step": 3103
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010070422535211268,
+      "loss": 0.7467,
+      "step": 3104
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010067221510883483,
+      "loss": 0.8381,
+      "step": 3105
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00010064020486555699,
+      "loss": 0.6798,
+      "step": 3106
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010060819462227913,
+      "loss": 0.7463,
+      "step": 3107
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010057618437900129,
+      "loss": 0.7776,
+      "step": 3108
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010054417413572344,
+      "loss": 0.7464,
+      "step": 3109
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010051216389244557,
+      "loss": 0.8203,
+      "step": 3110
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010048015364916774,
+      "loss": 0.7048,
+      "step": 3111
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010044814340588988,
+      "loss": 0.8408,
+      "step": 3112
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010041613316261205,
+      "loss": 0.8052,
+      "step": 3113
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001003841229193342,
+      "loss": 0.6494,
+      "step": 3114
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010035211267605636,
+      "loss": 0.8301,
+      "step": 3115
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001003201024327785,
+      "loss": 0.7847,
+      "step": 3116
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010028809218950063,
+      "loss": 0.8566,
+      "step": 3117
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00010025608194622281,
+      "loss": 0.6464,
+      "step": 3118
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010022407170294494,
+      "loss": 0.6818,
+      "step": 3119
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010019206145966711,
+      "loss": 0.7421,
+      "step": 3120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010016005121638925,
+      "loss": 0.7746,
+      "step": 3121
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010012804097311139,
+      "loss": 0.679,
+      "step": 3122
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010009603072983355,
+      "loss": 0.8434,
+      "step": 3123
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001000640204865557,
+      "loss": 0.7942,
+      "step": 3124
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00010003201024327786,
+      "loss": 0.919,
+      "step": 3125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001,
+      "loss": 0.7921,
+      "step": 3126
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.996798975672216e-05,
+      "loss": 0.7138,
+      "step": 3127
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.993597951344431e-05,
+      "loss": 0.7953,
+      "step": 3128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.990396927016646e-05,
+      "loss": 0.7121,
+      "step": 3129
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.98719590268886e-05,
+      "loss": 0.6707,
+      "step": 3130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.983994878361076e-05,
+      "loss": 0.6766,
+      "step": 3131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.980793854033291e-05,
+      "loss": 0.7304,
+      "step": 3132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.977592829705507e-05,
+      "loss": 0.7607,
+      "step": 3133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.974391805377722e-05,
+      "loss": 0.6566,
+      "step": 3134
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.971190781049936e-05,
+      "loss": 0.7113,
+      "step": 3135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.967989756722151e-05,
+      "loss": 0.7325,
+      "step": 3136
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.964788732394367e-05,
+      "loss": 0.758,
+      "step": 3137
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.961587708066581e-05,
+      "loss": 0.7019,
+      "step": 3138
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.958386683738797e-05,
+      "loss": 0.9103,
+      "step": 3139
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.955185659411012e-05,
+      "loss": 0.9091,
+      "step": 3140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.951984635083227e-05,
+      "loss": 0.7263,
+      "step": 3141
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.948783610755443e-05,
+      "loss": 0.7479,
+      "step": 3142
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.945582586427657e-05,
+      "loss": 0.7853,
+      "step": 3143
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.942381562099873e-05,
+      "loss": 0.8859,
+      "step": 3144
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.939180537772086e-05,
+      "loss": 0.6567,
+      "step": 3145
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.935979513444302e-05,
+      "loss": 0.7457,
+      "step": 3146
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.932778489116517e-05,
+      "loss": 0.788,
+      "step": 3147
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.929577464788733e-05,
+      "loss": 0.7907,
+      "step": 3148
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.926376440460948e-05,
+      "loss": 0.8579,
+      "step": 3149
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.923175416133164e-05,
+      "loss": 0.8234,
+      "step": 3150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.919974391805378e-05,
+      "loss": 0.813,
+      "step": 3151
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.916773367477593e-05,
+      "loss": 0.8119,
+      "step": 3152
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.913572343149809e-05,
+      "loss": 0.7834,
+      "step": 3153
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.910371318822023e-05,
+      "loss": 0.7309,
+      "step": 3154
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.90717029449424e-05,
+      "loss": 0.7642,
+      "step": 3155
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.903969270166453e-05,
+      "loss": 0.752,
+      "step": 3156
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.900768245838669e-05,
+      "loss": 0.7264,
+      "step": 3157
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.897567221510883e-05,
+      "loss": 0.7314,
+      "step": 3158
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.894366197183099e-05,
+      "loss": 0.8204,
+      "step": 3159
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.891165172855314e-05,
+      "loss": 0.7574,
+      "step": 3160
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.88796414852753e-05,
+      "loss": 0.7494,
+      "step": 3161
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.884763124199745e-05,
+      "loss": 0.835,
+      "step": 3162
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.881562099871959e-05,
+      "loss": 0.7666,
+      "step": 3163
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.878361075544175e-05,
+      "loss": 0.6899,
+      "step": 3164
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.87516005121639e-05,
+      "loss": 0.8076,
+      "step": 3165
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.871959026888606e-05,
+      "loss": 0.8082,
+      "step": 3166
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.868758002560819e-05,
+      "loss": 0.7648,
+      "step": 3167
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.865556978233035e-05,
+      "loss": 0.7868,
+      "step": 3168
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.86235595390525e-05,
+      "loss": 0.7117,
+      "step": 3169
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.859154929577466e-05,
+      "loss": 0.775,
+      "step": 3170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.85595390524968e-05,
+      "loss": 0.8082,
+      "step": 3171
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.852752880921895e-05,
+      "loss": 0.6562,
+      "step": 3172
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.849551856594111e-05,
+      "loss": 0.6746,
+      "step": 3173
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.846350832266325e-05,
+      "loss": 0.7261,
+      "step": 3174
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.843149807938541e-05,
+      "loss": 0.7227,
+      "step": 3175
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.839948783610756e-05,
+      "loss": 0.7806,
+      "step": 3176
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.836747759282972e-05,
+      "loss": 0.7027,
+      "step": 3177
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.833546734955185e-05,
+      "loss": 0.6752,
+      "step": 3178
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.830345710627401e-05,
+      "loss": 0.6945,
+      "step": 3179
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.827144686299616e-05,
+      "loss": 0.7129,
+      "step": 3180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.823943661971832e-05,
+      "loss": 0.8153,
+      "step": 3181
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.820742637644046e-05,
+      "loss": 0.6587,
+      "step": 3182
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.817541613316261e-05,
+      "loss": 0.7747,
+      "step": 3183
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.814340588988477e-05,
+      "loss": 0.6913,
+      "step": 3184
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.811139564660692e-05,
+      "loss": 0.7571,
+      "step": 3185
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.807938540332908e-05,
+      "loss": 0.8504,
+      "step": 3186
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.804737516005122e-05,
+      "loss": 0.9203,
+      "step": 3187
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.801536491677337e-05,
+      "loss": 0.6594,
+      "step": 3188
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.798335467349552e-05,
+      "loss": 0.8461,
+      "step": 3189
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.795134443021768e-05,
+      "loss": 0.7134,
+      "step": 3190
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.791933418693982e-05,
+      "loss": 0.7093,
+      "step": 3191
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.788732394366198e-05,
+      "loss": 0.7466,
+      "step": 3192
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.785531370038413e-05,
+      "loss": 0.7291,
+      "step": 3193
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.782330345710627e-05,
+      "loss": 0.8144,
+      "step": 3194
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.779129321382843e-05,
+      "loss": 0.759,
+      "step": 3195
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.775928297055058e-05,
+      "loss": 0.7432,
+      "step": 3196
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.772727272727274e-05,
+      "loss": 0.8008,
+      "step": 3197
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.769526248399489e-05,
+      "loss": 0.7309,
+      "step": 3198
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.766325224071703e-05,
+      "loss": 0.7451,
+      "step": 3199
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.763124199743918e-05,
+      "loss": 0.7399,
+      "step": 3200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.759923175416134e-05,
+      "loss": 0.7847,
+      "step": 3201
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.756722151088348e-05,
+      "loss": 0.7251,
+      "step": 3202
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.753521126760564e-05,
+      "loss": 0.6384,
+      "step": 3203
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.750320102432779e-05,
+      "loss": 0.7483,
+      "step": 3204
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.747119078104994e-05,
+      "loss": 0.7533,
+      "step": 3205
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.74391805377721e-05,
+      "loss": 0.811,
+      "step": 3206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.740717029449424e-05,
+      "loss": 0.7699,
+      "step": 3207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.73751600512164e-05,
+      "loss": 0.7438,
+      "step": 3208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.734314980793855e-05,
+      "loss": 0.7137,
+      "step": 3209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.73111395646607e-05,
+      "loss": 0.7488,
+      "step": 3210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.727912932138284e-05,
+      "loss": 0.7743,
+      "step": 3211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.7247119078105e-05,
+      "loss": 0.7308,
+      "step": 3212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.721510883482715e-05,
+      "loss": 0.6731,
+      "step": 3213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.718309859154931e-05,
+      "loss": 0.8083,
+      "step": 3214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.715108834827144e-05,
+      "loss": 0.6929,
+      "step": 3215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.71190781049936e-05,
+      "loss": 0.7657,
+      "step": 3216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.708706786171576e-05,
+      "loss": 0.7774,
+      "step": 3217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.70550576184379e-05,
+      "loss": 0.7326,
+      "step": 3218
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.702304737516007e-05,
+      "loss": 0.7106,
+      "step": 3219
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.699103713188221e-05,
+      "loss": 0.7056,
+      "step": 3220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.695902688860436e-05,
+      "loss": 0.6642,
+      "step": 3221
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.69270166453265e-05,
+      "loss": 0.7049,
+      "step": 3222
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.689500640204866e-05,
+      "loss": 0.8625,
+      "step": 3223
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.686299615877081e-05,
+      "loss": 0.6616,
+      "step": 3224
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.683098591549297e-05,
+      "loss": 0.73,
+      "step": 3225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.67989756722151e-05,
+      "loss": 0.6945,
+      "step": 3226
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.676696542893726e-05,
+      "loss": 0.816,
+      "step": 3227
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.673495518565941e-05,
+      "loss": 0.766,
+      "step": 3228
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.670294494238157e-05,
+      "loss": 0.7024,
+      "step": 3229
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.667093469910373e-05,
+      "loss": 0.7118,
+      "step": 3230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.663892445582587e-05,
+      "loss": 0.9246,
+      "step": 3231
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.660691421254802e-05,
+      "loss": 0.7255,
+      "step": 3232
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.657490396927017e-05,
+      "loss": 0.7413,
+      "step": 3233
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.654289372599233e-05,
+      "loss": 0.8422,
+      "step": 3234
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.651088348271447e-05,
+      "loss": 0.7779,
+      "step": 3235
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.647887323943663e-05,
+      "loss": 0.7067,
+      "step": 3236
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.644686299615877e-05,
+      "loss": 0.7671,
+      "step": 3237
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.641485275288092e-05,
+      "loss": 0.6652,
+      "step": 3238
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.638284250960307e-05,
+      "loss": 0.7769,
+      "step": 3239
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.635083226632523e-05,
+      "loss": 0.7696,
+      "step": 3240
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.631882202304738e-05,
+      "loss": 0.8159,
+      "step": 3241
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.628681177976952e-05,
+      "loss": 0.7393,
+      "step": 3242
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.625480153649168e-05,
+      "loss": 0.6135,
+      "step": 3243
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.622279129321383e-05,
+      "loss": 0.6894,
+      "step": 3244
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.619078104993599e-05,
+      "loss": 0.7813,
+      "step": 3245
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.615877080665814e-05,
+      "loss": 0.8276,
+      "step": 3246
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.61267605633803e-05,
+      "loss": 0.7217,
+      "step": 3247
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.609475032010243e-05,
+      "loss": 0.7306,
+      "step": 3248
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.606274007682459e-05,
+      "loss": 0.622,
+      "step": 3249
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.603072983354673e-05,
+      "loss": 0.6793,
+      "step": 3250
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.59987195902689e-05,
+      "loss": 0.752,
+      "step": 3251
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.596670934699104e-05,
+      "loss": 0.6745,
+      "step": 3252
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.593469910371319e-05,
+      "loss": 0.92,
+      "step": 3253
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.590268886043535e-05,
+      "loss": 0.6253,
+      "step": 3254
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.587067861715749e-05,
+      "loss": 0.7764,
+      "step": 3255
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.583866837387965e-05,
+      "loss": 0.7902,
+      "step": 3256
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.58066581306018e-05,
+      "loss": 0.7994,
+      "step": 3257
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.577464788732394e-05,
+      "loss": 0.6058,
+      "step": 3258
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.574263764404609e-05,
+      "loss": 0.7169,
+      "step": 3259
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.571062740076825e-05,
+      "loss": 0.7929,
+      "step": 3260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.56786171574904e-05,
+      "loss": 0.8975,
+      "step": 3261
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.564660691421256e-05,
+      "loss": 0.9405,
+      "step": 3262
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.56145966709347e-05,
+      "loss": 0.6401,
+      "step": 3263
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.558258642765685e-05,
+      "loss": 0.7699,
+      "step": 3264
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.555057618437901e-05,
+      "loss": 0.6407,
+      "step": 3265
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.551856594110115e-05,
+      "loss": 0.8884,
+      "step": 3266
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.548655569782331e-05,
+      "loss": 0.773,
+      "step": 3267
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.545454545454546e-05,
+      "loss": 0.9361,
+      "step": 3268
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.542253521126761e-05,
+      "loss": 0.7068,
+      "step": 3269
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.539052496798975e-05,
+      "loss": 0.888,
+      "step": 3270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.535851472471191e-05,
+      "loss": 0.7185,
+      "step": 3271
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.532650448143406e-05,
+      "loss": 0.7151,
+      "step": 3272
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.529449423815622e-05,
+      "loss": 0.7069,
+      "step": 3273
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.526248399487837e-05,
+      "loss": 0.7717,
+      "step": 3274
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.523047375160051e-05,
+      "loss": 0.7086,
+      "step": 3275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.519846350832267e-05,
+      "loss": 0.7758,
+      "step": 3276
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.516645326504482e-05,
+      "loss": 0.7608,
+      "step": 3277
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.513444302176698e-05,
+      "loss": 0.6706,
+      "step": 3278
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.510243277848912e-05,
+      "loss": 0.6834,
+      "step": 3279
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.507042253521127e-05,
+      "loss": 0.6608,
+      "step": 3280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.503841229193342e-05,
+      "loss": 0.6952,
+      "step": 3281
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.500640204865558e-05,
+      "loss": 0.6871,
+      "step": 3282
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.497439180537772e-05,
+      "loss": 0.7976,
+      "step": 3283
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.494238156209988e-05,
+      "loss": 0.7723,
+      "step": 3284
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.491037131882203e-05,
+      "loss": 0.6766,
+      "step": 3285
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.487836107554417e-05,
+      "loss": 0.7226,
+      "step": 3286
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.484635083226633e-05,
+      "loss": 0.658,
+      "step": 3287
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.481434058898848e-05,
+      "loss": 0.734,
+      "step": 3288
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.478233034571064e-05,
+      "loss": 0.7734,
+      "step": 3289
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.475032010243279e-05,
+      "loss": 0.7657,
+      "step": 3290
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.471830985915493e-05,
+      "loss": 0.7224,
+      "step": 3291
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.468629961587708e-05,
+      "loss": 0.8735,
+      "step": 3292
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.465428937259924e-05,
+      "loss": 0.7189,
+      "step": 3293
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.462227912932138e-05,
+      "loss": 0.7735,
+      "step": 3294
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.459026888604354e-05,
+      "loss": 0.7231,
+      "step": 3295
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.455825864276569e-05,
+      "loss": 0.7243,
+      "step": 3296
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.452624839948784e-05,
+      "loss": 0.8121,
+      "step": 3297
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.449423815621e-05,
+      "loss": 0.8405,
+      "step": 3298
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.446222791293214e-05,
+      "loss": 0.701,
+      "step": 3299
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.44302176696543e-05,
+      "loss": 0.7691,
+      "step": 3300
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.439820742637645e-05,
+      "loss": 0.7158,
+      "step": 3301
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.43661971830986e-05,
+      "loss": 0.8439,
+      "step": 3302
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.433418693982074e-05,
+      "loss": 0.6681,
+      "step": 3303
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.43021766965429e-05,
+      "loss": 0.6909,
+      "step": 3304
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.427016645326505e-05,
+      "loss": 0.7372,
+      "step": 3305
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.423815620998721e-05,
+      "loss": 0.7026,
+      "step": 3306
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.420614596670935e-05,
+      "loss": 0.7896,
+      "step": 3307
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.41741357234315e-05,
+      "loss": 0.7206,
+      "step": 3308
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.414212548015366e-05,
+      "loss": 0.7918,
+      "step": 3309
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.41101152368758e-05,
+      "loss": 0.8121,
+      "step": 3310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.407810499359797e-05,
+      "loss": 0.7619,
+      "step": 3311
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.40460947503201e-05,
+      "loss": 0.6473,
+      "step": 3312
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.401408450704226e-05,
+      "loss": 0.6538,
+      "step": 3313
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.39820742637644e-05,
+      "loss": 0.6144,
+      "step": 3314
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.395006402048656e-05,
+      "loss": 0.6979,
+      "step": 3315
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.391805377720871e-05,
+      "loss": 0.8324,
+      "step": 3316
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.388604353393087e-05,
+      "loss": 0.7983,
+      "step": 3317
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.3854033290653e-05,
+      "loss": 0.7099,
+      "step": 3318
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.382202304737516e-05,
+      "loss": 0.8352,
+      "step": 3319
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.379001280409732e-05,
+      "loss": 0.678,
+      "step": 3320
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.375800256081947e-05,
+      "loss": 0.6794,
+      "step": 3321
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.372599231754163e-05,
+      "loss": 0.7741,
+      "step": 3322
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.369398207426376e-05,
+      "loss": 0.744,
+      "step": 3323
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.366197183098592e-05,
+      "loss": 0.635,
+      "step": 3324
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.362996158770807e-05,
+      "loss": 0.8191,
+      "step": 3325
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.359795134443023e-05,
+      "loss": 0.7491,
+      "step": 3326
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.356594110115237e-05,
+      "loss": 0.7097,
+      "step": 3327
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.353393085787452e-05,
+      "loss": 0.8563,
+      "step": 3328
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.350192061459667e-05,
+      "loss": 0.7315,
+      "step": 3329
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.346991037131883e-05,
+      "loss": 0.8195,
+      "step": 3330
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.343790012804097e-05,
+      "loss": 0.7377,
+      "step": 3331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.340588988476313e-05,
+      "loss": 0.7839,
+      "step": 3332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.337387964148528e-05,
+      "loss": 0.7064,
+      "step": 3333
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.334186939820742e-05,
+      "loss": 0.787,
+      "step": 3334
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.330985915492958e-05,
+      "loss": 0.9049,
+      "step": 3335
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.327784891165173e-05,
+      "loss": 0.7069,
+      "step": 3336
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.324583866837389e-05,
+      "loss": 0.7193,
+      "step": 3337
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.321382842509604e-05,
+      "loss": 0.6776,
+      "step": 3338
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.318181818181818e-05,
+      "loss": 0.7164,
+      "step": 3339
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.314980793854033e-05,
+      "loss": 0.7305,
+      "step": 3340
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.311779769526249e-05,
+      "loss": 0.6901,
+      "step": 3341
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.308578745198463e-05,
+      "loss": 0.709,
+      "step": 3342
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.30537772087068e-05,
+      "loss": 0.7376,
+      "step": 3343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.302176696542894e-05,
+      "loss": 0.7679,
+      "step": 3344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298975672215109e-05,
+      "loss": 0.8033,
+      "step": 3345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.295774647887325e-05,
+      "loss": 0.5762,
+      "step": 3346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.292573623559539e-05,
+      "loss": 0.569,
+      "step": 3347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.289372599231755e-05,
+      "loss": 0.7459,
+      "step": 3348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.28617157490397e-05,
+      "loss": 0.703,
+      "step": 3349
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.282970550576185e-05,
+      "loss": 0.8826,
+      "step": 3350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.279769526248399e-05,
+      "loss": 0.7886,
+      "step": 3351
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.276568501920615e-05,
+      "loss": 0.8043,
+      "step": 3352
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.27336747759283e-05,
+      "loss": 0.751,
+      "step": 3353
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.270166453265046e-05,
+      "loss": 0.7804,
+      "step": 3354
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.26696542893726e-05,
+      "loss": 0.8256,
+      "step": 3355
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.263764404609475e-05,
+      "loss": 0.7921,
+      "step": 3356
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.260563380281691e-05,
+      "loss": 0.6966,
+      "step": 3357
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.257362355953906e-05,
+      "loss": 0.7518,
+      "step": 3358
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.254161331626122e-05,
+      "loss": 0.8085,
+      "step": 3359
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.250960307298336e-05,
+      "loss": 0.8015,
+      "step": 3360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.247759282970551e-05,
+      "loss": 0.7543,
+      "step": 3361
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244558258642765e-05,
+      "loss": 0.6892,
+      "step": 3362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.241357234314981e-05,
+      "loss": 0.7246,
+      "step": 3363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.238156209987196e-05,
+      "loss": 0.8146,
+      "step": 3364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234955185659412e-05,
+      "loss": 0.6579,
+      "step": 3365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231754161331627e-05,
+      "loss": 0.7171,
+      "step": 3366
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228553137003841e-05,
+      "loss": 0.7129,
+      "step": 3367
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.225352112676057e-05,
+      "loss": 0.8106,
+      "step": 3368
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.222151088348272e-05,
+      "loss": 0.7325,
+      "step": 3369
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.218950064020488e-05,
+      "loss": 0.7534,
+      "step": 3370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.215749039692702e-05,
+      "loss": 0.7747,
+      "step": 3371
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.212548015364917e-05,
+      "loss": 0.7761,
+      "step": 3372
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.209346991037132e-05,
+      "loss": 0.6322,
+      "step": 3373
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.206145966709348e-05,
+      "loss": 0.7122,
+      "step": 3374
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.202944942381562e-05,
+      "loss": 0.8123,
+      "step": 3375
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.199743918053778e-05,
+      "loss": 0.6652,
+      "step": 3376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.196542893725993e-05,
+      "loss": 0.8518,
+      "step": 3377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.193341869398208e-05,
+      "loss": 0.7832,
+      "step": 3378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.190140845070423e-05,
+      "loss": 0.649,
+      "step": 3379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.186939820742638e-05,
+      "loss": 0.7824,
+      "step": 3380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.183738796414854e-05,
+      "loss": 0.7179,
+      "step": 3381
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.180537772087067e-05,
+      "loss": 0.7984,
+      "step": 3382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.177336747759283e-05,
+      "loss": 0.8131,
+      "step": 3383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.174135723431498e-05,
+      "loss": 0.7362,
+      "step": 3384
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.170934699103714e-05,
+      "loss": 0.7143,
+      "step": 3385
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.167733674775929e-05,
+      "loss": 0.6642,
+      "step": 3386
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.164532650448145e-05,
+      "loss": 0.7525,
+      "step": 3387
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.161331626120359e-05,
+      "loss": 0.7647,
+      "step": 3388
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.158130601792574e-05,
+      "loss": 0.7053,
+      "step": 3389
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15492957746479e-05,
+      "loss": 0.6969,
+      "step": 3390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151728553137004e-05,
+      "loss": 0.6927,
+      "step": 3391
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14852752880922e-05,
+      "loss": 0.5547,
+      "step": 3392
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.145326504481434e-05,
+      "loss": 0.7429,
+      "step": 3393
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.14212548015365e-05,
+      "loss": 0.7565,
+      "step": 3394
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.138924455825864e-05,
+      "loss": 0.7839,
+      "step": 3395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.13572343149808e-05,
+      "loss": 0.7561,
+      "step": 3396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.132522407170295e-05,
+      "loss": 0.7354,
+      "step": 3397
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.129321382842511e-05,
+      "loss": 0.8195,
+      "step": 3398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.126120358514725e-05,
+      "loss": 0.7426,
+      "step": 3399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.12291933418694e-05,
+      "loss": 0.789,
+      "step": 3400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.119718309859156e-05,
+      "loss": 0.7711,
+      "step": 3401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.11651728553137e-05,
+      "loss": 0.7351,
+      "step": 3402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.113316261203587e-05,
+      "loss": 0.788,
+      "step": 3403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.1101152368758e-05,
+      "loss": 0.6213,
+      "step": 3404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.106914212548016e-05,
+      "loss": 0.8,
+      "step": 3405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.10371318822023e-05,
+      "loss": 0.8201,
+      "step": 3406
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.100512163892446e-05,
+      "loss": 0.7709,
+      "step": 3407
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.097311139564661e-05,
+      "loss": 0.6818,
+      "step": 3408
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.094110115236876e-05,
+      "loss": 0.7694,
+      "step": 3409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 0.7752,
+      "step": 3410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.087708066581306e-05,
+      "loss": 0.8655,
+      "step": 3411
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.084507042253522e-05,
+      "loss": 0.7461,
+      "step": 3412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.081306017925737e-05,
+      "loss": 0.7515,
+      "step": 3413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.078104993597953e-05,
+      "loss": 0.8181,
+      "step": 3414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.074903969270166e-05,
+      "loss": 0.6523,
+      "step": 3415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.071702944942382e-05,
+      "loss": 0.6652,
+      "step": 3416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.068501920614597e-05,
+      "loss": 0.643,
+      "step": 3417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.065300896286813e-05,
+      "loss": 0.7795,
+      "step": 3418
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.062099871959027e-05,
+      "loss": 0.8565,
+      "step": 3419
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.058898847631242e-05,
+      "loss": 0.8027,
+      "step": 3420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.055697823303457e-05,
+      "loss": 0.7643,
+      "step": 3421
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.052496798975673e-05,
+      "loss": 0.678,
+      "step": 3422
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.049295774647887e-05,
+      "loss": 0.6925,
+      "step": 3423
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.046094750320103e-05,
+      "loss": 0.8498,
+      "step": 3424
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.042893725992318e-05,
+      "loss": 0.7826,
+      "step": 3425
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.039692701664532e-05,
+      "loss": 0.7964,
+      "step": 3426
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.036491677336748e-05,
+      "loss": 0.7492,
+      "step": 3427
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.033290653008963e-05,
+      "loss": 0.6843,
+      "step": 3428
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.030089628681179e-05,
+      "loss": 0.8092,
+      "step": 3429
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.026888604353394e-05,
+      "loss": 0.7252,
+      "step": 3430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.023687580025608e-05,
+      "loss": 0.8138,
+      "step": 3431
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.020486555697823e-05,
+      "loss": 0.8043,
+      "step": 3432
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.017285531370039e-05,
+      "loss": 0.8653,
+      "step": 3433
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.014084507042254e-05,
+      "loss": 0.6592,
+      "step": 3434
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.01088348271447e-05,
+      "loss": 0.7373,
+      "step": 3435
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.007682458386684e-05,
+      "loss": 0.7495,
+      "step": 3436
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.004481434058899e-05,
+      "loss": 0.7694,
+      "step": 3437
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.001280409731115e-05,
+      "loss": 0.627,
+      "step": 3438
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.99807938540333e-05,
+      "loss": 0.8217,
+      "step": 3439
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.994878361075545e-05,
+      "loss": 0.849,
+      "step": 3440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.99167733674776e-05,
+      "loss": 0.7179,
+      "step": 3441
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.988476312419975e-05,
+      "loss": 0.8216,
+      "step": 3442
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.985275288092189e-05,
+      "loss": 0.6905,
+      "step": 3443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.982074263764405e-05,
+      "loss": 0.7441,
+      "step": 3444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.97887323943662e-05,
+      "loss": 0.7386,
+      "step": 3445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.975672215108836e-05,
+      "loss": 0.7467,
+      "step": 3446
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.97247119078105e-05,
+      "loss": 0.8003,
+      "step": 3447
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.969270166453265e-05,
+      "loss": 0.6559,
+      "step": 3448
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.966069142125481e-05,
+      "loss": 0.7651,
+      "step": 3449
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.962868117797696e-05,
+      "loss": 0.7277,
+      "step": 3450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.959667093469912e-05,
+      "loss": 0.8058,
+      "step": 3451
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.956466069142125e-05,
+      "loss": 0.7561,
+      "step": 3452
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.953265044814341e-05,
+      "loss": 0.8854,
+      "step": 3453
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.950064020486555e-05,
+      "loss": 0.6462,
+      "step": 3454
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.946862996158771e-05,
+      "loss": 0.7167,
+      "step": 3455
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.943661971830986e-05,
+      "loss": 0.8101,
+      "step": 3456
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.940460947503202e-05,
+      "loss": 0.7403,
+      "step": 3457
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.937259923175417e-05,
+      "loss": 0.7644,
+      "step": 3458
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.934058898847631e-05,
+      "loss": 0.8233,
+      "step": 3459
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.930857874519847e-05,
+      "loss": 0.826,
+      "step": 3460
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.927656850192062e-05,
+      "loss": 0.7811,
+      "step": 3461
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.924455825864278e-05,
+      "loss": 0.8728,
+      "step": 3462
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.921254801536491e-05,
+      "loss": 0.794,
+      "step": 3463
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.918053777208707e-05,
+      "loss": 0.753,
+      "step": 3464
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.914852752880922e-05,
+      "loss": 0.9353,
+      "step": 3465
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.911651728553138e-05,
+      "loss": 0.7191,
+      "step": 3466
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.908450704225352e-05,
+      "loss": 0.8647,
+      "step": 3467
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.905249679897568e-05,
+      "loss": 0.7085,
+      "step": 3468
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.902048655569783e-05,
+      "loss": 0.7354,
+      "step": 3469
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.898847631241998e-05,
+      "loss": 0.7524,
+      "step": 3470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.895646606914214e-05,
+      "loss": 0.7218,
+      "step": 3471
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.892445582586428e-05,
+      "loss": 0.7224,
+      "step": 3472
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.889244558258644e-05,
+      "loss": 0.7233,
+      "step": 3473
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.886043533930857e-05,
+      "loss": 0.6786,
+      "step": 3474
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.882842509603073e-05,
+      "loss": 0.7517,
+      "step": 3475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.879641485275288e-05,
+      "loss": 0.6633,
+      "step": 3476
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.876440460947504e-05,
+      "loss": 0.7414,
+      "step": 3477
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.873239436619719e-05,
+      "loss": 0.899,
+      "step": 3478
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.870038412291933e-05,
+      "loss": 0.62,
+      "step": 3479
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.866837387964149e-05,
+      "loss": 0.6395,
+      "step": 3480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.863636363636364e-05,
+      "loss": 0.7531,
+      "step": 3481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.86043533930858e-05,
+      "loss": 0.7821,
+      "step": 3482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.857234314980794e-05,
+      "loss": 0.698,
+      "step": 3483
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.85403329065301e-05,
+      "loss": 0.7784,
+      "step": 3484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.850832266325224e-05,
+      "loss": 0.6644,
+      "step": 3485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.84763124199744e-05,
+      "loss": 0.7533,
+      "step": 3486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.844430217669654e-05,
+      "loss": 0.6341,
+      "step": 3487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.84122919334187e-05,
+      "loss": 0.7153,
+      "step": 3488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.838028169014085e-05,
+      "loss": 0.7649,
+      "step": 3489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.8348271446863e-05,
+      "loss": 0.7077,
+      "step": 3490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.831626120358515e-05,
+      "loss": 0.765,
+      "step": 3491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.82842509603073e-05,
+      "loss": 0.7951,
+      "step": 3492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.825224071702946e-05,
+      "loss": 0.6694,
+      "step": 3493
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.822023047375161e-05,
+      "loss": 0.7926,
+      "step": 3494
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.818822023047375e-05,
+      "loss": 0.7126,
+      "step": 3495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.81562099871959e-05,
+      "loss": 0.8151,
+      "step": 3496
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.812419974391806e-05,
+      "loss": 0.7275,
+      "step": 3497
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.80921895006402e-05,
+      "loss": 0.7165,
+      "step": 3498
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.806017925736237e-05,
+      "loss": 0.7686,
+      "step": 3499
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.802816901408451e-05,
+      "loss": 0.6949,
+      "step": 3500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.799615877080666e-05,
+      "loss": 0.765,
+      "step": 3501
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.796414852752882e-05,
+      "loss": 0.6952,
+      "step": 3502
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.793213828425096e-05,
+      "loss": 0.8083,
+      "step": 3503
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.790012804097312e-05,
+      "loss": 0.8343,
+      "step": 3504
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.786811779769527e-05,
+      "loss": 0.7339,
+      "step": 3505
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.783610755441742e-05,
+      "loss": 0.6442,
+      "step": 3506
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.780409731113956e-05,
+      "loss": 0.8082,
+      "step": 3507
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.777208706786172e-05,
+      "loss": 0.7931,
+      "step": 3508
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.774007682458387e-05,
+      "loss": 0.7291,
+      "step": 3509
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.770806658130603e-05,
+      "loss": 0.6949,
+      "step": 3510
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.767605633802817e-05,
+      "loss": 0.8709,
+      "step": 3511
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.764404609475032e-05,
+      "loss": 0.739,
+      "step": 3512
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.761203585147247e-05,
+      "loss": 0.7337,
+      "step": 3513
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.758002560819463e-05,
+      "loss": 0.7553,
+      "step": 3514
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.754801536491679e-05,
+      "loss": 0.7309,
+      "step": 3515
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.751600512163893e-05,
+      "loss": 0.7127,
+      "step": 3516
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.748399487836108e-05,
+      "loss": 0.8748,
+      "step": 3517
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.745198463508323e-05,
+      "loss": 0.7254,
+      "step": 3518
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.741997439180539e-05,
+      "loss": 0.6833,
+      "step": 3519
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.738796414852753e-05,
+      "loss": 0.6106,
+      "step": 3520
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.735595390524969e-05,
+      "loss": 0.7595,
+      "step": 3521
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.732394366197182e-05,
+      "loss": 0.6917,
+      "step": 3522
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.729193341869398e-05,
+      "loss": 0.7469,
+      "step": 3523
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.725992317541613e-05,
+      "loss": 0.8689,
+      "step": 3524
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.722791293213829e-05,
+      "loss": 0.721,
+      "step": 3525
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.719590268886044e-05,
+      "loss": 0.736,
+      "step": 3526
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.71638924455826e-05,
+      "loss": 0.8983,
+      "step": 3527
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.713188220230474e-05,
+      "loss": 0.5996,
+      "step": 3528
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.709987195902689e-05,
+      "loss": 0.7452,
+      "step": 3529
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.706786171574905e-05,
+      "loss": 0.704,
+      "step": 3530
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.70358514724712e-05,
+      "loss": 0.7187,
+      "step": 3531
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.700384122919335e-05,
+      "loss": 0.7269,
+      "step": 3532
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.697183098591549e-05,
+      "loss": 0.7099,
+      "step": 3533
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.693982074263765e-05,
+      "loss": 0.736,
+      "step": 3534
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.690781049935979e-05,
+      "loss": 0.6892,
+      "step": 3535
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.687580025608195e-05,
+      "loss": 0.6823,
+      "step": 3536
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.68437900128041e-05,
+      "loss": 0.672,
+      "step": 3537
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.681177976952626e-05,
+      "loss": 0.7149,
+      "step": 3538
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.67797695262484e-05,
+      "loss": 0.7193,
+      "step": 3539
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.674775928297055e-05,
+      "loss": 0.7369,
+      "step": 3540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.671574903969271e-05,
+      "loss": 0.7006,
+      "step": 3541
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.668373879641486e-05,
+      "loss": 0.8033,
+      "step": 3542
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.665172855313702e-05,
+      "loss": 0.7879,
+      "step": 3543
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.661971830985915e-05,
+      "loss": 0.7547,
+      "step": 3544
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.658770806658131e-05,
+      "loss": 0.7087,
+      "step": 3545
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.655569782330346e-05,
+      "loss": 0.6548,
+      "step": 3546
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.652368758002562e-05,
+      "loss": 0.761,
+      "step": 3547
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.649167733674776e-05,
+      "loss": 0.7218,
+      "step": 3548
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.645966709346991e-05,
+      "loss": 0.7695,
+      "step": 3549
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.642765685019207e-05,
+      "loss": 0.7379,
+      "step": 3550
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.639564660691421e-05,
+      "loss": 0.7341,
+      "step": 3551
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.636363636363637e-05,
+      "loss": 0.7479,
+      "step": 3552
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.633162612035852e-05,
+      "loss": 0.7947,
+      "step": 3553
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.629961587708068e-05,
+      "loss": 0.7179,
+      "step": 3554
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.626760563380281e-05,
+      "loss": 0.7165,
+      "step": 3555
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.623559539052497e-05,
+      "loss": 0.5968,
+      "step": 3556
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.620358514724712e-05,
+      "loss": 0.8374,
+      "step": 3557
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.617157490396928e-05,
+      "loss": 0.724,
+      "step": 3558
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.613956466069142e-05,
+      "loss": 0.7045,
+      "step": 3559
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.610755441741357e-05,
+      "loss": 0.8342,
+      "step": 3560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.607554417413573e-05,
+      "loss": 0.796,
+      "step": 3561
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.604353393085788e-05,
+      "loss": 0.6322,
+      "step": 3562
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.601152368758004e-05,
+      "loss": 0.744,
+      "step": 3563
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.597951344430218e-05,
+      "loss": 0.7394,
+      "step": 3564
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.594750320102433e-05,
+      "loss": 0.7487,
+      "step": 3565
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.591549295774647e-05,
+      "loss": 0.8474,
+      "step": 3566
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.588348271446863e-05,
+      "loss": 0.6097,
+      "step": 3567
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.585147247119078e-05,
+      "loss": 0.7167,
+      "step": 3568
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.581946222791294e-05,
+      "loss": 0.7208,
+      "step": 3569
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.578745198463509e-05,
+      "loss": 0.7994,
+      "step": 3570
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.575544174135723e-05,
+      "loss": 0.7622,
+      "step": 3571
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.572343149807939e-05,
+      "loss": 0.8342,
+      "step": 3572
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.569142125480154e-05,
+      "loss": 0.6941,
+      "step": 3573
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.56594110115237e-05,
+      "loss": 0.6905,
+      "step": 3574
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.562740076824585e-05,
+      "loss": 0.736,
+      "step": 3575
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.559539052496799e-05,
+      "loss": 0.6689,
+      "step": 3576
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.556338028169014e-05,
+      "loss": 0.6119,
+      "step": 3577
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.55313700384123e-05,
+      "loss": 0.6254,
+      "step": 3578
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.549935979513444e-05,
+      "loss": 0.7582,
+      "step": 3579
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.54673495518566e-05,
+      "loss": 0.6999,
+      "step": 3580
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.543533930857875e-05,
+      "loss": 0.7559,
+      "step": 3581
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.54033290653009e-05,
+      "loss": 0.8938,
+      "step": 3582
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.537131882202306e-05,
+      "loss": 0.7718,
+      "step": 3583
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.53393085787452e-05,
+      "loss": 0.7892,
+      "step": 3584
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.530729833546736e-05,
+      "loss": 0.736,
+      "step": 3585
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.527528809218951e-05,
+      "loss": 0.8282,
+      "step": 3586
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.524327784891165e-05,
+      "loss": 0.8009,
+      "step": 3587
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.52112676056338e-05,
+      "loss": 0.7819,
+      "step": 3588
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.517925736235596e-05,
+      "loss": 0.6807,
+      "step": 3589
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.51472471190781e-05,
+      "loss": 0.684,
+      "step": 3590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.511523687580027e-05,
+      "loss": 0.6703,
+      "step": 3591
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.508322663252241e-05,
+      "loss": 0.7143,
+      "step": 3592
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.505121638924456e-05,
+      "loss": 0.8074,
+      "step": 3593
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.501920614596672e-05,
+      "loss": 0.8541,
+      "step": 3594
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.498719590268886e-05,
+      "loss": 0.6999,
+      "step": 3595
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.495518565941102e-05,
+      "loss": 0.8229,
+      "step": 3596
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.492317541613317e-05,
+      "loss": 0.8044,
+      "step": 3597
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.489116517285532e-05,
+      "loss": 0.8352,
+      "step": 3598
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.485915492957746e-05,
+      "loss": 0.9506,
+      "step": 3599
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.482714468629962e-05,
+      "loss": 0.7067,
+      "step": 3600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.479513444302177e-05,
+      "loss": 0.7517,
+      "step": 3601
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.476312419974393e-05,
+      "loss": 0.7876,
+      "step": 3602
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.473111395646606e-05,
+      "loss": 0.7107,
+      "step": 3603
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.469910371318822e-05,
+      "loss": 0.751,
+      "step": 3604
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466709346991038e-05,
+      "loss": 0.7623,
+      "step": 3605
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.463508322663253e-05,
+      "loss": 0.7535,
+      "step": 3606
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.460307298335469e-05,
+      "loss": 0.8407,
+      "step": 3607
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.457106274007683e-05,
+      "loss": 0.7418,
+      "step": 3608
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.453905249679898e-05,
+      "loss": 0.8314,
+      "step": 3609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.450704225352113e-05,
+      "loss": 0.7362,
+      "step": 3610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.447503201024329e-05,
+      "loss": 0.7249,
+      "step": 3611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.444302176696543e-05,
+      "loss": 0.7365,
+      "step": 3612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.441101152368759e-05,
+      "loss": 0.7132,
+      "step": 3613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.437900128040972e-05,
+      "loss": 0.7178,
+      "step": 3614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.434699103713188e-05,
+      "loss": 0.7676,
+      "step": 3615
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.431498079385403e-05,
+      "loss": 0.7015,
+      "step": 3616
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.428297055057619e-05,
+      "loss": 0.8731,
+      "step": 3617
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.425096030729835e-05,
+      "loss": 0.8121,
+      "step": 3618
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.421895006402048e-05,
+      "loss": 0.7956,
+      "step": 3619
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.418693982074264e-05,
+      "loss": 0.6545,
+      "step": 3620
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.415492957746479e-05,
+      "loss": 0.7237,
+      "step": 3621
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.412291933418695e-05,
+      "loss": 0.7644,
+      "step": 3622
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.40909090909091e-05,
+      "loss": 0.7515,
+      "step": 3623
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.405889884763125e-05,
+      "loss": 0.6998,
+      "step": 3624
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.402688860435339e-05,
+      "loss": 0.7206,
+      "step": 3625
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.399487836107555e-05,
+      "loss": 0.864,
+      "step": 3626
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.396286811779769e-05,
+      "loss": 0.6613,
+      "step": 3627
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.393085787451985e-05,
+      "loss": 0.8676,
+      "step": 3628
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.3898847631242e-05,
+      "loss": 0.7627,
+      "step": 3629
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.386683738796415e-05,
+      "loss": 0.7353,
+      "step": 3630
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.38348271446863e-05,
+      "loss": 0.6283,
+      "step": 3631
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.380281690140845e-05,
+      "loss": 0.6804,
+      "step": 3632
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.377080665813061e-05,
+      "loss": 0.7415,
+      "step": 3633
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.373879641485276e-05,
+      "loss": 0.6528,
+      "step": 3634
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.37067861715749e-05,
+      "loss": 0.8479,
+      "step": 3635
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.367477592829705e-05,
+      "loss": 0.6502,
+      "step": 3636
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.364276568501921e-05,
+      "loss": 0.8436,
+      "step": 3637
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.361075544174136e-05,
+      "loss": 0.7867,
+      "step": 3638
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.357874519846352e-05,
+      "loss": 0.6749,
+      "step": 3639
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.354673495518566e-05,
+      "loss": 0.674,
+      "step": 3640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.351472471190781e-05,
+      "loss": 0.7523,
+      "step": 3641
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.348271446862997e-05,
+      "loss": 0.774,
+      "step": 3642
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.345070422535211e-05,
+      "loss": 0.7996,
+      "step": 3643
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.341869398207427e-05,
+      "loss": 0.8567,
+      "step": 3644
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.338668373879642e-05,
+      "loss": 0.7258,
+      "step": 3645
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.335467349551857e-05,
+      "loss": 0.6709,
+      "step": 3646
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.332266325224071e-05,
+      "loss": 0.7065,
+      "step": 3647
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.329065300896287e-05,
+      "loss": 0.7192,
+      "step": 3648
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.325864276568502e-05,
+      "loss": 0.6909,
+      "step": 3649
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.322663252240718e-05,
+      "loss": 0.777,
+      "step": 3650
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.319462227912932e-05,
+      "loss": 0.7595,
+      "step": 3651
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.316261203585147e-05,
+      "loss": 0.6704,
+      "step": 3652
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.313060179257363e-05,
+      "loss": 0.8164,
+      "step": 3653
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.309859154929578e-05,
+      "loss": 0.7913,
+      "step": 3654
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.306658130601794e-05,
+      "loss": 0.7166,
+      "step": 3655
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.303457106274008e-05,
+      "loss": 0.8283,
+      "step": 3656
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.300256081946223e-05,
+      "loss": 0.8721,
+      "step": 3657
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.297055057618438e-05,
+      "loss": 0.7705,
+      "step": 3658
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.293854033290654e-05,
+      "loss": 0.6945,
+      "step": 3659
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.290653008962868e-05,
+      "loss": 0.6937,
+      "step": 3660
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.287451984635084e-05,
+      "loss": 0.706,
+      "step": 3661
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.284250960307299e-05,
+      "loss": 0.8217,
+      "step": 3662
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.281049935979513e-05,
+      "loss": 0.7291,
+      "step": 3663
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.27784891165173e-05,
+      "loss": 0.8045,
+      "step": 3664
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.274647887323944e-05,
+      "loss": 0.7404,
+      "step": 3665
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.27144686299616e-05,
+      "loss": 0.8582,
+      "step": 3666
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.268245838668375e-05,
+      "loss": 0.719,
+      "step": 3667
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.265044814340589e-05,
+      "loss": 0.8395,
+      "step": 3668
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.261843790012804e-05,
+      "loss": 0.6515,
+      "step": 3669
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.25864276568502e-05,
+      "loss": 0.7112,
+      "step": 3670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.255441741357234e-05,
+      "loss": 0.8511,
+      "step": 3671
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.25224071702945e-05,
+      "loss": 0.6876,
+      "step": 3672
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.249039692701665e-05,
+      "loss": 0.847,
+      "step": 3673
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.24583866837388e-05,
+      "loss": 0.8409,
+      "step": 3674
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.242637644046096e-05,
+      "loss": 0.7164,
+      "step": 3675
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.23943661971831e-05,
+      "loss": 0.8521,
+      "step": 3676
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.236235595390526e-05,
+      "loss": 0.7586,
+      "step": 3677
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.233034571062741e-05,
+      "loss": 0.8678,
+      "step": 3678
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.229833546734955e-05,
+      "loss": 0.704,
+      "step": 3679
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.22663252240717e-05,
+      "loss": 0.7828,
+      "step": 3680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.223431498079386e-05,
+      "loss": 0.7941,
+      "step": 3681
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.220230473751601e-05,
+      "loss": 0.5931,
+      "step": 3682
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.217029449423817e-05,
+      "loss": 0.8906,
+      "step": 3683
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.213828425096031e-05,
+      "loss": 0.7975,
+      "step": 3684
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.210627400768246e-05,
+      "loss": 0.7403,
+      "step": 3685
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.207426376440462e-05,
+      "loss": 0.7005,
+      "step": 3686
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.204225352112677e-05,
+      "loss": 0.9106,
+      "step": 3687
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.201024327784892e-05,
+      "loss": 0.8138,
+      "step": 3688
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.197823303457106e-05,
+      "loss": 0.9695,
+      "step": 3689
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.194622279129322e-05,
+      "loss": 0.8233,
+      "step": 3690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.191421254801536e-05,
+      "loss": 0.8088,
+      "step": 3691
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.188220230473752e-05,
+      "loss": 0.7851,
+      "step": 3692
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.185019206145967e-05,
+      "loss": 0.7749,
+      "step": 3693
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 0.7636,
+      "step": 3694
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.178617157490398e-05,
+      "loss": 0.7958,
+      "step": 3695
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.175416133162612e-05,
+      "loss": 0.7392,
+      "step": 3696
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.172215108834828e-05,
+      "loss": 0.7222,
+      "step": 3697
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.169014084507043e-05,
+      "loss": 0.7105,
+      "step": 3698
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.165813060179259e-05,
+      "loss": 0.7828,
+      "step": 3699
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.162612035851472e-05,
+      "loss": 0.7534,
+      "step": 3700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.159411011523688e-05,
+      "loss": 0.8409,
+      "step": 3701
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.156209987195903e-05,
+      "loss": 0.8948,
+      "step": 3702
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.153008962868119e-05,
+      "loss": 0.7348,
+      "step": 3703
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.149807938540333e-05,
+      "loss": 0.7404,
+      "step": 3704
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.146606914212549e-05,
+      "loss": 0.6841,
+      "step": 3705
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.143405889884762e-05,
+      "loss": 0.9227,
+      "step": 3706
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.140204865556978e-05,
+      "loss": 0.7035,
+      "step": 3707
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.137003841229194e-05,
+      "loss": 0.7337,
+      "step": 3708
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.133802816901409e-05,
+      "loss": 0.7892,
+      "step": 3709
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.130601792573625e-05,
+      "loss": 0.7154,
+      "step": 3710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.127400768245838e-05,
+      "loss": 0.7102,
+      "step": 3711
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.124199743918054e-05,
+      "loss": 0.7606,
+      "step": 3712
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.120998719590269e-05,
+      "loss": 0.668,
+      "step": 3713
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.117797695262485e-05,
+      "loss": 0.7594,
+      "step": 3714
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.1145966709347e-05,
+      "loss": 0.6388,
+      "step": 3715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.111395646606914e-05,
+      "loss": 0.8139,
+      "step": 3716
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.108194622279129e-05,
+      "loss": 0.7878,
+      "step": 3717
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.104993597951345e-05,
+      "loss": 0.7383,
+      "step": 3718
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.10179257362356e-05,
+      "loss": 0.7418,
+      "step": 3719
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.098591549295775e-05,
+      "loss": 0.8998,
+      "step": 3720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.095390524967991e-05,
+      "loss": 0.623,
+      "step": 3721
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.092189500640205e-05,
+      "loss": 0.7278,
+      "step": 3722
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.08898847631242e-05,
+      "loss": 0.672,
+      "step": 3723
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.085787451984635e-05,
+      "loss": 0.6928,
+      "step": 3724
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.082586427656851e-05,
+      "loss": 0.7628,
+      "step": 3725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.079385403329066e-05,
+      "loss": 0.8186,
+      "step": 3726
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.07618437900128e-05,
+      "loss": 0.6842,
+      "step": 3727
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.072983354673495e-05,
+      "loss": 0.7558,
+      "step": 3728
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.069782330345711e-05,
+      "loss": 0.7225,
+      "step": 3729
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.066581306017926e-05,
+      "loss": 0.8523,
+      "step": 3730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.063380281690142e-05,
+      "loss": 0.6979,
+      "step": 3731
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.060179257362356e-05,
+      "loss": 0.6162,
+      "step": 3732
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.056978233034571e-05,
+      "loss": 0.7674,
+      "step": 3733
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.053777208706787e-05,
+      "loss": 0.7547,
+      "step": 3734
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.050576184379001e-05,
+      "loss": 0.7031,
+      "step": 3735
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.047375160051217e-05,
+      "loss": 0.8008,
+      "step": 3736
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.044174135723432e-05,
+      "loss": 0.7118,
+      "step": 3737
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.040973111395647e-05,
+      "loss": 0.8047,
+      "step": 3738
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.037772087067861e-05,
+      "loss": 0.7383,
+      "step": 3739
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.034571062740077e-05,
+      "loss": 0.7255,
+      "step": 3740
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.031370038412292e-05,
+      "loss": 0.6616,
+      "step": 3741
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.028169014084508e-05,
+      "loss": 0.8024,
+      "step": 3742
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.024967989756723e-05,
+      "loss": 0.7322,
+      "step": 3743
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.021766965428937e-05,
+      "loss": 0.7495,
+      "step": 3744
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.018565941101153e-05,
+      "loss": 0.7516,
+      "step": 3745
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.015364916773368e-05,
+      "loss": 0.6855,
+      "step": 3746
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.012163892445584e-05,
+      "loss": 0.7152,
+      "step": 3747
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.008962868117798e-05,
+      "loss": 0.7055,
+      "step": 3748
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.005761843790013e-05,
+      "loss": 0.7009,
+      "step": 3749
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.002560819462228e-05,
+      "loss": 0.7697,
+      "step": 3750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.999359795134444e-05,
+      "loss": 0.74,
+      "step": 3751
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.996158770806658e-05,
+      "loss": 0.6802,
+      "step": 3752
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.992957746478874e-05,
+      "loss": 0.6822,
+      "step": 3753
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.989756722151089e-05,
+      "loss": 0.7562,
+      "step": 3754
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.986555697823303e-05,
+      "loss": 0.6544,
+      "step": 3755
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.98335467349552e-05,
+      "loss": 0.7546,
+      "step": 3756
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.980153649167734e-05,
+      "loss": 0.7098,
+      "step": 3757
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.97695262483995e-05,
+      "loss": 0.7313,
+      "step": 3758
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.973751600512163e-05,
+      "loss": 0.6639,
+      "step": 3759
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.970550576184379e-05,
+      "loss": 0.6337,
+      "step": 3760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.967349551856594e-05,
+      "loss": 0.7449,
+      "step": 3761
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.96414852752881e-05,
+      "loss": 0.6779,
+      "step": 3762
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.960947503201024e-05,
+      "loss": 0.637,
+      "step": 3763
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.95774647887324e-05,
+      "loss": 0.735,
+      "step": 3764
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.954545454545455e-05,
+      "loss": 0.7193,
+      "step": 3765
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.95134443021767e-05,
+      "loss": 0.7259,
+      "step": 3766
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.948143405889886e-05,
+      "loss": 0.6523,
+      "step": 3767
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.9449423815621e-05,
+      "loss": 0.6978,
+      "step": 3768
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.941741357234316e-05,
+      "loss": 0.7216,
+      "step": 3769
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.93854033290653e-05,
+      "loss": 0.599,
+      "step": 3770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.935339308578746e-05,
+      "loss": 0.7014,
+      "step": 3771
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.93213828425096e-05,
+      "loss": 0.6542,
+      "step": 3772
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.928937259923176e-05,
+      "loss": 0.802,
+      "step": 3773
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.925736235595391e-05,
+      "loss": 0.778,
+      "step": 3774
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.922535211267607e-05,
+      "loss": 0.6697,
+      "step": 3775
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.919334186939821e-05,
+      "loss": 0.7476,
+      "step": 3776
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.916133162612036e-05,
+      "loss": 0.7107,
+      "step": 3777
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.912932138284252e-05,
+      "loss": 0.7611,
+      "step": 3778
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.909731113956467e-05,
+      "loss": 0.5822,
+      "step": 3779
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.906530089628683e-05,
+      "loss": 0.6586,
+      "step": 3780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.903329065300896e-05,
+      "loss": 0.7613,
+      "step": 3781
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.900128040973112e-05,
+      "loss": 0.7804,
+      "step": 3782
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.896927016645326e-05,
+      "loss": 0.7141,
+      "step": 3783
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.893725992317542e-05,
+      "loss": 0.7197,
+      "step": 3784
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.890524967989757e-05,
+      "loss": 0.7279,
+      "step": 3785
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.887323943661972e-05,
+      "loss": 0.7078,
+      "step": 3786
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.884122919334188e-05,
+      "loss": 0.6666,
+      "step": 3787
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.880921895006402e-05,
+      "loss": 0.8436,
+      "step": 3788
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.877720870678618e-05,
+      "loss": 0.7693,
+      "step": 3789
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.874519846350833e-05,
+      "loss": 0.6345,
+      "step": 3790
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.871318822023049e-05,
+      "loss": 0.5932,
+      "step": 3791
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.868117797695262e-05,
+      "loss": 0.7152,
+      "step": 3792
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.864916773367478e-05,
+      "loss": 0.7746,
+      "step": 3793
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.861715749039693e-05,
+      "loss": 0.6455,
+      "step": 3794
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.858514724711909e-05,
+      "loss": 0.6861,
+      "step": 3795
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.855313700384123e-05,
+      "loss": 0.719,
+      "step": 3796
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.852112676056338e-05,
+      "loss": 0.7621,
+      "step": 3797
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.848911651728554e-05,
+      "loss": 0.5869,
+      "step": 3798
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.845710627400769e-05,
+      "loss": 0.7731,
+      "step": 3799
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.842509603072985e-05,
+      "loss": 0.7095,
+      "step": 3800
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.839308578745199e-05,
+      "loss": 0.8139,
+      "step": 3801
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.836107554417414e-05,
+      "loss": 0.7516,
+      "step": 3802
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.832906530089628e-05,
+      "loss": 0.6285,
+      "step": 3803
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.829705505761844e-05,
+      "loss": 0.689,
+      "step": 3804
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.826504481434059e-05,
+      "loss": 0.6985,
+      "step": 3805
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.823303457106275e-05,
+      "loss": 0.7888,
+      "step": 3806
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.82010243277849e-05,
+      "loss": 0.636,
+      "step": 3807
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.816901408450704e-05,
+      "loss": 0.7139,
+      "step": 3808
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.813700384122919e-05,
+      "loss": 0.6646,
+      "step": 3809
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.810499359795135e-05,
+      "loss": 0.6906,
+      "step": 3810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.807298335467351e-05,
+      "loss": 0.6766,
+      "step": 3811
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.804097311139565e-05,
+      "loss": 0.7609,
+      "step": 3812
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.80089628681178e-05,
+      "loss": 0.6974,
+      "step": 3813
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.797695262483995e-05,
+      "loss": 0.7329,
+      "step": 3814
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.79449423815621e-05,
+      "loss": 0.6753,
+      "step": 3815
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.791293213828425e-05,
+      "loss": 0.791,
+      "step": 3816
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.788092189500641e-05,
+      "loss": 0.5685,
+      "step": 3817
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.784891165172856e-05,
+      "loss": 0.7327,
+      "step": 3818
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.78169014084507e-05,
+      "loss": 0.5643,
+      "step": 3819
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.778489116517285e-05,
+      "loss": 0.6848,
+      "step": 3820
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.775288092189501e-05,
+      "loss": 0.682,
+      "step": 3821
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.772087067861716e-05,
+      "loss": 0.7658,
+      "step": 3822
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.768886043533932e-05,
+      "loss": 0.7359,
+      "step": 3823
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.765685019206146e-05,
+      "loss": 0.7371,
+      "step": 3824
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.762483994878361e-05,
+      "loss": 0.8346,
+      "step": 3825
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.759282970550577e-05,
+      "loss": 0.6247,
+      "step": 3826
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.756081946222792e-05,
+      "loss": 0.7248,
+      "step": 3827
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.752880921895008e-05,
+      "loss": 0.6926,
+      "step": 3828
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.749679897567221e-05,
+      "loss": 0.8102,
+      "step": 3829
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.746478873239437e-05,
+      "loss": 0.7796,
+      "step": 3830
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.743277848911651e-05,
+      "loss": 0.6873,
+      "step": 3831
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.740076824583867e-05,
+      "loss": 0.6759,
+      "step": 3832
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.736875800256082e-05,
+      "loss": 0.7456,
+      "step": 3833
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.733674775928298e-05,
+      "loss": 0.7496,
+      "step": 3834
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.730473751600513e-05,
+      "loss": 0.7198,
+      "step": 3835
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.727272727272727e-05,
+      "loss": 0.7493,
+      "step": 3836
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.724071702944943e-05,
+      "loss": 0.7094,
+      "step": 3837
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.720870678617158e-05,
+      "loss": 0.6554,
+      "step": 3838
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.717669654289374e-05,
+      "loss": 0.7692,
+      "step": 3839
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.714468629961587e-05,
+      "loss": 0.6579,
+      "step": 3840
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.711267605633803e-05,
+      "loss": 0.7497,
+      "step": 3841
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.708066581306018e-05,
+      "loss": 0.7548,
+      "step": 3842
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.704865556978234e-05,
+      "loss": 0.8296,
+      "step": 3843
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.701664532650448e-05,
+      "loss": 0.6827,
+      "step": 3844
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.698463508322664e-05,
+      "loss": 0.7529,
+      "step": 3845
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.695262483994879e-05,
+      "loss": 0.7063,
+      "step": 3846
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.692061459667093e-05,
+      "loss": 0.7692,
+      "step": 3847
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.68886043533931e-05,
+      "loss": 0.6763,
+      "step": 3848
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.685659411011524e-05,
+      "loss": 0.6263,
+      "step": 3849
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.68245838668374e-05,
+      "loss": 0.7003,
+      "step": 3850
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.679257362355953e-05,
+      "loss": 0.6913,
+      "step": 3851
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.676056338028169e-05,
+      "loss": 0.708,
+      "step": 3852
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.672855313700384e-05,
+      "loss": 0.7072,
+      "step": 3853
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.6696542893726e-05,
+      "loss": 0.5925,
+      "step": 3854
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.666453265044815e-05,
+      "loss": 0.7152,
+      "step": 3855
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.663252240717029e-05,
+      "loss": 0.6288,
+      "step": 3856
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.660051216389245e-05,
+      "loss": 0.7122,
+      "step": 3857
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.65685019206146e-05,
+      "loss": 0.6737,
+      "step": 3858
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.653649167733676e-05,
+      "loss": 0.5596,
+      "step": 3859
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.65044814340589e-05,
+      "loss": 0.7865,
+      "step": 3860
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.647247119078106e-05,
+      "loss": 0.651,
+      "step": 3861
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.64404609475032e-05,
+      "loss": 0.5874,
+      "step": 3862
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.640845070422536e-05,
+      "loss": 0.6591,
+      "step": 3863
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.63764404609475e-05,
+      "loss": 0.6755,
+      "step": 3864
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.634443021766966e-05,
+      "loss": 0.6302,
+      "step": 3865
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.631241997439181e-05,
+      "loss": 0.6733,
+      "step": 3866
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.628040973111395e-05,
+      "loss": 0.6811,
+      "step": 3867
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.624839948783611e-05,
+      "loss": 0.7985,
+      "step": 3868
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.621638924455826e-05,
+      "loss": 0.7298,
+      "step": 3869
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.618437900128042e-05,
+      "loss": 0.6692,
+      "step": 3870
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.615236875800257e-05,
+      "loss": 0.7209,
+      "step": 3871
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.612035851472471e-05,
+      "loss": 0.684,
+      "step": 3872
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.608834827144686e-05,
+      "loss": 0.7312,
+      "step": 3873
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.605633802816902e-05,
+      "loss": 0.6406,
+      "step": 3874
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.602432778489116e-05,
+      "loss": 0.7578,
+      "step": 3875
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.599231754161332e-05,
+      "loss": 0.6969,
+      "step": 3876
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.596030729833547e-05,
+      "loss": 0.6936,
+      "step": 3877
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.592829705505762e-05,
+      "loss": 0.6553,
+      "step": 3878
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.589628681177978e-05,
+      "loss": 0.852,
+      "step": 3879
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.586427656850192e-05,
+      "loss": 0.6684,
+      "step": 3880
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.583226632522408e-05,
+      "loss": 0.7124,
+      "step": 3881
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.580025608194623e-05,
+      "loss": 0.6617,
+      "step": 3882
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.576824583866838e-05,
+      "loss": 0.6071,
+      "step": 3883
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.573623559539052e-05,
+      "loss": 0.6147,
+      "step": 3884
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.570422535211268e-05,
+      "loss": 0.6941,
+      "step": 3885
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.567221510883483e-05,
+      "loss": 0.751,
+      "step": 3886
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.564020486555699e-05,
+      "loss": 0.6525,
+      "step": 3887
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.560819462227913e-05,
+      "loss": 0.6945,
+      "step": 3888
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.557618437900128e-05,
+      "loss": 0.7396,
+      "step": 3889
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.554417413572344e-05,
+      "loss": 0.6141,
+      "step": 3890
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.551216389244559e-05,
+      "loss": 0.5816,
+      "step": 3891
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.548015364916775e-05,
+      "loss": 0.779,
+      "step": 3892
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.544814340588989e-05,
+      "loss": 0.6224,
+      "step": 3893
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.541613316261204e-05,
+      "loss": 0.7992,
+      "step": 3894
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.538412291933418e-05,
+      "loss": 0.5978,
+      "step": 3895
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.535211267605634e-05,
+      "loss": 0.6513,
+      "step": 3896
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.532010243277849e-05,
+      "loss": 0.7274,
+      "step": 3897
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.528809218950065e-05,
+      "loss": 0.7398,
+      "step": 3898
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.525608194622278e-05,
+      "loss": 0.734,
+      "step": 3899
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.522407170294494e-05,
+      "loss": 0.7032,
+      "step": 3900
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.51920614596671e-05,
+      "loss": 0.6029,
+      "step": 3901
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.516005121638925e-05,
+      "loss": 0.6849,
+      "step": 3902
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.512804097311141e-05,
+      "loss": 0.7357,
+      "step": 3903
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.509603072983355e-05,
+      "loss": 0.7815,
+      "step": 3904
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.50640204865557e-05,
+      "loss": 0.6738,
+      "step": 3905
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.503201024327785e-05,
+      "loss": 0.5864,
+      "step": 3906
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.7187,
+      "step": 3907
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.496798975672215e-05,
+      "loss": 0.675,
+      "step": 3908
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.493597951344431e-05,
+      "loss": 0.7426,
+      "step": 3909
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.490396927016645e-05,
+      "loss": 0.8085,
+      "step": 3910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.48719590268886e-05,
+      "loss": 0.6964,
+      "step": 3911
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.483994878361075e-05,
+      "loss": 0.6115,
+      "step": 3912
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.480793854033291e-05,
+      "loss": 0.4978,
+      "step": 3913
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.477592829705507e-05,
+      "loss": 0.7665,
+      "step": 3914
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.474391805377722e-05,
+      "loss": 0.7467,
+      "step": 3915
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.471190781049936e-05,
+      "loss": 0.7245,
+      "step": 3916
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.467989756722151e-05,
+      "loss": 0.7538,
+      "step": 3917
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.464788732394367e-05,
+      "loss": 0.7665,
+      "step": 3918
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.461587708066582e-05,
+      "loss": 0.6474,
+      "step": 3919
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.458386683738798e-05,
+      "loss": 0.7168,
+      "step": 3920
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.455185659411011e-05,
+      "loss": 0.8139,
+      "step": 3921
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.451984635083227e-05,
+      "loss": 0.8239,
+      "step": 3922
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.448783610755441e-05,
+      "loss": 0.7044,
+      "step": 3923
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.445582586427657e-05,
+      "loss": 0.7448,
+      "step": 3924
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.442381562099872e-05,
+      "loss": 0.8341,
+      "step": 3925
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.439180537772087e-05,
+      "loss": 0.6223,
+      "step": 3926
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.435979513444303e-05,
+      "loss": 0.6243,
+      "step": 3927
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.432778489116517e-05,
+      "loss": 0.7108,
+      "step": 3928
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.429577464788733e-05,
+      "loss": 0.7716,
+      "step": 3929
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.426376440460948e-05,
+      "loss": 0.669,
+      "step": 3930
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.423175416133164e-05,
+      "loss": 0.7142,
+      "step": 3931
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.419974391805377e-05,
+      "loss": 0.7969,
+      "step": 3932
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.416773367477593e-05,
+      "loss": 0.6874,
+      "step": 3933
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.413572343149808e-05,
+      "loss": 0.631,
+      "step": 3934
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.410371318822024e-05,
+      "loss": 0.6382,
+      "step": 3935
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.407170294494238e-05,
+      "loss": 0.7322,
+      "step": 3936
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.403969270166453e-05,
+      "loss": 0.5992,
+      "step": 3937
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.400768245838669e-05,
+      "loss": 0.7708,
+      "step": 3938
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.397567221510884e-05,
+      "loss": 0.7074,
+      "step": 3939
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.3943661971831e-05,
+      "loss": 0.663,
+      "step": 3940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.391165172855314e-05,
+      "loss": 0.7439,
+      "step": 3941
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.38796414852753e-05,
+      "loss": 0.6702,
+      "step": 3942
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.384763124199743e-05,
+      "loss": 0.628,
+      "step": 3943
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.38156209987196e-05,
+      "loss": 0.7106,
+      "step": 3944
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.378361075544174e-05,
+      "loss": 0.7252,
+      "step": 3945
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.37516005121639e-05,
+      "loss": 0.6537,
+      "step": 3946
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.371959026888605e-05,
+      "loss": 0.6919,
+      "step": 3947
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.368758002560819e-05,
+      "loss": 0.6448,
+      "step": 3948
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.365556978233035e-05,
+      "loss": 0.7117,
+      "step": 3949
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.36235595390525e-05,
+      "loss": 0.6988,
+      "step": 3950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.359154929577466e-05,
+      "loss": 0.6457,
+      "step": 3951
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.35595390524968e-05,
+      "loss": 0.6315,
+      "step": 3952
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.352752880921895e-05,
+      "loss": 0.7792,
+      "step": 3953
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.34955185659411e-05,
+      "loss": 0.8546,
+      "step": 3954
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.346350832266326e-05,
+      "loss": 0.7533,
+      "step": 3955
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.34314980793854e-05,
+      "loss": 0.6927,
+      "step": 3956
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.339948783610756e-05,
+      "loss": 0.6552,
+      "step": 3957
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.336747759282971e-05,
+      "loss": 0.7095,
+      "step": 3958
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.333546734955186e-05,
+      "loss": 0.8116,
+      "step": 3959
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.330345710627401e-05,
+      "loss": 0.7088,
+      "step": 3960
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.327144686299616e-05,
+      "loss": 0.7238,
+      "step": 3961
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.323943661971832e-05,
+      "loss": 0.7483,
+      "step": 3962
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.320742637644047e-05,
+      "loss": 0.6379,
+      "step": 3963
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.317541613316261e-05,
+      "loss": 0.6307,
+      "step": 3964
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.314340588988476e-05,
+      "loss": 0.7813,
+      "step": 3965
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.311139564660692e-05,
+      "loss": 0.7655,
+      "step": 3966
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.307938540332907e-05,
+      "loss": 0.698,
+      "step": 3967
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.304737516005123e-05,
+      "loss": 0.666,
+      "step": 3968
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.301536491677337e-05,
+      "loss": 0.708,
+      "step": 3969
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.298335467349552e-05,
+      "loss": 0.6959,
+      "step": 3970
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.295134443021768e-05,
+      "loss": 0.6826,
+      "step": 3971
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.291933418693982e-05,
+      "loss": 0.6443,
+      "step": 3972
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.288732394366198e-05,
+      "loss": 0.7073,
+      "step": 3973
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.285531370038413e-05,
+      "loss": 0.6569,
+      "step": 3974
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.282330345710628e-05,
+      "loss": 0.6656,
+      "step": 3975
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.279129321382842e-05,
+      "loss": 0.656,
+      "step": 3976
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.275928297055058e-05,
+      "loss": 0.7316,
+      "step": 3977
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.8091,
+      "step": 3978
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.269526248399489e-05,
+      "loss": 0.6174,
+      "step": 3979
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.266325224071703e-05,
+      "loss": 0.6677,
+      "step": 3980
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.263124199743918e-05,
+      "loss": 0.6702,
+      "step": 3981
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.259923175416134e-05,
+      "loss": 0.6666,
+      "step": 3982
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.256722151088349e-05,
+      "loss": 0.6259,
+      "step": 3983
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.253521126760565e-05,
+      "loss": 0.6483,
+      "step": 3984
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.250320102432779e-05,
+      "loss": 0.6655,
+      "step": 3985
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.247119078104994e-05,
+      "loss": 0.6544,
+      "step": 3986
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.243918053777209e-05,
+      "loss": 0.6454,
+      "step": 3987
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.240717029449424e-05,
+      "loss": 0.7302,
+      "step": 3988
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.237516005121639e-05,
+      "loss": 0.6294,
+      "step": 3989
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.234314980793855e-05,
+      "loss": 0.6582,
+      "step": 3990
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.23111395646607e-05,
+      "loss": 0.7087,
+      "step": 3991
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.227912932138284e-05,
+      "loss": 0.6124,
+      "step": 3992
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.2247119078105e-05,
+      "loss": 0.7594,
+      "step": 3993
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.221510883482715e-05,
+      "loss": 0.7401,
+      "step": 3994
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.218309859154931e-05,
+      "loss": 0.6867,
+      "step": 3995
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.215108834827144e-05,
+      "loss": 0.8066,
+      "step": 3996
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.21190781049936e-05,
+      "loss": 0.7565,
+      "step": 3997
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.208706786171575e-05,
+      "loss": 0.7574,
+      "step": 3998
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.205505761843791e-05,
+      "loss": 0.6164,
+      "step": 3999
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.202304737516005e-05,
+      "loss": 0.8545,
+      "step": 4000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.199103713188221e-05,
+      "loss": 0.7048,
+      "step": 4001
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.195902688860435e-05,
+      "loss": 0.7342,
+      "step": 4002
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.19270166453265e-05,
+      "loss": 0.7089,
+      "step": 4003
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.189500640204867e-05,
+      "loss": 0.6058,
+      "step": 4004
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.186299615877081e-05,
+      "loss": 0.6454,
+      "step": 4005
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.183098591549297e-05,
+      "loss": 0.667,
+      "step": 4006
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.17989756722151e-05,
+      "loss": 0.7388,
+      "step": 4007
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.176696542893726e-05,
+      "loss": 0.6944,
+      "step": 4008
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.173495518565941e-05,
+      "loss": 0.7652,
+      "step": 4009
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.170294494238157e-05,
+      "loss": 0.5636,
+      "step": 4010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.167093469910372e-05,
+      "loss": 0.6254,
+      "step": 4011
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.163892445582588e-05,
+      "loss": 0.6377,
+      "step": 4012
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.160691421254801e-05,
+      "loss": 0.6997,
+      "step": 4013
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.157490396927017e-05,
+      "loss": 0.6629,
+      "step": 4014
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.154289372599232e-05,
+      "loss": 0.7809,
+      "step": 4015
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.151088348271447e-05,
+      "loss": 0.6747,
+      "step": 4016
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.147887323943662e-05,
+      "loss": 0.6807,
+      "step": 4017
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.144686299615877e-05,
+      "loss": 0.7186,
+      "step": 4018
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.141485275288093e-05,
+      "loss": 0.7022,
+      "step": 4019
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.138284250960307e-05,
+      "loss": 0.6441,
+      "step": 4020
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.135083226632523e-05,
+      "loss": 0.7646,
+      "step": 4021
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.131882202304738e-05,
+      "loss": 0.5846,
+      "step": 4022
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.128681177976953e-05,
+      "loss": 0.6752,
+      "step": 4023
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.125480153649167e-05,
+      "loss": 0.8251,
+      "step": 4024
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.122279129321383e-05,
+      "loss": 0.7023,
+      "step": 4025
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.119078104993598e-05,
+      "loss": 0.6423,
+      "step": 4026
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.115877080665814e-05,
+      "loss": 0.6532,
+      "step": 4027
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.112676056338028e-05,
+      "loss": 0.6383,
+      "step": 4028
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.109475032010243e-05,
+      "loss": 0.7451,
+      "step": 4029
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.106274007682459e-05,
+      "loss": 0.7421,
+      "step": 4030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.103072983354674e-05,
+      "loss": 0.7432,
+      "step": 4031
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.09987195902689e-05,
+      "loss": 0.6473,
+      "step": 4032
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.096670934699104e-05,
+      "loss": 0.5982,
+      "step": 4033
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.093469910371319e-05,
+      "loss": 0.6696,
+      "step": 4034
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.090268886043533e-05,
+      "loss": 0.7402,
+      "step": 4035
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.08706786171575e-05,
+      "loss": 0.6765,
+      "step": 4036
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.083866837387964e-05,
+      "loss": 0.6614,
+      "step": 4037
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.08066581306018e-05,
+      "loss": 0.7413,
+      "step": 4038
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.077464788732395e-05,
+      "loss": 0.5844,
+      "step": 4039
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.074263764404609e-05,
+      "loss": 0.668,
+      "step": 4040
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.071062740076825e-05,
+      "loss": 0.6593,
+      "step": 4041
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.06786171574904e-05,
+      "loss": 0.6445,
+      "step": 4042
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.064660691421256e-05,
+      "loss": 0.7949,
+      "step": 4043
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.06145966709347e-05,
+      "loss": 0.5542,
+      "step": 4044
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.058258642765685e-05,
+      "loss": 0.7106,
+      "step": 4045
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.0550576184379e-05,
+      "loss": 0.7065,
+      "step": 4046
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.051856594110116e-05,
+      "loss": 0.6263,
+      "step": 4047
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.04865556978233e-05,
+      "loss": 0.658,
+      "step": 4048
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.045454545454546e-05,
+      "loss": 0.5963,
+      "step": 4049
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.042253521126761e-05,
+      "loss": 0.6989,
+      "step": 4050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.039052496798976e-05,
+      "loss": 0.6599,
+      "step": 4051
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.035851472471192e-05,
+      "loss": 0.8016,
+      "step": 4052
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.032650448143406e-05,
+      "loss": 0.589,
+      "step": 4053
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.029449423815622e-05,
+      "loss": 0.7169,
+      "step": 4054
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.026248399487837e-05,
+      "loss": 0.8287,
+      "step": 4055
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.023047375160051e-05,
+      "loss": 0.6686,
+      "step": 4056
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.019846350832266e-05,
+      "loss": 0.6445,
+      "step": 4057
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.016645326504482e-05,
+      "loss": 0.6958,
+      "step": 4058
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.013444302176697e-05,
+      "loss": 0.8017,
+      "step": 4059
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.010243277848913e-05,
+      "loss": 0.673,
+      "step": 4060
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.007042253521127e-05,
+      "loss": 0.7549,
+      "step": 4061
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.003841229193342e-05,
+      "loss": 0.6873,
+      "step": 4062
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.000640204865558e-05,
+      "loss": 0.7235,
+      "step": 4063
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.997439180537772e-05,
+      "loss": 0.7815,
+      "step": 4064
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.994238156209988e-05,
+      "loss": 0.8735,
+      "step": 4065
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.991037131882202e-05,
+      "loss": 0.7173,
+      "step": 4066
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.987836107554418e-05,
+      "loss": 0.6896,
+      "step": 4067
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.984635083226632e-05,
+      "loss": 0.7077,
+      "step": 4068
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.981434058898848e-05,
+      "loss": 0.6673,
+      "step": 4069
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.978233034571063e-05,
+      "loss": 0.6717,
+      "step": 4070
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.975032010243279e-05,
+      "loss": 0.6405,
+      "step": 4071
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.971830985915493e-05,
+      "loss": 0.6583,
+      "step": 4072
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.968629961587708e-05,
+      "loss": 0.6751,
+      "step": 4073
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.965428937259924e-05,
+      "loss": 0.7041,
+      "step": 4074
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.962227912932139e-05,
+      "loss": 0.5831,
+      "step": 4075
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.959026888604355e-05,
+      "loss": 0.6955,
+      "step": 4076
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.955825864276568e-05,
+      "loss": 0.6883,
+      "step": 4077
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.952624839948784e-05,
+      "loss": 0.7435,
+      "step": 4078
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.949423815620999e-05,
+      "loss": 0.7523,
+      "step": 4079
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.946222791293215e-05,
+      "loss": 0.5529,
+      "step": 4080
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.943021766965429e-05,
+      "loss": 0.7944,
+      "step": 4081
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.939820742637645e-05,
+      "loss": 0.726,
+      "step": 4082
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.93661971830986e-05,
+      "loss": 0.6664,
+      "step": 4083
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.933418693982074e-05,
+      "loss": 0.6828,
+      "step": 4084
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.93021766965429e-05,
+      "loss": 0.8541,
+      "step": 4085
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.927016645326505e-05,
+      "loss": 0.7191,
+      "step": 4086
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.923815620998721e-05,
+      "loss": 0.7801,
+      "step": 4087
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.920614596670934e-05,
+      "loss": 0.6431,
+      "step": 4088
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.91741357234315e-05,
+      "loss": 0.6282,
+      "step": 4089
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.914212548015365e-05,
+      "loss": 0.7305,
+      "step": 4090
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.911011523687581e-05,
+      "loss": 0.6772,
+      "step": 4091
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.907810499359795e-05,
+      "loss": 0.6481,
+      "step": 4092
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.90460947503201e-05,
+      "loss": 0.6414,
+      "step": 4093
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.901408450704226e-05,
+      "loss": 0.8044,
+      "step": 4094
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.89820742637644e-05,
+      "loss": 0.7181,
+      "step": 4095
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.895006402048657e-05,
+      "loss": 0.6738,
+      "step": 4096
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.891805377720871e-05,
+      "loss": 0.6641,
+      "step": 4097
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.888604353393087e-05,
+      "loss": 0.8143,
+      "step": 4098
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8854033290653e-05,
+      "loss": 0.7516,
+      "step": 4099
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.882202304737516e-05,
+      "loss": 0.752,
+      "step": 4100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.879001280409731e-05,
+      "loss": 0.7657,
+      "step": 4101
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.875800256081947e-05,
+      "loss": 0.8042,
+      "step": 4102
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.872599231754162e-05,
+      "loss": 0.7716,
+      "step": 4103
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.869398207426376e-05,
+      "loss": 0.6802,
+      "step": 4104
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.866197183098591e-05,
+      "loss": 0.8591,
+      "step": 4105
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.862996158770807e-05,
+      "loss": 0.711,
+      "step": 4106
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.859795134443022e-05,
+      "loss": 0.6691,
+      "step": 4107
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.856594110115238e-05,
+      "loss": 0.6871,
+      "step": 4108
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.853393085787452e-05,
+      "loss": 0.7233,
+      "step": 4109
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.850192061459667e-05,
+      "loss": 0.7697,
+      "step": 4110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.846991037131883e-05,
+      "loss": 0.6935,
+      "step": 4111
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.843790012804097e-05,
+      "loss": 0.6202,
+      "step": 4112
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.840588988476313e-05,
+      "loss": 0.7316,
+      "step": 4113
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.837387964148528e-05,
+      "loss": 0.6224,
+      "step": 4114
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.834186939820743e-05,
+      "loss": 0.7215,
+      "step": 4115
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.830985915492957e-05,
+      "loss": 0.8197,
+      "step": 4116
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.827784891165173e-05,
+      "loss": 0.6945,
+      "step": 4117
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.824583866837388e-05,
+      "loss": 0.5186,
+      "step": 4118
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.821382842509604e-05,
+      "loss": 0.5521,
+      "step": 4119
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.818181818181818e-05,
+      "loss": 0.5786,
+      "step": 4120
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.814980793854033e-05,
+      "loss": 0.788,
+      "step": 4121
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.811779769526249e-05,
+      "loss": 0.8562,
+      "step": 4122
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.808578745198464e-05,
+      "loss": 0.7054,
+      "step": 4123
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.80537772087068e-05,
+      "loss": 0.7037,
+      "step": 4124
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.802176696542894e-05,
+      "loss": 0.7512,
+      "step": 4125
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.798975672215109e-05,
+      "loss": 0.7021,
+      "step": 4126
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.795774647887324e-05,
+      "loss": 0.7644,
+      "step": 4127
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.79257362355954e-05,
+      "loss": 0.6773,
+      "step": 4128
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.789372599231754e-05,
+      "loss": 0.7256,
+      "step": 4129
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.78617157490397e-05,
+      "loss": 0.6771,
+      "step": 4130
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.782970550576185e-05,
+      "loss": 0.612,
+      "step": 4131
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7797695262484e-05,
+      "loss": 0.6913,
+      "step": 4132
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.776568501920615e-05,
+      "loss": 0.7272,
+      "step": 4133
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.77336747759283e-05,
+      "loss": 0.6781,
+      "step": 4134
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.770166453265046e-05,
+      "loss": 0.834,
+      "step": 4135
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.766965428937259e-05,
+      "loss": 0.6203,
+      "step": 4136
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.763764404609475e-05,
+      "loss": 0.6205,
+      "step": 4137
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.76056338028169e-05,
+      "loss": 0.6366,
+      "step": 4138
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.757362355953906e-05,
+      "loss": 0.6332,
+      "step": 4139
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.75416133162612e-05,
+      "loss": 0.6586,
+      "step": 4140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.750960307298336e-05,
+      "loss": 0.6773,
+      "step": 4141
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.747759282970551e-05,
+      "loss": 0.7366,
+      "step": 4142
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.744558258642766e-05,
+      "loss": 0.7146,
+      "step": 4143
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.741357234314982e-05,
+      "loss": 0.9265,
+      "step": 4144
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.738156209987196e-05,
+      "loss": 0.666,
+      "step": 4145
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.734955185659412e-05,
+      "loss": 0.8244,
+      "step": 4146
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.731754161331625e-05,
+      "loss": 0.7909,
+      "step": 4147
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.728553137003841e-05,
+      "loss": 0.6953,
+      "step": 4148
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.725352112676056e-05,
+      "loss": 0.7129,
+      "step": 4149
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.722151088348272e-05,
+      "loss": 0.6658,
+      "step": 4150
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.718950064020487e-05,
+      "loss": 0.7012,
+      "step": 4151
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.715749039692703e-05,
+      "loss": 0.7427,
+      "step": 4152
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.712548015364917e-05,
+      "loss": 0.6988,
+      "step": 4153
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.709346991037132e-05,
+      "loss": 0.7723,
+      "step": 4154
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.706145966709348e-05,
+      "loss": 0.7739,
+      "step": 4155
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.702944942381563e-05,
+      "loss": 0.7877,
+      "step": 4156
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.699743918053778e-05,
+      "loss": 0.6356,
+      "step": 4157
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.696542893725992e-05,
+      "loss": 0.6921,
+      "step": 4158
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.693341869398208e-05,
+      "loss": 0.6771,
+      "step": 4159
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.690140845070422e-05,
+      "loss": 0.7159,
+      "step": 4160
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.686939820742638e-05,
+      "loss": 0.6351,
+      "step": 4161
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.683738796414853e-05,
+      "loss": 0.7888,
+      "step": 4162
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.680537772087068e-05,
+      "loss": 0.8826,
+      "step": 4163
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.677336747759284e-05,
+      "loss": 0.7291,
+      "step": 4164
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.674135723431498e-05,
+      "loss": 0.7871,
+      "step": 4165
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.670934699103714e-05,
+      "loss": 0.7448,
+      "step": 4166
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.667733674775929e-05,
+      "loss": 0.6442,
+      "step": 4167
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.664532650448145e-05,
+      "loss": 0.7476,
+      "step": 4168
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.661331626120358e-05,
+      "loss": 0.7438,
+      "step": 4169
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.658130601792574e-05,
+      "loss": 0.7108,
+      "step": 4170
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.654929577464789e-05,
+      "loss": 0.7285,
+      "step": 4171
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.651728553137005e-05,
+      "loss": 0.6494,
+      "step": 4172
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.648527528809219e-05,
+      "loss": 0.6914,
+      "step": 4173
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.645326504481434e-05,
+      "loss": 0.7461,
+      "step": 4174
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.64212548015365e-05,
+      "loss": 0.7641,
+      "step": 4175
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.638924455825864e-05,
+      "loss": 0.5777,
+      "step": 4176
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.63572343149808e-05,
+      "loss": 0.6993,
+      "step": 4177
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.632522407170295e-05,
+      "loss": 0.7002,
+      "step": 4178
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.629321382842511e-05,
+      "loss": 0.8369,
+      "step": 4179
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.626120358514724e-05,
+      "loss": 0.6883,
+      "step": 4180
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.62291933418694e-05,
+      "loss": 0.603,
+      "step": 4181
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.619718309859155e-05,
+      "loss": 0.7024,
+      "step": 4182
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.616517285531371e-05,
+      "loss": 0.6882,
+      "step": 4183
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.613316261203586e-05,
+      "loss": 0.8854,
+      "step": 4184
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.6101152368758e-05,
+      "loss": 0.7489,
+      "step": 4185
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.606914212548016e-05,
+      "loss": 0.733,
+      "step": 4186
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.603713188220231e-05,
+      "loss": 0.6671,
+      "step": 4187
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.600512163892447e-05,
+      "loss": 0.787,
+      "step": 4188
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.597311139564661e-05,
+      "loss": 0.6873,
+      "step": 4189
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.594110115236876e-05,
+      "loss": 0.8123,
+      "step": 4190
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.59090909090909e-05,
+      "loss": 0.6701,
+      "step": 4191
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.587708066581307e-05,
+      "loss": 0.6187,
+      "step": 4192
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.584507042253521e-05,
+      "loss": 0.6348,
+      "step": 4193
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.581306017925737e-05,
+      "loss": 0.685,
+      "step": 4194
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.578104993597952e-05,
+      "loss": 0.7194,
+      "step": 4195
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.574903969270166e-05,
+      "loss": 0.6414,
+      "step": 4196
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.571702944942381e-05,
+      "loss": 0.6913,
+      "step": 4197
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.568501920614597e-05,
+      "loss": 0.722,
+      "step": 4198
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.565300896286813e-05,
+      "loss": 0.7322,
+      "step": 4199
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.562099871959028e-05,
+      "loss": 0.7205,
+      "step": 4200
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.558898847631242e-05,
+      "loss": 0.802,
+      "step": 4201
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.555697823303457e-05,
+      "loss": 0.625,
+      "step": 4202
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.552496798975673e-05,
+      "loss": 0.5546,
+      "step": 4203
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.549295774647887e-05,
+      "loss": 0.7215,
+      "step": 4204
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.546094750320103e-05,
+      "loss": 0.6653,
+      "step": 4205
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.542893725992317e-05,
+      "loss": 0.65,
+      "step": 4206
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.539692701664533e-05,
+      "loss": 0.6912,
+      "step": 4207
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.536491677336747e-05,
+      "loss": 0.842,
+      "step": 4208
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.533290653008963e-05,
+      "loss": 0.6956,
+      "step": 4209
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.530089628681178e-05,
+      "loss": 0.7455,
+      "step": 4210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.526888604353394e-05,
+      "loss": 0.6202,
+      "step": 4211
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.523687580025609e-05,
+      "loss": 0.6646,
+      "step": 4212
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.520486555697823e-05,
+      "loss": 0.6089,
+      "step": 4213
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.517285531370039e-05,
+      "loss": 0.6428,
+      "step": 4214
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.514084507042254e-05,
+      "loss": 0.7097,
+      "step": 4215
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.51088348271447e-05,
+      "loss": 0.6738,
+      "step": 4216
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.507682458386683e-05,
+      "loss": 0.7185,
+      "step": 4217
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.504481434058899e-05,
+      "loss": 0.5816,
+      "step": 4218
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.501280409731114e-05,
+      "loss": 0.7458,
+      "step": 4219
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.49807938540333e-05,
+      "loss": 0.7311,
+      "step": 4220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.494878361075544e-05,
+      "loss": 0.7029,
+      "step": 4221
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.49167733674776e-05,
+      "loss": 0.7268,
+      "step": 4222
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.488476312419975e-05,
+      "loss": 0.754,
+      "step": 4223
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.48527528809219e-05,
+      "loss": 0.684,
+      "step": 4224
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.482074263764405e-05,
+      "loss": 0.6252,
+      "step": 4225
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.47887323943662e-05,
+      "loss": 0.5758,
+      "step": 4226
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.475672215108836e-05,
+      "loss": 0.7831,
+      "step": 4227
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.472471190781049e-05,
+      "loss": 0.8511,
+      "step": 4228
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.469270166453265e-05,
+      "loss": 0.7063,
+      "step": 4229
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.46606914212548e-05,
+      "loss": 0.7723,
+      "step": 4230
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.462868117797696e-05,
+      "loss": 0.6786,
+      "step": 4231
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.45966709346991e-05,
+      "loss": 0.6675,
+      "step": 4232
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.456466069142125e-05,
+      "loss": 0.7688,
+      "step": 4233
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.453265044814341e-05,
+      "loss": 0.5662,
+      "step": 4234
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.450064020486556e-05,
+      "loss": 0.6406,
+      "step": 4235
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.446862996158772e-05,
+      "loss": 0.6893,
+      "step": 4236
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.443661971830986e-05,
+      "loss": 0.6588,
+      "step": 4237
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.440460947503202e-05,
+      "loss": 0.7816,
+      "step": 4238
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.437259923175416e-05,
+      "loss": 0.8686,
+      "step": 4239
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.434058898847632e-05,
+      "loss": 0.5776,
+      "step": 4240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.430857874519846e-05,
+      "loss": 0.6695,
+      "step": 4241
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.427656850192062e-05,
+      "loss": 0.7307,
+      "step": 4242
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.424455825864277e-05,
+      "loss": 0.787,
+      "step": 4243
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.421254801536491e-05,
+      "loss": 0.7329,
+      "step": 4244
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.418053777208707e-05,
+      "loss": 0.8,
+      "step": 4245
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.414852752880922e-05,
+      "loss": 0.6676,
+      "step": 4246
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.411651728553138e-05,
+      "loss": 0.6894,
+      "step": 4247
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.408450704225353e-05,
+      "loss": 0.6839,
+      "step": 4248
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.405249679897569e-05,
+      "loss": 0.6021,
+      "step": 4249
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.402048655569782e-05,
+      "loss": 0.6692,
+      "step": 4250
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.398847631241998e-05,
+      "loss": 0.7682,
+      "step": 4251
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.395646606914212e-05,
+      "loss": 0.66,
+      "step": 4252
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.392445582586428e-05,
+      "loss": 0.5609,
+      "step": 4253
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.389244558258643e-05,
+      "loss": 0.795,
+      "step": 4254
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.386043533930858e-05,
+      "loss": 0.6564,
+      "step": 4255
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.382842509603074e-05,
+      "loss": 0.5752,
+      "step": 4256
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.379641485275288e-05,
+      "loss": 0.6477,
+      "step": 4257
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.376440460947504e-05,
+      "loss": 0.7006,
+      "step": 4258
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.373239436619719e-05,
+      "loss": 0.7159,
+      "step": 4259
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.370038412291933e-05,
+      "loss": 0.7437,
+      "step": 4260
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.366837387964148e-05,
+      "loss": 0.7208,
+      "step": 4261
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 0.7812,
+      "step": 4262
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.360435339308579e-05,
+      "loss": 0.6444,
+      "step": 4263
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.357234314980795e-05,
+      "loss": 0.7154,
+      "step": 4264
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.354033290653009e-05,
+      "loss": 0.6947,
+      "step": 4265
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.350832266325224e-05,
+      "loss": 0.5887,
+      "step": 4266
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.34763124199744e-05,
+      "loss": 0.7377,
+      "step": 4267
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.344430217669655e-05,
+      "loss": 0.7933,
+      "step": 4268
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.34122919334187e-05,
+      "loss": 0.6699,
+      "step": 4269
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.338028169014085e-05,
+      "loss": 0.7103,
+      "step": 4270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.3348271446863e-05,
+      "loss": 0.7648,
+      "step": 4271
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.331626120358514e-05,
+      "loss": 0.7612,
+      "step": 4272
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.32842509603073e-05,
+      "loss": 0.6397,
+      "step": 4273
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.325224071702945e-05,
+      "loss": 0.6972,
+      "step": 4274
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.322023047375161e-05,
+      "loss": 0.6618,
+      "step": 4275
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.318822023047376e-05,
+      "loss": 0.6795,
+      "step": 4276
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.31562099871959e-05,
+      "loss": 0.6572,
+      "step": 4277
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.312419974391806e-05,
+      "loss": 0.6309,
+      "step": 4278
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.309218950064021e-05,
+      "loss": 0.7571,
+      "step": 4279
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.306017925736237e-05,
+      "loss": 0.6433,
+      "step": 4280
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.302816901408451e-05,
+      "loss": 0.8364,
+      "step": 4281
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.299615877080666e-05,
+      "loss": 0.7657,
+      "step": 4282
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.29641485275288e-05,
+      "loss": 0.6094,
+      "step": 4283
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.293213828425097e-05,
+      "loss": 0.7642,
+      "step": 4284
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.290012804097311e-05,
+      "loss": 0.8367,
+      "step": 4285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.286811779769527e-05,
+      "loss": 0.6295,
+      "step": 4286
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.28361075544174e-05,
+      "loss": 0.654,
+      "step": 4287
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.280409731113956e-05,
+      "loss": 0.6797,
+      "step": 4288
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.277208706786172e-05,
+      "loss": 0.6715,
+      "step": 4289
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.274007682458387e-05,
+      "loss": 0.7307,
+      "step": 4290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.270806658130603e-05,
+      "loss": 0.7332,
+      "step": 4291
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.267605633802818e-05,
+      "loss": 0.6255,
+      "step": 4292
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.264404609475032e-05,
+      "loss": 0.6085,
+      "step": 4293
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.261203585147247e-05,
+      "loss": 0.6571,
+      "step": 4294
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.258002560819463e-05,
+      "loss": 0.6628,
+      "step": 4295
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.254801536491678e-05,
+      "loss": 0.6376,
+      "step": 4296
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.251600512163893e-05,
+      "loss": 0.7457,
+      "step": 4297
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.248399487836107e-05,
+      "loss": 0.6695,
+      "step": 4298
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.245198463508323e-05,
+      "loss": 0.6622,
+      "step": 4299
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.241997439180537e-05,
+      "loss": 0.7241,
+      "step": 4300
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.238796414852753e-05,
+      "loss": 0.7057,
+      "step": 4301
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.23559539052497e-05,
+      "loss": 0.6475,
+      "step": 4302
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.232394366197183e-05,
+      "loss": 0.6679,
+      "step": 4303
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.229193341869399e-05,
+      "loss": 0.7028,
+      "step": 4304
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.225992317541613e-05,
+      "loss": 0.6229,
+      "step": 4305
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.222791293213829e-05,
+      "loss": 0.5842,
+      "step": 4306
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.219590268886044e-05,
+      "loss": 0.7773,
+      "step": 4307
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.21638924455826e-05,
+      "loss": 0.8217,
+      "step": 4308
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.213188220230473e-05,
+      "loss": 0.6783,
+      "step": 4309
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.209987195902689e-05,
+      "loss": 0.6717,
+      "step": 4310
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.206786171574904e-05,
+      "loss": 0.7146,
+      "step": 4311
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.20358514724712e-05,
+      "loss": 0.6831,
+      "step": 4312
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.200384122919334e-05,
+      "loss": 0.6629,
+      "step": 4313
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.197183098591549e-05,
+      "loss": 0.5963,
+      "step": 4314
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.193982074263765e-05,
+      "loss": 0.7725,
+      "step": 4315
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.19078104993598e-05,
+      "loss": 0.6398,
+      "step": 4316
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.187580025608195e-05,
+      "loss": 0.825,
+      "step": 4317
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.18437900128041e-05,
+      "loss": 0.7755,
+      "step": 4318
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.181177976952626e-05,
+      "loss": 0.6861,
+      "step": 4319
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.177976952624839e-05,
+      "loss": 0.763,
+      "step": 4320
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.174775928297055e-05,
+      "loss": 0.6056,
+      "step": 4321
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.17157490396927e-05,
+      "loss": 0.7769,
+      "step": 4322
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.168373879641486e-05,
+      "loss": 0.712,
+      "step": 4323
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.1651728553137e-05,
+      "loss": 0.764,
+      "step": 4324
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.161971830985915e-05,
+      "loss": 0.6934,
+      "step": 4325
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.158770806658131e-05,
+      "loss": 0.8549,
+      "step": 4326
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.155569782330346e-05,
+      "loss": 0.6893,
+      "step": 4327
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.152368758002562e-05,
+      "loss": 0.652,
+      "step": 4328
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.149167733674776e-05,
+      "loss": 0.6919,
+      "step": 4329
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.145966709346991e-05,
+      "loss": 0.7116,
+      "step": 4330
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.142765685019206e-05,
+      "loss": 0.5749,
+      "step": 4331
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.139564660691422e-05,
+      "loss": 0.6806,
+      "step": 4332
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.136363636363636e-05,
+      "loss": 0.7629,
+      "step": 4333
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.133162612035852e-05,
+      "loss": 0.6681,
+      "step": 4334
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.129961587708067e-05,
+      "loss": 0.6679,
+      "step": 4335
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.126760563380281e-05,
+      "loss": 0.7106,
+      "step": 4336
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.123559539052497e-05,
+      "loss": 0.6676,
+      "step": 4337
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.120358514724712e-05,
+      "loss": 0.7281,
+      "step": 4338
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.117157490396928e-05,
+      "loss": 0.6987,
+      "step": 4339
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.113956466069143e-05,
+      "loss": 0.7291,
+      "step": 4340
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.110755441741357e-05,
+      "loss": 0.8109,
+      "step": 4341
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.107554417413572e-05,
+      "loss": 0.7585,
+      "step": 4342
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.104353393085788e-05,
+      "loss": 0.8332,
+      "step": 4343
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.101152368758003e-05,
+      "loss": 0.7058,
+      "step": 4344
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0979513444302185e-05,
+      "loss": 0.744,
+      "step": 4345
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0947503201024324e-05,
+      "loss": 0.7556,
+      "step": 4346
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.091549295774648e-05,
+      "loss": 0.7647,
+      "step": 4347
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.088348271446863e-05,
+      "loss": 0.74,
+      "step": 4348
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.085147247119078e-05,
+      "loss": 0.7043,
+      "step": 4349
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0819462227912936e-05,
+      "loss": 0.7566,
+      "step": 4350
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.078745198463509e-05,
+      "loss": 0.7474,
+      "step": 4351
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0755441741357235e-05,
+      "loss": 0.687,
+      "step": 4352
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.072343149807939e-05,
+      "loss": 0.6193,
+      "step": 4353
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.069142125480154e-05,
+      "loss": 0.644,
+      "step": 4354
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0659411011523694e-05,
+      "loss": 0.7766,
+      "step": 4355
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.062740076824585e-05,
+      "loss": 0.7196,
+      "step": 4356
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.059539052496799e-05,
+      "loss": 0.6246,
+      "step": 4357
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.056338028169014e-05,
+      "loss": 0.7459,
+      "step": 4358
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.053137003841229e-05,
+      "loss": 0.7069,
+      "step": 4359
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0499359795134446e-05,
+      "loss": 0.8074,
+      "step": 4360
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.04673495518566e-05,
+      "loss": 0.7689,
+      "step": 4361
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.043533930857875e-05,
+      "loss": 0.695,
+      "step": 4362
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.04033290653009e-05,
+      "loss": 0.7715,
+      "step": 4363
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.037131882202305e-05,
+      "loss": 0.6939,
+      "step": 4364
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0339308578745204e-05,
+      "loss": 0.7074,
+      "step": 4365
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.030729833546736e-05,
+      "loss": 0.7525,
+      "step": 4366
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.027528809218951e-05,
+      "loss": 0.7212,
+      "step": 4367
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.024327784891165e-05,
+      "loss": 0.6988,
+      "step": 4368
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.02112676056338e-05,
+      "loss": 0.5779,
+      "step": 4369
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0179257362355956e-05,
+      "loss": 0.7394,
+      "step": 4370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.014724711907811e-05,
+      "loss": 0.6771,
+      "step": 4371
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.011523687580026e-05,
+      "loss": 0.7383,
+      "step": 4372
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.008322663252241e-05,
+      "loss": 0.688,
+      "step": 4373
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.005121638924456e-05,
+      "loss": 0.7324,
+      "step": 4374
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0019206145966714e-05,
+      "loss": 0.7246,
+      "step": 4375
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.998719590268887e-05,
+      "loss": 0.6393,
+      "step": 4376
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.995518565941102e-05,
+      "loss": 0.6148,
+      "step": 4377
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.992317541613317e-05,
+      "loss": 0.7735,
+      "step": 4378
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.989116517285531e-05,
+      "loss": 0.6065,
+      "step": 4379
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.9859154929577465e-05,
+      "loss": 0.826,
+      "step": 4380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.982714468629962e-05,
+      "loss": 0.7072,
+      "step": 4381
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.979513444302177e-05,
+      "loss": 0.8663,
+      "step": 4382
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9763124199743924e-05,
+      "loss": 0.757,
+      "step": 4383
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9731113956466064e-05,
+      "loss": 0.7774,
+      "step": 4384
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9699103713188224e-05,
+      "loss": 0.6458,
+      "step": 4385
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9667093469910377e-05,
+      "loss": 0.7119,
+      "step": 4386
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.963508322663253e-05,
+      "loss": 0.7223,
+      "step": 4387
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.960307298335468e-05,
+      "loss": 0.6482,
+      "step": 4388
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9571062740076836e-05,
+      "loss": 0.6542,
+      "step": 4389
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9539052496798975e-05,
+      "loss": 0.7969,
+      "step": 4390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.950704225352113e-05,
+      "loss": 0.6902,
+      "step": 4391
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.947503201024328e-05,
+      "loss": 0.7135,
+      "step": 4392
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9443021766965434e-05,
+      "loss": 0.6631,
+      "step": 4393
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.941101152368759e-05,
+      "loss": 0.7771,
+      "step": 4394
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9379001280409727e-05,
+      "loss": 0.673,
+      "step": 4395
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.934699103713188e-05,
+      "loss": 0.7324,
+      "step": 4396
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.931498079385403e-05,
+      "loss": 0.5876,
+      "step": 4397
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.928297055057619e-05,
+      "loss": 0.7206,
+      "step": 4398
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9250960307298345e-05,
+      "loss": 0.6627,
+      "step": 4399
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9218950064020485e-05,
+      "loss": 0.7185,
+      "step": 4400
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.918693982074264e-05,
+      "loss": 0.6543,
+      "step": 4401
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.915492957746479e-05,
+      "loss": 0.7285,
+      "step": 4402
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9122919334186944e-05,
+      "loss": 0.588,
+      "step": 4403
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.90909090909091e-05,
+      "loss": 0.6729,
+      "step": 4404
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.905889884763125e-05,
+      "loss": 0.8435,
+      "step": 4405
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.902688860435339e-05,
+      "loss": 0.612,
+      "step": 4406
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.899487836107554e-05,
+      "loss": 0.7692,
+      "step": 4407
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8962868117797695e-05,
+      "loss": 0.691,
+      "step": 4408
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.893085787451985e-05,
+      "loss": 0.6881,
+      "step": 4409
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8898847631242e-05,
+      "loss": 0.7316,
+      "step": 4410
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.886683738796415e-05,
+      "loss": 0.6368,
+      "step": 4411
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.88348271446863e-05,
+      "loss": 0.717,
+      "step": 4412
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8802816901408454e-05,
+      "loss": 0.7198,
+      "step": 4413
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8770806658130607e-05,
+      "loss": 0.6673,
+      "step": 4414
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.873879641485276e-05,
+      "loss": 0.5749,
+      "step": 4415
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.87067861715749e-05,
+      "loss": 0.6923,
+      "step": 4416
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.867477592829705e-05,
+      "loss": 0.7641,
+      "step": 4417
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8642765685019205e-05,
+      "loss": 0.7041,
+      "step": 4418
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.861075544174136e-05,
+      "loss": 0.7096,
+      "step": 4419
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.857874519846351e-05,
+      "loss": 0.6236,
+      "step": 4420
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8546734955185664e-05,
+      "loss": 0.6812,
+      "step": 4421
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.851472471190781e-05,
+      "loss": 0.6029,
+      "step": 4422
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.848271446862996e-05,
+      "loss": 0.6844,
+      "step": 4423
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8450704225352116e-05,
+      "loss": 0.6449,
+      "step": 4424
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.841869398207427e-05,
+      "loss": 0.6183,
+      "step": 4425
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.838668373879642e-05,
+      "loss": 0.7123,
+      "step": 4426
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.835467349551856e-05,
+      "loss": 0.6372,
+      "step": 4427
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8322663252240715e-05,
+      "loss": 0.6719,
+      "step": 4428
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.829065300896287e-05,
+      "loss": 0.7343,
+      "step": 4429
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.825864276568502e-05,
+      "loss": 0.726,
+      "step": 4430
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8226632522407174e-05,
+      "loss": 0.6236,
+      "step": 4431
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.819462227912933e-05,
+      "loss": 0.7199,
+      "step": 4432
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.816261203585147e-05,
+      "loss": 0.7615,
+      "step": 4433
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8130601792573626e-05,
+      "loss": 0.6132,
+      "step": 4434
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.809859154929578e-05,
+      "loss": 0.6717,
+      "step": 4435
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.806658130601793e-05,
+      "loss": 0.7703,
+      "step": 4436
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8034571062740085e-05,
+      "loss": 0.6671,
+      "step": 4437
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8002560819462225e-05,
+      "loss": 0.6972,
+      "step": 4438
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.797055057618438e-05,
+      "loss": 0.7542,
+      "step": 4439
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.793854033290653e-05,
+      "loss": 0.77,
+      "step": 4440
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.7906530089628684e-05,
+      "loss": 0.7279,
+      "step": 4441
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.7874519846350837e-05,
+      "loss": 0.6639,
+      "step": 4442
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.784250960307298e-05,
+      "loss": 0.6736,
+      "step": 4443
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7810499359795136e-05,
+      "loss": 0.7558,
+      "step": 4444
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.777848911651729e-05,
+      "loss": 0.7892,
+      "step": 4445
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.774647887323944e-05,
+      "loss": 0.7413,
+      "step": 4446
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7714468629961595e-05,
+      "loss": 0.674,
+      "step": 4447
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.768245838668375e-05,
+      "loss": 0.688,
+      "step": 4448
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.765044814340589e-05,
+      "loss": 0.6525,
+      "step": 4449
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.761843790012804e-05,
+      "loss": 0.734,
+      "step": 4450
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.758642765685019e-05,
+      "loss": 0.6769,
+      "step": 4451
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7554417413572346e-05,
+      "loss": 0.7794,
+      "step": 4452
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.75224071702945e-05,
+      "loss": 0.6997,
+      "step": 4453
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7490396927016646e-05,
+      "loss": 0.6943,
+      "step": 4454
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.74583866837388e-05,
+      "loss": 0.6885,
+      "step": 4455
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.742637644046095e-05,
+      "loss": 0.7245,
+      "step": 4456
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7394366197183105e-05,
+      "loss": 0.7294,
+      "step": 4457
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.736235595390526e-05,
+      "loss": 0.7344,
+      "step": 4458
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.733034571062741e-05,
+      "loss": 0.5778,
+      "step": 4459
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.729833546734955e-05,
+      "loss": 0.6848,
+      "step": 4460
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.72663252240717e-05,
+      "loss": 0.6988,
+      "step": 4461
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7234314980793856e-05,
+      "loss": 0.65,
+      "step": 4462
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.720230473751601e-05,
+      "loss": 0.714,
+      "step": 4463
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.717029449423816e-05,
+      "loss": 0.7178,
+      "step": 4464
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.713828425096031e-05,
+      "loss": 0.6363,
+      "step": 4465
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.710627400768246e-05,
+      "loss": 0.7004,
+      "step": 4466
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7074263764404614e-05,
+      "loss": 0.6963,
+      "step": 4467
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.704225352112677e-05,
+      "loss": 0.7882,
+      "step": 4468
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.701024327784892e-05,
+      "loss": 0.6559,
+      "step": 4469
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.697823303457106e-05,
+      "loss": 0.752,
+      "step": 4470
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.694622279129321e-05,
+      "loss": 0.6198,
+      "step": 4471
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6914212548015366e-05,
+      "loss": 0.6761,
+      "step": 4472
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.688220230473752e-05,
+      "loss": 0.7993,
+      "step": 4473
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.685019206145967e-05,
+      "loss": 0.7153,
+      "step": 4474
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6818181818181825e-05,
+      "loss": 0.6228,
+      "step": 4475
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.678617157490397e-05,
+      "loss": 0.7156,
+      "step": 4476
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6754161331626124e-05,
+      "loss": 0.6742,
+      "step": 4477
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.672215108834828e-05,
+      "loss": 0.7513,
+      "step": 4478
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.669014084507043e-05,
+      "loss": 0.6541,
+      "step": 4479
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.665813060179258e-05,
+      "loss": 0.7303,
+      "step": 4480
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.662612035851472e-05,
+      "loss": 0.6607,
+      "step": 4481
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6594110115236876e-05,
+      "loss": 0.7702,
+      "step": 4482
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.656209987195903e-05,
+      "loss": 0.7051,
+      "step": 4483
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.653008962868118e-05,
+      "loss": 0.6601,
+      "step": 4484
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6498079385403335e-05,
+      "loss": 0.8148,
+      "step": 4485
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.646606914212549e-05,
+      "loss": 0.7127,
+      "step": 4486
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.643405889884763e-05,
+      "loss": 0.7085,
+      "step": 4487
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.640204865556979e-05,
+      "loss": 0.6758,
+      "step": 4488
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.637003841229194e-05,
+      "loss": 0.7578,
+      "step": 4489
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.633802816901409e-05,
+      "loss": 0.6775,
+      "step": 4490
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6306017925736246e-05,
+      "loss": 0.6774,
+      "step": 4491
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6274007682458385e-05,
+      "loss": 0.6557,
+      "step": 4492
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.624199743918054e-05,
+      "loss": 0.611,
+      "step": 4493
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.620998719590269e-05,
+      "loss": 0.6374,
+      "step": 4494
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6177976952624844e-05,
+      "loss": 0.7335,
+      "step": 4495
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6145966709347e-05,
+      "loss": 0.7986,
+      "step": 4496
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.611395646606914e-05,
+      "loss": 0.6636,
+      "step": 4497
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.608194622279129e-05,
+      "loss": 0.6688,
+      "step": 4498
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.604993597951344e-05,
+      "loss": 0.776,
+      "step": 4499
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6017925736235596e-05,
+      "loss": 0.7217,
+      "step": 4500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.598591549295775e-05,
+      "loss": 0.7535,
+      "step": 4501
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.595390524967991e-05,
+      "loss": 0.7302,
+      "step": 4502
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.592189500640205e-05,
+      "loss": 0.6364,
+      "step": 4503
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.58898847631242e-05,
+      "loss": 0.871,
+      "step": 4504
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5857874519846354e-05,
+      "loss": 0.752,
+      "step": 4505
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.582586427656851e-05,
+      "loss": 0.6547,
+      "step": 4506
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.579385403329066e-05,
+      "loss": 0.6955,
+      "step": 4507
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.57618437900128e-05,
+      "loss": 0.6206,
+      "step": 4508
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.572983354673495e-05,
+      "loss": 0.7641,
+      "step": 4509
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5697823303457106e-05,
+      "loss": 0.6265,
+      "step": 4510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.566581306017926e-05,
+      "loss": 0.7897,
+      "step": 4511
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.563380281690141e-05,
+      "loss": 0.7344,
+      "step": 4512
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.560179257362356e-05,
+      "loss": 0.6338,
+      "step": 4513
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.556978233034571e-05,
+      "loss": 0.5628,
+      "step": 4514
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5537772087067864e-05,
+      "loss": 0.6696,
+      "step": 4515
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.550576184379002e-05,
+      "loss": 0.7186,
+      "step": 4516
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.547375160051217e-05,
+      "loss": 0.6831,
+      "step": 4517
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.544174135723432e-05,
+      "loss": 0.7082,
+      "step": 4518
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.540973111395646e-05,
+      "loss": 0.6906,
+      "step": 4519
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5377720870678615e-05,
+      "loss": 0.6643,
+      "step": 4520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.534571062740077e-05,
+      "loss": 0.6674,
+      "step": 4521
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.531370038412292e-05,
+      "loss": 0.6604,
+      "step": 4522
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5281690140845074e-05,
+      "loss": 0.7554,
+      "step": 4523
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.524967989756722e-05,
+      "loss": 0.7378,
+      "step": 4524
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5217669654289374e-05,
+      "loss": 0.6454,
+      "step": 4525
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.518565941101153e-05,
+      "loss": 0.67,
+      "step": 4526
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.515364916773368e-05,
+      "loss": 0.6865,
+      "step": 4527
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.512163892445583e-05,
+      "loss": 0.7828,
+      "step": 4528
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5089628681177986e-05,
+      "loss": 0.7646,
+      "step": 4529
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5057618437900125e-05,
+      "loss": 0.5903,
+      "step": 4530
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.502560819462228e-05,
+      "loss": 0.7172,
+      "step": 4531
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.499359795134443e-05,
+      "loss": 0.7709,
+      "step": 4532
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4961587708066584e-05,
+      "loss": 0.7749,
+      "step": 4533
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.492957746478874e-05,
+      "loss": 0.6878,
+      "step": 4534
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4897567221510883e-05,
+      "loss": 0.7425,
+      "step": 4535
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4865556978233036e-05,
+      "loss": 0.6696,
+      "step": 4536
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.483354673495519e-05,
+      "loss": 0.8072,
+      "step": 4537
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.480153649167734e-05,
+      "loss": 0.7721,
+      "step": 4538
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4769526248399495e-05,
+      "loss": 0.7294,
+      "step": 4539
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4737516005121635e-05,
+      "loss": 0.6726,
+      "step": 4540
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.470550576184379e-05,
+      "loss": 0.7604,
+      "step": 4541
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.467349551856594e-05,
+      "loss": 0.7294,
+      "step": 4542
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4641485275288094e-05,
+      "loss": 0.6283,
+      "step": 4543
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.460947503201025e-05,
+      "loss": 0.5859,
+      "step": 4544
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.45774647887324e-05,
+      "loss": 0.7952,
+      "step": 4545
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.6305,
+      "step": 4546
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.45134443021767e-05,
+      "loss": 0.6847,
+      "step": 4547
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.448143405889885e-05,
+      "loss": 0.6314,
+      "step": 4548
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4449423815621005e-05,
+      "loss": 0.645,
+      "step": 4549
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.441741357234316e-05,
+      "loss": 0.7204,
+      "step": 4550
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.43854033290653e-05,
+      "loss": 0.7403,
+      "step": 4551
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.435339308578745e-05,
+      "loss": 0.7754,
+      "step": 4552
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4321382842509604e-05,
+      "loss": 0.6294,
+      "step": 4553
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.428937259923176e-05,
+      "loss": 0.6786,
+      "step": 4554
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.425736235595391e-05,
+      "loss": 0.8126,
+      "step": 4555
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.422535211267606e-05,
+      "loss": 0.629,
+      "step": 4556
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.419334186939821e-05,
+      "loss": 0.7313,
+      "step": 4557
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.416133162612036e-05,
+      "loss": 0.7051,
+      "step": 4558
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4129321382842515e-05,
+      "loss": 0.7117,
+      "step": 4559
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.409731113956467e-05,
+      "loss": 0.7072,
+      "step": 4560
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.406530089628682e-05,
+      "loss": 0.7881,
+      "step": 4561
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.403329065300896e-05,
+      "loss": 0.7981,
+      "step": 4562
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4001280409731113e-05,
+      "loss": 0.6737,
+      "step": 4563
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.3969270166453266e-05,
+      "loss": 0.7759,
+      "step": 4564
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.393725992317542e-05,
+      "loss": 0.5139,
+      "step": 4565
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.390524967989757e-05,
+      "loss": 0.6484,
+      "step": 4566
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.387323943661972e-05,
+      "loss": 0.6612,
+      "step": 4567
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.384122919334187e-05,
+      "loss": 0.6611,
+      "step": 4568
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3809218950064025e-05,
+      "loss": 0.7351,
+      "step": 4569
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.377720870678618e-05,
+      "loss": 0.6955,
+      "step": 4570
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.374519846350833e-05,
+      "loss": 0.8184,
+      "step": 4571
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3713188220230484e-05,
+      "loss": 0.5937,
+      "step": 4572
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.368117797695262e-05,
+      "loss": 0.6761,
+      "step": 4573
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3649167733674776e-05,
+      "loss": 0.7019,
+      "step": 4574
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.361715749039693e-05,
+      "loss": 0.7517,
+      "step": 4575
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.358514724711908e-05,
+      "loss": 0.7624,
+      "step": 4576
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3553137003841235e-05,
+      "loss": 0.767,
+      "step": 4577
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.352112676056338e-05,
+      "loss": 0.8186,
+      "step": 4578
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3489116517285535e-05,
+      "loss": 0.7373,
+      "step": 4579
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.345710627400769e-05,
+      "loss": 0.8818,
+      "step": 4580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.342509603072984e-05,
+      "loss": 0.5795,
+      "step": 4581
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3393085787451994e-05,
+      "loss": 0.6596,
+      "step": 4582
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.336107554417413e-05,
+      "loss": 0.6605,
+      "step": 4583
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3329065300896286e-05,
+      "loss": 0.6837,
+      "step": 4584
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.329705505761844e-05,
+      "loss": 0.7729,
+      "step": 4585
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.326504481434059e-05,
+      "loss": 0.6443,
+      "step": 4586
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3233034571062745e-05,
+      "loss": 0.6873,
+      "step": 4587
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.32010243277849e-05,
+      "loss": 0.7204,
+      "step": 4588
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.316901408450704e-05,
+      "loss": 0.662,
+      "step": 4589
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.313700384122919e-05,
+      "loss": 0.6044,
+      "step": 4590
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3104993597951344e-05,
+      "loss": 0.7239,
+      "step": 4591
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.30729833546735e-05,
+      "loss": 0.6173,
+      "step": 4592
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3040973111395656e-05,
+      "loss": 0.6767,
+      "step": 4593
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.3008962868117796e-05,
+      "loss": 0.6335,
+      "step": 4594
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.297695262483995e-05,
+      "loss": 0.646,
+      "step": 4595
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.29449423815621e-05,
+      "loss": 0.6795,
+      "step": 4596
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2912932138284255e-05,
+      "loss": 0.7227,
+      "step": 4597
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.288092189500641e-05,
+      "loss": 0.8293,
+      "step": 4598
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.284891165172856e-05,
+      "loss": 0.8062,
+      "step": 4599
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.28169014084507e-05,
+      "loss": 0.6663,
+      "step": 4600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.278489116517285e-05,
+      "loss": 0.694,
+      "step": 4601
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2752880921895006e-05,
+      "loss": 0.7954,
+      "step": 4602
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.272087067861716e-05,
+      "loss": 0.7402,
+      "step": 4603
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.268886043533931e-05,
+      "loss": 0.6941,
+      "step": 4604
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.265685019206146e-05,
+      "loss": 0.7546,
+      "step": 4605
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.262483994878361e-05,
+      "loss": 0.8806,
+      "step": 4606
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2592829705505765e-05,
+      "loss": 0.6971,
+      "step": 4607
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.256081946222792e-05,
+      "loss": 0.6406,
+      "step": 4608
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.252880921895007e-05,
+      "loss": 0.768,
+      "step": 4609
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.249679897567221e-05,
+      "loss": 0.7656,
+      "step": 4610
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.246478873239436e-05,
+      "loss": 0.6593,
+      "step": 4611
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2432778489116516e-05,
+      "loss": 0.6196,
+      "step": 4612
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.240076824583867e-05,
+      "loss": 0.7465,
+      "step": 4613
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.236875800256082e-05,
+      "loss": 0.7951,
+      "step": 4614
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2336747759282975e-05,
+      "loss": 0.6763,
+      "step": 4615
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.230473751600512e-05,
+      "loss": 0.7806,
+      "step": 4616
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.2272727272727274e-05,
+      "loss": 0.69,
+      "step": 4617
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.224071702944943e-05,
+      "loss": 0.7296,
+      "step": 4618
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.220870678617158e-05,
+      "loss": 0.6436,
+      "step": 4619
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.217669654289373e-05,
+      "loss": 0.7271,
+      "step": 4620
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.214468629961587e-05,
+      "loss": 0.6177,
+      "step": 4621
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.2112676056338026e-05,
+      "loss": 0.7715,
+      "step": 4622
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.208066581306018e-05,
+      "loss": 0.7322,
+      "step": 4623
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.204865556978233e-05,
+      "loss": 0.7109,
+      "step": 4624
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.2016645326504485e-05,
+      "loss": 0.6097,
+      "step": 4625
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.198463508322664e-05,
+      "loss": 0.7534,
+      "step": 4626
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.1952624839948784e-05,
+      "loss": 0.691,
+      "step": 4627
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.192061459667094e-05,
+      "loss": 0.6339,
+      "step": 4628
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.188860435339309e-05,
+      "loss": 0.7437,
+      "step": 4629
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.185659411011524e-05,
+      "loss": 0.7362,
+      "step": 4630
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.1824583866837396e-05,
+      "loss": 0.7501,
+      "step": 4631
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1792573623559536e-05,
+      "loss": 0.6319,
+      "step": 4632
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.176056338028169e-05,
+      "loss": 0.5973,
+      "step": 4633
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.172855313700384e-05,
+      "loss": 0.7697,
+      "step": 4634
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1696542893725995e-05,
+      "loss": 0.7166,
+      "step": 4635
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.166453265044815e-05,
+      "loss": 0.7797,
+      "step": 4636
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1632522407170294e-05,
+      "loss": 0.7228,
+      "step": 4637
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.160051216389245e-05,
+      "loss": 0.7139,
+      "step": 4638
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.15685019206146e-05,
+      "loss": 0.6829,
+      "step": 4639
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.153649167733675e-05,
+      "loss": 0.8862,
+      "step": 4640
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1504481434058906e-05,
+      "loss": 0.749,
+      "step": 4641
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.147247119078106e-05,
+      "loss": 0.6643,
+      "step": 4642
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.14404609475032e-05,
+      "loss": 0.6733,
+      "step": 4643
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.140845070422535e-05,
+      "loss": 0.6581,
+      "step": 4644
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1376440460947504e-05,
+      "loss": 0.6859,
+      "step": 4645
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.134443021766966e-05,
+      "loss": 0.687,
+      "step": 4646
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.131241997439181e-05,
+      "loss": 0.8113,
+      "step": 4647
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1280409731113957e-05,
+      "loss": 0.748,
+      "step": 4648
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.124839948783611e-05,
+      "loss": 0.672,
+      "step": 4649
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.121638924455826e-05,
+      "loss": 0.7456,
+      "step": 4650
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1184379001280416e-05,
+      "loss": 0.7437,
+      "step": 4651
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.115236875800257e-05,
+      "loss": 0.7205,
+      "step": 4652
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.112035851472471e-05,
+      "loss": 0.7365,
+      "step": 4653
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.108834827144686e-05,
+      "loss": 0.7663,
+      "step": 4654
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1056338028169014e-05,
+      "loss": 0.6298,
+      "step": 4655
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.102432778489117e-05,
+      "loss": 0.6288,
+      "step": 4656
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.099231754161332e-05,
+      "loss": 0.6648,
+      "step": 4657
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.096030729833547e-05,
+      "loss": 0.6963,
+      "step": 4658
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.092829705505762e-05,
+      "loss": 0.5815,
+      "step": 4659
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.089628681177977e-05,
+      "loss": 0.7524,
+      "step": 4660
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.0864276568501925e-05,
+      "loss": 0.7411,
+      "step": 4661
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.083226632522408e-05,
+      "loss": 0.6453,
+      "step": 4662
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.080025608194623e-05,
+      "loss": 0.6834,
+      "step": 4663
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.076824583866837e-05,
+      "loss": 0.6964,
+      "step": 4664
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.0736235595390524e-05,
+      "loss": 0.739,
+      "step": 4665
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.070422535211268e-05,
+      "loss": 0.7338,
+      "step": 4666
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.067221510883483e-05,
+      "loss": 0.6449,
+      "step": 4667
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.064020486555698e-05,
+      "loss": 0.7762,
+      "step": 4668
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0608194622279136e-05,
+      "loss": 0.6977,
+      "step": 4669
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.057618437900128e-05,
+      "loss": 0.7494,
+      "step": 4670
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0544174135723435e-05,
+      "loss": 0.7308,
+      "step": 4671
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.051216389244559e-05,
+      "loss": 0.7815,
+      "step": 4672
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.048015364916774e-05,
+      "loss": 0.6845,
+      "step": 4673
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0448143405889894e-05,
+      "loss": 0.6972,
+      "step": 4674
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0416133162612034e-05,
+      "loss": 0.7622,
+      "step": 4675
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.038412291933419e-05,
+      "loss": 0.7243,
+      "step": 4676
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.035211267605634e-05,
+      "loss": 0.7498,
+      "step": 4677
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.032010243277849e-05,
+      "loss": 0.6695,
+      "step": 4678
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0288092189500646e-05,
+      "loss": 0.8248,
+      "step": 4679
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0256081946222785e-05,
+      "loss": 0.7516,
+      "step": 4680
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.022407170294494e-05,
+      "loss": 0.6587,
+      "step": 4681
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.01920614596671e-05,
+      "loss": 0.7755,
+      "step": 4682
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.016005121638925e-05,
+      "loss": 0.6435,
+      "step": 4683
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.0128040973111404e-05,
+      "loss": 0.7501,
+      "step": 4684
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.009603072983356e-05,
+      "loss": 0.6473,
+      "step": 4685
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.0064020486555696e-05,
+      "loss": 0.6043,
+      "step": 4686
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.003201024327785e-05,
+      "loss": 0.6149,
+      "step": 4687
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-05,
+      "loss": 0.7631,
+      "step": 4688
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9967989756722155e-05,
+      "loss": 0.7342,
+      "step": 4689
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.99359795134443e-05,
+      "loss": 0.6764,
+      "step": 4690
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9903969270166455e-05,
+      "loss": 0.7917,
+      "step": 4691
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.987195902688861e-05,
+      "loss": 0.6885,
+      "step": 4692
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9839948783610754e-05,
+      "loss": 0.6854,
+      "step": 4693
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.980793854033291e-05,
+      "loss": 0.822,
+      "step": 4694
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.977592829705506e-05,
+      "loss": 0.727,
+      "step": 4695
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.974391805377721e-05,
+      "loss": 0.7155,
+      "step": 4696
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9711907810499366e-05,
+      "loss": 0.6015,
+      "step": 4697
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.967989756722151e-05,
+      "loss": 0.6175,
+      "step": 4698
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9647887323943665e-05,
+      "loss": 0.8339,
+      "step": 4699
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.961587708066582e-05,
+      "loss": 0.7489,
+      "step": 4700
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9583866837387964e-05,
+      "loss": 0.7504,
+      "step": 4701
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.955185659411012e-05,
+      "loss": 0.772,
+      "step": 4702
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9519846350832264e-05,
+      "loss": 0.6389,
+      "step": 4703
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.948783610755442e-05,
+      "loss": 0.676,
+      "step": 4704
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.945582586427657e-05,
+      "loss": 0.6815,
+      "step": 4705
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.942381562099872e-05,
+      "loss": 0.6939,
+      "step": 4706
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9391805377720876e-05,
+      "loss": 0.7511,
+      "step": 4707
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.935979513444303e-05,
+      "loss": 0.6534,
+      "step": 4708
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9327784891165175e-05,
+      "loss": 0.7317,
+      "step": 4709
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.929577464788733e-05,
+      "loss": 0.7103,
+      "step": 4710
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9263764404609474e-05,
+      "loss": 0.735,
+      "step": 4711
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.923175416133163e-05,
+      "loss": 0.7393,
+      "step": 4712
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.919974391805378e-05,
+      "loss": 0.7038,
+      "step": 4713
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9167733674775926e-05,
+      "loss": 0.7441,
+      "step": 4714
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.913572343149808e-05,
+      "loss": 0.7103,
+      "step": 4715
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.910371318822023e-05,
+      "loss": 0.669,
+      "step": 4716
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9071702944942385e-05,
+      "loss": 0.7988,
+      "step": 4717
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.903969270166454e-05,
+      "loss": 0.6224,
+      "step": 4718
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.9007682458386685e-05,
+      "loss": 0.6805,
+      "step": 4719
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.897567221510884e-05,
+      "loss": 0.7823,
+      "step": 4720
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.894366197183099e-05,
+      "loss": 0.7855,
+      "step": 4721
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.891165172855314e-05,
+      "loss": 0.6678,
+      "step": 4722
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.887964148527529e-05,
+      "loss": 0.6468,
+      "step": 4723
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.884763124199744e-05,
+      "loss": 0.738,
+      "step": 4724
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.881562099871959e-05,
+      "loss": 0.6066,
+      "step": 4725
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.878361075544174e-05,
+      "loss": 0.74,
+      "step": 4726
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8751600512163895e-05,
+      "loss": 0.7714,
+      "step": 4727
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.871959026888605e-05,
+      "loss": 0.6609,
+      "step": 4728
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.86875800256082e-05,
+      "loss": 0.7967,
+      "step": 4729
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.865556978233035e-05,
+      "loss": 0.6913,
+      "step": 4730
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.86235595390525e-05,
+      "loss": 0.6086,
+      "step": 4731
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8591549295774653e-05,
+      "loss": 0.9007,
+      "step": 4732
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.85595390524968e-05,
+      "loss": 0.81,
+      "step": 4733
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.852752880921895e-05,
+      "loss": 0.8083,
+      "step": 4734
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8495518565941106e-05,
+      "loss": 0.6475,
+      "step": 4735
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.846350832266325e-05,
+      "loss": 0.6277,
+      "step": 4736
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8431498079385405e-05,
+      "loss": 0.7513,
+      "step": 4737
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.839948783610755e-05,
+      "loss": 0.7177,
+      "step": 4738
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8367477592829704e-05,
+      "loss": 0.7063,
+      "step": 4739
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8335467349551864e-05,
+      "loss": 0.7433,
+      "step": 4740
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.830345710627401e-05,
+      "loss": 0.592,
+      "step": 4741
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.827144686299616e-05,
+      "loss": 0.7767,
+      "step": 4742
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8239436619718316e-05,
+      "loss": 0.7925,
+      "step": 4743
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.820742637644046e-05,
+      "loss": 0.7216,
+      "step": 4744
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8175416133162615e-05,
+      "loss": 0.7004,
+      "step": 4745
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.814340588988476e-05,
+      "loss": 0.7201,
+      "step": 4746
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8111395646606915e-05,
+      "loss": 0.6837,
+      "step": 4747
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.807938540332907e-05,
+      "loss": 0.7215,
+      "step": 4748
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8047375160051214e-05,
+      "loss": 0.6084,
+      "step": 4749
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.801536491677337e-05,
+      "loss": 0.6654,
+      "step": 4750
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.798335467349552e-05,
+      "loss": 0.6987,
+      "step": 4751
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.795134443021767e-05,
+      "loss": 0.6457,
+      "step": 4752
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.7919334186939826e-05,
+      "loss": 0.7662,
+      "step": 4753
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.788732394366197e-05,
+      "loss": 0.6535,
+      "step": 4754
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.7855313700384125e-05,
+      "loss": 0.633,
+      "step": 4755
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.782330345710628e-05,
+      "loss": 0.706,
+      "step": 4756
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7791293213828424e-05,
+      "loss": 0.6286,
+      "step": 4757
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.775928297055058e-05,
+      "loss": 0.7717,
+      "step": 4758
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.772727272727273e-05,
+      "loss": 0.6654,
+      "step": 4759
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.769526248399488e-05,
+      "loss": 0.6696,
+      "step": 4760
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.766325224071703e-05,
+      "loss": 0.6599,
+      "step": 4761
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.763124199743918e-05,
+      "loss": 0.7847,
+      "step": 4762
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7599231754161336e-05,
+      "loss": 0.6446,
+      "step": 4763
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.756722151088349e-05,
+      "loss": 0.6399,
+      "step": 4764
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7535211267605635e-05,
+      "loss": 0.7396,
+      "step": 4765
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.750320102432779e-05,
+      "loss": 0.8012,
+      "step": 4766
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.747119078104994e-05,
+      "loss": 0.6369,
+      "step": 4767
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.743918053777209e-05,
+      "loss": 0.7713,
+      "step": 4768
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.740717029449424e-05,
+      "loss": 0.8271,
+      "step": 4769
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.737516005121639e-05,
+      "loss": 0.7934,
+      "step": 4770
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.734314980793854e-05,
+      "loss": 0.7019,
+      "step": 4771
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.731113956466069e-05,
+      "loss": 0.6811,
+      "step": 4772
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7279129321382845e-05,
+      "loss": 0.7796,
+      "step": 4773
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7247119078105e-05,
+      "loss": 0.7124,
+      "step": 4774
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.721510883482715e-05,
+      "loss": 0.7255,
+      "step": 4775
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.71830985915493e-05,
+      "loss": 0.659,
+      "step": 4776
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.715108834827145e-05,
+      "loss": 0.6891,
+      "step": 4777
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7119078104993604e-05,
+      "loss": 0.5844,
+      "step": 4778
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.708706786171575e-05,
+      "loss": 0.6601,
+      "step": 4779
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.70550576184379e-05,
+      "loss": 0.67,
+      "step": 4780
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.702304737516005e-05,
+      "loss": 0.6631,
+      "step": 4781
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.69910371318822e-05,
+      "loss": 0.6761,
+      "step": 4782
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6959026888604355e-05,
+      "loss": 0.7824,
+      "step": 4783
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.69270166453265e-05,
+      "loss": 0.7102,
+      "step": 4784
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.689500640204866e-05,
+      "loss": 0.6236,
+      "step": 4785
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6862996158770814e-05,
+      "loss": 0.7507,
+      "step": 4786
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.683098591549296e-05,
+      "loss": 0.6627,
+      "step": 4787
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6798975672215114e-05,
+      "loss": 0.6767,
+      "step": 4788
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.676696542893726e-05,
+      "loss": 0.6936,
+      "step": 4789
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.673495518565941e-05,
+      "loss": 0.7076,
+      "step": 4790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6702944942381566e-05,
+      "loss": 0.7149,
+      "step": 4791
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.667093469910371e-05,
+      "loss": 0.6917,
+      "step": 4792
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6638924455825865e-05,
+      "loss": 0.6995,
+      "step": 4793
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.660691421254802e-05,
+      "loss": 0.7391,
+      "step": 4794
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6574903969270164e-05,
+      "loss": 0.6332,
+      "step": 4795
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.654289372599232e-05,
+      "loss": 0.7822,
+      "step": 4796
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.651088348271447e-05,
+      "loss": 0.6418,
+      "step": 4797
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.647887323943662e-05,
+      "loss": 0.7166,
+      "step": 4798
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6446862996158776e-05,
+      "loss": 0.7868,
+      "step": 4799
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.641485275288092e-05,
+      "loss": 0.6917,
+      "step": 4800
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6382842509603076e-05,
+      "loss": 0.8245,
+      "step": 4801
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.635083226632523e-05,
+      "loss": 0.7129,
+      "step": 4802
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6318822023047375e-05,
+      "loss": 0.7607,
+      "step": 4803
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.628681177976953e-05,
+      "loss": 0.6587,
+      "step": 4804
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.625480153649168e-05,
+      "loss": 0.6796,
+      "step": 4805
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.622279129321383e-05,
+      "loss": 0.7939,
+      "step": 4806
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.619078104993598e-05,
+      "loss": 0.7774,
+      "step": 4807
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.615877080665813e-05,
+      "loss": 0.6594,
+      "step": 4808
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.6126760563380286e-05,
+      "loss": 0.6571,
+      "step": 4809
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.609475032010244e-05,
+      "loss": 0.6773,
+      "step": 4810
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.6062740076824585e-05,
+      "loss": 0.6974,
+      "step": 4811
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.603072983354674e-05,
+      "loss": 0.6961,
+      "step": 4812
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.599871959026889e-05,
+      "loss": 0.7599,
+      "step": 4813
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.596670934699104e-05,
+      "loss": 0.7203,
+      "step": 4814
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.593469910371319e-05,
+      "loss": 0.603,
+      "step": 4815
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.590268886043534e-05,
+      "loss": 0.7209,
+      "step": 4816
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.587067861715749e-05,
+      "loss": 0.7206,
+      "step": 4817
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.583866837387964e-05,
+      "loss": 0.6556,
+      "step": 4818
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5806658130601796e-05,
+      "loss": 0.6754,
+      "step": 4819
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.577464788732395e-05,
+      "loss": 0.6918,
+      "step": 4820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.57426376440461e-05,
+      "loss": 0.7462,
+      "step": 4821
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.571062740076825e-05,
+      "loss": 0.7148,
+      "step": 4822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.56786171574904e-05,
+      "loss": 0.6594,
+      "step": 4823
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5646606914212554e-05,
+      "loss": 0.7812,
+      "step": 4824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.56145966709347e-05,
+      "loss": 0.7585,
+      "step": 4825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.558258642765685e-05,
+      "loss": 0.7097,
+      "step": 4826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5550576184379e-05,
+      "loss": 0.7252,
+      "step": 4827
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.551856594110115e-05,
+      "loss": 0.7843,
+      "step": 4828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5486555697823306e-05,
+      "loss": 0.7404,
+      "step": 4829
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.7354,
+      "step": 4830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.542253521126761e-05,
+      "loss": 0.771,
+      "step": 4831
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5390524967989765e-05,
+      "loss": 0.8947,
+      "step": 4832
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.535851472471191e-05,
+      "loss": 0.7144,
+      "step": 4833
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5326504481434064e-05,
+      "loss": 0.6156,
+      "step": 4834
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.529449423815621e-05,
+      "loss": 0.7101,
+      "step": 4835
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.526248399487836e-05,
+      "loss": 0.6609,
+      "step": 4836
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5230473751600516e-05,
+      "loss": 0.655,
+      "step": 4837
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.519846350832266e-05,
+      "loss": 0.7163,
+      "step": 4838
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5166453265044815e-05,
+      "loss": 0.8927,
+      "step": 4839
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.513444302176697e-05,
+      "loss": 0.7132,
+      "step": 4840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5102432778489115e-05,
+      "loss": 0.7936,
+      "step": 4841
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.507042253521127e-05,
+      "loss": 0.7244,
+      "step": 4842
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.503841229193342e-05,
+      "loss": 0.684,
+      "step": 4843
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.5006402048655574e-05,
+      "loss": 0.6798,
+      "step": 4844
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4974391805377727e-05,
+      "loss": 0.6836,
+      "step": 4845
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.494238156209987e-05,
+      "loss": 0.6527,
+      "step": 4846
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4910371318822026e-05,
+      "loss": 0.7609,
+      "step": 4847
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.487836107554418e-05,
+      "loss": 0.7714,
+      "step": 4848
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4846350832266325e-05,
+      "loss": 0.7606,
+      "step": 4849
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.481434058898848e-05,
+      "loss": 0.8353,
+      "step": 4850
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4782330345710624e-05,
+      "loss": 0.7442,
+      "step": 4851
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.475032010243278e-05,
+      "loss": 0.7138,
+      "step": 4852
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.471830985915493e-05,
+      "loss": 0.8241,
+      "step": 4853
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.468629961587708e-05,
+      "loss": 0.7959,
+      "step": 4854
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4654289372599236e-05,
+      "loss": 0.6873,
+      "step": 4855
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.462227912932139e-05,
+      "loss": 0.646,
+      "step": 4856
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4590268886043536e-05,
+      "loss": 0.5776,
+      "step": 4857
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.455825864276569e-05,
+      "loss": 0.6885,
+      "step": 4858
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.452624839948784e-05,
+      "loss": 0.7945,
+      "step": 4859
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.449423815620999e-05,
+      "loss": 0.6438,
+      "step": 4860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.446222791293214e-05,
+      "loss": 0.694,
+      "step": 4861
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.443021766965429e-05,
+      "loss": 0.649,
+      "step": 4862
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.439820742637644e-05,
+      "loss": 0.8262,
+      "step": 4863
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.436619718309859e-05,
+      "loss": 0.758,
+      "step": 4864
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4334186939820746e-05,
+      "loss": 0.7239,
+      "step": 4865
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.43021766965429e-05,
+      "loss": 0.707,
+      "step": 4866
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.427016645326505e-05,
+      "loss": 0.6235,
+      "step": 4867
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.42381562099872e-05,
+      "loss": 0.6687,
+      "step": 4868
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.420614596670935e-05,
+      "loss": 0.6954,
+      "step": 4869
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.41741357234315e-05,
+      "loss": 0.7443,
+      "step": 4870
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.414212548015365e-05,
+      "loss": 0.7698,
+      "step": 4871
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.4110115236875804e-05,
+      "loss": 0.721,
+      "step": 4872
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.407810499359795e-05,
+      "loss": 0.7284,
+      "step": 4873
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.40460947503201e-05,
+      "loss": 0.5948,
+      "step": 4874
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.4014084507042256e-05,
+      "loss": 0.6881,
+      "step": 4875
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.398207426376441e-05,
+      "loss": 0.6502,
+      "step": 4876
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.395006402048656e-05,
+      "loss": 0.6559,
+      "step": 4877
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.391805377720871e-05,
+      "loss": 0.7202,
+      "step": 4878
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.388604353393086e-05,
+      "loss": 0.7502,
+      "step": 4879
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.3854033290653014e-05,
+      "loss": 0.5989,
+      "step": 4880
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.382202304737516e-05,
+      "loss": 0.6977,
+      "step": 4881
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.379001280409731e-05,
+      "loss": 0.6239,
+      "step": 4882
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3758002560819466e-05,
+      "loss": 0.6674,
+      "step": 4883
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.372599231754161e-05,
+      "loss": 0.7696,
+      "step": 4884
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3693982074263766e-05,
+      "loss": 0.7317,
+      "step": 4885
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.366197183098591e-05,
+      "loss": 0.7287,
+      "step": 4886
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3629961587708065e-05,
+      "loss": 0.6441,
+      "step": 4887
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.359795134443022e-05,
+      "loss": 0.7778,
+      "step": 4888
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.356594110115237e-05,
+      "loss": 0.8689,
+      "step": 4889
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3533930857874524e-05,
+      "loss": 0.7159,
+      "step": 4890
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.350192061459668e-05,
+      "loss": 0.6478,
+      "step": 4891
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.346991037131882e-05,
+      "loss": 0.7319,
+      "step": 4892
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3437900128040976e-05,
+      "loss": 0.5295,
+      "step": 4893
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.340588988476313e-05,
+      "loss": 0.6784,
+      "step": 4894
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3373879641485275e-05,
+      "loss": 0.6931,
+      "step": 4895
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.334186939820743e-05,
+      "loss": 0.7339,
+      "step": 4896
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3309859154929575e-05,
+      "loss": 0.7547,
+      "step": 4897
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.327784891165173e-05,
+      "loss": 0.8045,
+      "step": 4898
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.324583866837388e-05,
+      "loss": 0.7108,
+      "step": 4899
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3213828425096034e-05,
+      "loss": 0.5908,
+      "step": 4900
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.318181818181819e-05,
+      "loss": 0.6588,
+      "step": 4901
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.314980793854034e-05,
+      "loss": 0.7149,
+      "step": 4902
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3117797695262486e-05,
+      "loss": 0.8251,
+      "step": 4903
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.308578745198464e-05,
+      "loss": 0.7329,
+      "step": 4904
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3053777208706785e-05,
+      "loss": 0.7612,
+      "step": 4905
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.302176696542894e-05,
+      "loss": 0.7761,
+      "step": 4906
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.298975672215109e-05,
+      "loss": 0.7173,
+      "step": 4907
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.295774647887324e-05,
+      "loss": 0.7757,
+      "step": 4908
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.292573623559539e-05,
+      "loss": 0.673,
+      "step": 4909
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2893725992317543e-05,
+      "loss": 0.6965,
+      "step": 4910
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2861715749039696e-05,
+      "loss": 0.5475,
+      "step": 4911
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.282970550576185e-05,
+      "loss": 0.6816,
+      "step": 4912
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2797695262483996e-05,
+      "loss": 0.6167,
+      "step": 4913
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.276568501920615e-05,
+      "loss": 0.722,
+      "step": 4914
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.27336747759283e-05,
+      "loss": 0.6437,
+      "step": 4915
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.270166453265045e-05,
+      "loss": 0.8197,
+      "step": 4916
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.26696542893726e-05,
+      "loss": 0.7746,
+      "step": 4917
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2637644046094754e-05,
+      "loss": 0.8358,
+      "step": 4918
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.26056338028169e-05,
+      "loss": 0.73,
+      "step": 4919
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.257362355953905e-05,
+      "loss": 0.6789,
+      "step": 4920
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2541613316261206e-05,
+      "loss": 0.7539,
+      "step": 4921
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.250960307298336e-05,
+      "loss": 0.7211,
+      "step": 4922
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.247759282970551e-05,
+      "loss": 0.5559,
+      "step": 4923
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.244558258642766e-05,
+      "loss": 0.7992,
+      "step": 4924
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.241357234314981e-05,
+      "loss": 0.7098,
+      "step": 4925
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2381562099871964e-05,
+      "loss": 0.7098,
+      "step": 4926
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.234955185659411e-05,
+      "loss": 0.6701,
+      "step": 4927
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2317541613316264e-05,
+      "loss": 0.6412,
+      "step": 4928
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.228553137003842e-05,
+      "loss": 0.5857,
+      "step": 4929
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.225352112676056e-05,
+      "loss": 0.6688,
+      "step": 4930
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2221510883482716e-05,
+      "loss": 0.7295,
+      "step": 4931
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.218950064020486e-05,
+      "loss": 0.8218,
+      "step": 4932
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.2157490396927015e-05,
+      "loss": 0.77,
+      "step": 4933
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.2125480153649175e-05,
+      "loss": 0.7448,
+      "step": 4934
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.209346991037132e-05,
+      "loss": 0.6085,
+      "step": 4935
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.2061459667093474e-05,
+      "loss": 0.6356,
+      "step": 4936
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.202944942381563e-05,
+      "loss": 0.6851,
+      "step": 4937
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1997439180537773e-05,
+      "loss": 0.7273,
+      "step": 4938
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1965428937259926e-05,
+      "loss": 0.8714,
+      "step": 4939
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.193341869398207e-05,
+      "loss": 0.7041,
+      "step": 4940
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1901408450704226e-05,
+      "loss": 0.764,
+      "step": 4941
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.186939820742638e-05,
+      "loss": 0.6206,
+      "step": 4942
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1837387964148525e-05,
+      "loss": 0.6843,
+      "step": 4943
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.180537772087068e-05,
+      "loss": 0.5801,
+      "step": 4944
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.177336747759283e-05,
+      "loss": 0.6745,
+      "step": 4945
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1741357234314984e-05,
+      "loss": 0.63,
+      "step": 4946
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.170934699103714e-05,
+      "loss": 0.8455,
+      "step": 4947
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.167733674775928e-05,
+      "loss": 0.7732,
+      "step": 4948
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1645326504481436e-05,
+      "loss": 0.7804,
+      "step": 4949
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.161331626120359e-05,
+      "loss": 0.6926,
+      "step": 4950
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1581306017925735e-05,
+      "loss": 0.6171,
+      "step": 4951
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.154929577464789e-05,
+      "loss": 0.6728,
+      "step": 4952
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.151728553137004e-05,
+      "loss": 0.675,
+      "step": 4953
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.148527528809219e-05,
+      "loss": 0.7139,
+      "step": 4954
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.145326504481434e-05,
+      "loss": 0.7632,
+      "step": 4955
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1421254801536494e-05,
+      "loss": 0.6265,
+      "step": 4956
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.138924455825865e-05,
+      "loss": 0.7263,
+      "step": 4957
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.13572343149808e-05,
+      "loss": 0.6807,
+      "step": 4958
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1325224071702946e-05,
+      "loss": 0.6963,
+      "step": 4959
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.12932138284251e-05,
+      "loss": 0.663,
+      "step": 4960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.126120358514725e-05,
+      "loss": 0.6964,
+      "step": 4961
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.12291933418694e-05,
+      "loss": 0.6245,
+      "step": 4962
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.119718309859155e-05,
+      "loss": 0.6982,
+      "step": 4963
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1165172855313704e-05,
+      "loss": 0.6507,
+      "step": 4964
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.113316261203585e-05,
+      "loss": 0.596,
+      "step": 4965
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1101152368758003e-05,
+      "loss": 0.6324,
+      "step": 4966
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1069142125480156e-05,
+      "loss": 0.7334,
+      "step": 4967
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.103713188220231e-05,
+      "loss": 0.679,
+      "step": 4968
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.100512163892446e-05,
+      "loss": 0.6683,
+      "step": 4969
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.097311139564661e-05,
+      "loss": 0.593,
+      "step": 4970
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.094110115236876e-05,
+      "loss": 0.7005,
+      "step": 4971
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0909090909090915e-05,
+      "loss": 0.7182,
+      "step": 4972
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.087708066581306e-05,
+      "loss": 0.7241,
+      "step": 4973
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0845070422535214e-05,
+      "loss": 0.7283,
+      "step": 4974
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.081306017925736e-05,
+      "loss": 0.7298,
+      "step": 4975
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.078104993597951e-05,
+      "loss": 0.75,
+      "step": 4976
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0749039692701666e-05,
+      "loss": 0.6942,
+      "step": 4977
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.071702944942381e-05,
+      "loss": 0.6125,
+      "step": 4978
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.068501920614597e-05,
+      "loss": 0.6476,
+      "step": 4979
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0653008962868125e-05,
+      "loss": 0.6045,
+      "step": 4980
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.062099871959027e-05,
+      "loss": 0.6285,
+      "step": 4981
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0588988476312425e-05,
+      "loss": 0.698,
+      "step": 4982
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.055697823303457e-05,
+      "loss": 0.7272,
+      "step": 4983
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0524967989756724e-05,
+      "loss": 0.6678,
+      "step": 4984
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.049295774647888e-05,
+      "loss": 0.7069,
+      "step": 4985
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.046094750320102e-05,
+      "loss": 0.6096,
+      "step": 4986
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0428937259923176e-05,
+      "loss": 0.7045,
+      "step": 4987
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.039692701664533e-05,
+      "loss": 0.7585,
+      "step": 4988
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0364916773367475e-05,
+      "loss": 0.6947,
+      "step": 4989
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.033290653008963e-05,
+      "loss": 0.7312,
+      "step": 4990
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.030089628681178e-05,
+      "loss": 0.7574,
+      "step": 4991
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0268886043533934e-05,
+      "loss": 0.6886,
+      "step": 4992
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.023687580025609e-05,
+      "loss": 0.7358,
+      "step": 4993
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0204865556978234e-05,
+      "loss": 0.7224,
+      "step": 4994
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0172855313700387e-05,
+      "loss": 0.7324,
+      "step": 4995
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.014084507042254e-05,
+      "loss": 0.7573,
+      "step": 4996
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0108834827144686e-05,
+      "loss": 0.8397,
+      "step": 4997
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.007682458386684e-05,
+      "loss": 0.6686,
+      "step": 4998
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.004481434058899e-05,
+      "loss": 0.6521,
+      "step": 4999
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.001280409731114e-05,
+      "loss": 0.6891,
+      "step": 5000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.998079385403329e-05,
+      "loss": 0.6021,
+      "step": 5001
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9948783610755444e-05,
+      "loss": 0.6857,
+      "step": 5002
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.99167733674776e-05,
+      "loss": 0.6741,
+      "step": 5003
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.988476312419975e-05,
+      "loss": 0.6328,
+      "step": 5004
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9852752880921896e-05,
+      "loss": 0.6008,
+      "step": 5005
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.982074263764405e-05,
+      "loss": 0.6513,
+      "step": 5006
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.97887323943662e-05,
+      "loss": 0.6408,
+      "step": 5007
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.975672215108835e-05,
+      "loss": 0.6809,
+      "step": 5008
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.97247119078105e-05,
+      "loss": 0.5319,
+      "step": 5009
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.969270166453265e-05,
+      "loss": 0.7417,
+      "step": 5010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.96606914212548e-05,
+      "loss": 0.581,
+      "step": 5011
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9628681177976954e-05,
+      "loss": 0.7453,
+      "step": 5012
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.959667093469911e-05,
+      "loss": 0.7403,
+      "step": 5013
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.956466069142126e-05,
+      "loss": 0.6948,
+      "step": 5014
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.953265044814341e-05,
+      "loss": 0.7438,
+      "step": 5015
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.950064020486556e-05,
+      "loss": 0.7043,
+      "step": 5016
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.946862996158771e-05,
+      "loss": 0.5647,
+      "step": 5017
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.943661971830986e-05,
+      "loss": 0.6786,
+      "step": 5018
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.940460947503201e-05,
+      "loss": 0.8185,
+      "step": 5019
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9372599231754164e-05,
+      "loss": 0.5862,
+      "step": 5020
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.934058898847631e-05,
+      "loss": 0.846,
+      "step": 5021
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9308578745198464e-05,
+      "loss": 0.7179,
+      "step": 5022
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9276568501920617e-05,
+      "loss": 0.6444,
+      "step": 5023
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.924455825864277e-05,
+      "loss": 0.6342,
+      "step": 5024
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.921254801536492e-05,
+      "loss": 0.5903,
+      "step": 5025
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.918053777208707e-05,
+      "loss": 0.678,
+      "step": 5026
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.914852752880922e-05,
+      "loss": 0.7005,
+      "step": 5027
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9116517285531375e-05,
+      "loss": 0.7838,
+      "step": 5028
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.908450704225352e-05,
+      "loss": 0.7213,
+      "step": 5029
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.9052496798975674e-05,
+      "loss": 0.6833,
+      "step": 5030
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.902048655569783e-05,
+      "loss": 0.5465,
+      "step": 5031
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.898847631241997e-05,
+      "loss": 0.6386,
+      "step": 5032
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8956466069142126e-05,
+      "loss": 0.6472,
+      "step": 5033
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.892445582586428e-05,
+      "loss": 0.6218,
+      "step": 5034
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8892445582586426e-05,
+      "loss": 0.507,
+      "step": 5035
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.886043533930858e-05,
+      "loss": 0.6736,
+      "step": 5036
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.882842509603073e-05,
+      "loss": 0.7069,
+      "step": 5037
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8796414852752885e-05,
+      "loss": 0.6537,
+      "step": 5038
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.876440460947504e-05,
+      "loss": 0.7206,
+      "step": 5039
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8732394366197184e-05,
+      "loss": 0.6005,
+      "step": 5040
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.870038412291934e-05,
+      "loss": 0.6545,
+      "step": 5041
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.866837387964149e-05,
+      "loss": 0.5872,
+      "step": 5042
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.5974,
+      "step": 5043
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.860435339308579e-05,
+      "loss": 0.6325,
+      "step": 5044
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8572343149807935e-05,
+      "loss": 0.6789,
+      "step": 5045
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.854033290653009e-05,
+      "loss": 0.6971,
+      "step": 5046
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.850832266325224e-05,
+      "loss": 0.6216,
+      "step": 5047
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8476312419974394e-05,
+      "loss": 0.6724,
+      "step": 5048
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.844430217669655e-05,
+      "loss": 0.6959,
+      "step": 5049
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.84122919334187e-05,
+      "loss": 0.8137,
+      "step": 5050
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8380281690140847e-05,
+      "loss": 0.7102,
+      "step": 5051
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8348271446863e-05,
+      "loss": 0.566,
+      "step": 5052
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8316261203585146e-05,
+      "loss": 0.6603,
+      "step": 5053
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.82842509603073e-05,
+      "loss": 0.6947,
+      "step": 5054
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.825224071702945e-05,
+      "loss": 0.6882,
+      "step": 5055
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.82202304737516e-05,
+      "loss": 0.7073,
+      "step": 5056
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.818822023047375e-05,
+      "loss": 0.6687,
+      "step": 5057
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8156209987195904e-05,
+      "loss": 0.6466,
+      "step": 5058
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.812419974391806e-05,
+      "loss": 0.7121,
+      "step": 5059
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.809218950064021e-05,
+      "loss": 0.6158,
+      "step": 5060
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8060179257362356e-05,
+      "loss": 0.7739,
+      "step": 5061
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.802816901408451e-05,
+      "loss": 0.6347,
+      "step": 5062
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.799615877080666e-05,
+      "loss": 0.7955,
+      "step": 5063
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.796414852752881e-05,
+      "loss": 0.6903,
+      "step": 5064
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.793213828425096e-05,
+      "loss": 0.6315,
+      "step": 5065
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7900128040973115e-05,
+      "loss": 0.7594,
+      "step": 5066
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.786811779769526e-05,
+      "loss": 0.6586,
+      "step": 5067
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7836107554417414e-05,
+      "loss": 0.6865,
+      "step": 5068
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.780409731113957e-05,
+      "loss": 0.6871,
+      "step": 5069
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.777208706786172e-05,
+      "loss": 0.7114,
+      "step": 5070
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.774007682458387e-05,
+      "loss": 0.7271,
+      "step": 5071
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.770806658130602e-05,
+      "loss": 0.6443,
+      "step": 5072
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.767605633802817e-05,
+      "loss": 0.6142,
+      "step": 5073
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7644046094750325e-05,
+      "loss": 0.676,
+      "step": 5074
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.761203585147247e-05,
+      "loss": 0.5874,
+      "step": 5075
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7580025608194624e-05,
+      "loss": 0.6664,
+      "step": 5076
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.754801536491678e-05,
+      "loss": 0.6475,
+      "step": 5077
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7516005121638924e-05,
+      "loss": 0.7103,
+      "step": 5078
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7483994878361077e-05,
+      "loss": 0.6781,
+      "step": 5079
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.745198463508322e-05,
+      "loss": 0.7135,
+      "step": 5080
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7419974391805376e-05,
+      "loss": 0.7155,
+      "step": 5081
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7387964148527536e-05,
+      "loss": 0.651,
+      "step": 5082
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.735595390524968e-05,
+      "loss": 0.6513,
+      "step": 5083
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7323943661971835e-05,
+      "loss": 0.615,
+      "step": 5084
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.729193341869399e-05,
+      "loss": 0.6525,
+      "step": 5085
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7259923175416134e-05,
+      "loss": 0.6284,
+      "step": 5086
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.722791293213829e-05,
+      "loss": 0.7466,
+      "step": 5087
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.719590268886043e-05,
+      "loss": 0.7019,
+      "step": 5088
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7163892445582586e-05,
+      "loss": 0.7374,
+      "step": 5089
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.713188220230474e-05,
+      "loss": 0.6759,
+      "step": 5090
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7099871959026886e-05,
+      "loss": 0.6021,
+      "step": 5091
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.706786171574904e-05,
+      "loss": 0.6808,
+      "step": 5092
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.703585147247119e-05,
+      "loss": 0.694,
+      "step": 5093
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.7003841229193345e-05,
+      "loss": 0.6169,
+      "step": 5094
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.69718309859155e-05,
+      "loss": 0.7538,
+      "step": 5095
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.693982074263765e-05,
+      "loss": 0.6137,
+      "step": 5096
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.69078104993598e-05,
+      "loss": 0.769,
+      "step": 5097
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.687580025608195e-05,
+      "loss": 0.791,
+      "step": 5098
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6843790012804096e-05,
+      "loss": 0.6481,
+      "step": 5099
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.681177976952625e-05,
+      "loss": 0.6299,
+      "step": 5100
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.67797695262484e-05,
+      "loss": 0.6473,
+      "step": 5101
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.674775928297055e-05,
+      "loss": 0.6989,
+      "step": 5102
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.67157490396927e-05,
+      "loss": 0.6066,
+      "step": 5103
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6683738796414854e-05,
+      "loss": 0.5847,
+      "step": 5104
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.665172855313701e-05,
+      "loss": 0.665,
+      "step": 5105
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.661971830985916e-05,
+      "loss": 0.7021,
+      "step": 5106
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.658770806658131e-05,
+      "loss": 0.7169,
+      "step": 5107
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.655569782330346e-05,
+      "loss": 0.6456,
+      "step": 5108
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.652368758002561e-05,
+      "loss": 0.7031,
+      "step": 5109
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.649167733674776e-05,
+      "loss": 0.6613,
+      "step": 5110
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.645966709346991e-05,
+      "loss": 0.6408,
+      "step": 5111
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6427656850192065e-05,
+      "loss": 0.5417,
+      "step": 5112
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.639564660691421e-05,
+      "loss": 0.6205,
+      "step": 5113
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.6498,
+      "step": 5114
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.633162612035852e-05,
+      "loss": 0.6895,
+      "step": 5115
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.629961587708067e-05,
+      "loss": 0.5582,
+      "step": 5116
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.626760563380282e-05,
+      "loss": 0.6216,
+      "step": 5117
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.623559539052497e-05,
+      "loss": 0.6735,
+      "step": 5118
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.620358514724712e-05,
+      "loss": 0.6704,
+      "step": 5119
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6171574903969275e-05,
+      "loss": 0.575,
+      "step": 5120
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.613956466069142e-05,
+      "loss": 0.6438,
+      "step": 5121
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6107554417413575e-05,
+      "loss": 0.6656,
+      "step": 5122
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.607554417413572e-05,
+      "loss": 0.5794,
+      "step": 5123
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6043533930857874e-05,
+      "loss": 0.7146,
+      "step": 5124
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.601152368758003e-05,
+      "loss": 0.5767,
+      "step": 5125
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.597951344430217e-05,
+      "loss": 0.644,
+      "step": 5126
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.594750320102433e-05,
+      "loss": 0.6613,
+      "step": 5127
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5915492957746486e-05,
+      "loss": 0.714,
+      "step": 5128
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.588348271446863e-05,
+      "loss": 0.7503,
+      "step": 5129
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5851472471190785e-05,
+      "loss": 0.6493,
+      "step": 5130
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.581946222791294e-05,
+      "loss": 0.748,
+      "step": 5131
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5787451984635084e-05,
+      "loss": 0.7554,
+      "step": 5132
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.575544174135724e-05,
+      "loss": 0.6935,
+      "step": 5133
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5723431498079384e-05,
+      "loss": 0.7088,
+      "step": 5134
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.569142125480154e-05,
+      "loss": 0.6163,
+      "step": 5135
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.565941101152369e-05,
+      "loss": 0.7323,
+      "step": 5136
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5627400768245836e-05,
+      "loss": 0.7044,
+      "step": 5137
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.559539052496799e-05,
+      "loss": 0.6816,
+      "step": 5138
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.556338028169014e-05,
+      "loss": 0.7235,
+      "step": 5139
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5531370038412295e-05,
+      "loss": 0.6668,
+      "step": 5140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.549935979513445e-05,
+      "loss": 0.7283,
+      "step": 5141
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5467349551856594e-05,
+      "loss": 0.6512,
+      "step": 5142
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.543533930857875e-05,
+      "loss": 0.6886,
+      "step": 5143
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.54033290653009e-05,
+      "loss": 0.5714,
+      "step": 5144
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5371318822023046e-05,
+      "loss": 0.7887,
+      "step": 5145
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.53393085787452e-05,
+      "loss": 0.7931,
+      "step": 5146
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.530729833546735e-05,
+      "loss": 0.7294,
+      "step": 5147
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.52752880921895e-05,
+      "loss": 0.6986,
+      "step": 5148
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.524327784891165e-05,
+      "loss": 0.6901,
+      "step": 5149
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5211267605633805e-05,
+      "loss": 0.6977,
+      "step": 5150
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.517925736235596e-05,
+      "loss": 0.5632,
+      "step": 5151
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.514724711907811e-05,
+      "loss": 0.6429,
+      "step": 5152
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.511523687580026e-05,
+      "loss": 0.6129,
+      "step": 5153
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.508322663252241e-05,
+      "loss": 0.5522,
+      "step": 5154
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.505121638924456e-05,
+      "loss": 0.6157,
+      "step": 5155
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.501920614596671e-05,
+      "loss": 0.6209,
+      "step": 5156
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.498719590268886e-05,
+      "loss": 0.5335,
+      "step": 5157
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.495518565941101e-05,
+      "loss": 0.842,
+      "step": 5158
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.492317541613316e-05,
+      "loss": 0.606,
+      "step": 5159
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4891165172855314e-05,
+      "loss": 0.645,
+      "step": 5160
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.485915492957747e-05,
+      "loss": 0.754,
+      "step": 5161
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.482714468629962e-05,
+      "loss": 0.6491,
+      "step": 5162
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4795134443021773e-05,
+      "loss": 0.712,
+      "step": 5163
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.476312419974392e-05,
+      "loss": 0.6288,
+      "step": 5164
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.473111395646607e-05,
+      "loss": 0.6678,
+      "step": 5165
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4699103713188226e-05,
+      "loss": 0.692,
+      "step": 5166
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.466709346991037e-05,
+      "loss": 0.7684,
+      "step": 5167
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4635083226632525e-05,
+      "loss": 0.6433,
+      "step": 5168
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.460307298335467e-05,
+      "loss": 0.6717,
+      "step": 5169
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4571062740076824e-05,
+      "loss": 0.6177,
+      "step": 5170
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.453905249679898e-05,
+      "loss": 0.5983,
+      "step": 5171
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.450704225352113e-05,
+      "loss": 0.6657,
+      "step": 5172
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.447503201024328e-05,
+      "loss": 0.6766,
+      "step": 5173
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4443021766965436e-05,
+      "loss": 0.7164,
+      "step": 5174
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.441101152368758e-05,
+      "loss": 0.69,
+      "step": 5175
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4379001280409735e-05,
+      "loss": 0.6156,
+      "step": 5176
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.434699103713188e-05,
+      "loss": 0.737,
+      "step": 5177
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4314980793854035e-05,
+      "loss": 0.6852,
+      "step": 5178
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.428297055057619e-05,
+      "loss": 0.673,
+      "step": 5179
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4250960307298334e-05,
+      "loss": 0.8903,
+      "step": 5180
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.421895006402049e-05,
+      "loss": 0.5664,
+      "step": 5181
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.418693982074264e-05,
+      "loss": 0.6454,
+      "step": 5182
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4154929577464786e-05,
+      "loss": 0.5052,
+      "step": 5183
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.412291933418694e-05,
+      "loss": 0.6183,
+      "step": 5184
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.409090909090909e-05,
+      "loss": 0.7045,
+      "step": 5185
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4058898847631245e-05,
+      "loss": 0.6927,
+      "step": 5186
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.40268886043534e-05,
+      "loss": 0.6974,
+      "step": 5187
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3994878361075544e-05,
+      "loss": 0.6301,
+      "step": 5188
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.39628681177977e-05,
+      "loss": 0.698,
+      "step": 5189
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.393085787451985e-05,
+      "loss": 0.6259,
+      "step": 5190
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3898847631242e-05,
+      "loss": 0.6823,
+      "step": 5191
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.386683738796415e-05,
+      "loss": 0.656,
+      "step": 5192
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3834827144686296e-05,
+      "loss": 0.7861,
+      "step": 5193
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.380281690140845e-05,
+      "loss": 0.544,
+      "step": 5194
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.37708066581306e-05,
+      "loss": 0.586,
+      "step": 5195
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3738796414852755e-05,
+      "loss": 0.724,
+      "step": 5196
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.370678617157491e-05,
+      "loss": 0.7102,
+      "step": 5197
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.367477592829706e-05,
+      "loss": 0.7212,
+      "step": 5198
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.364276568501921e-05,
+      "loss": 0.6213,
+      "step": 5199
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.361075544174136e-05,
+      "loss": 0.6506,
+      "step": 5200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.357874519846351e-05,
+      "loss": 0.6582,
+      "step": 5201
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.354673495518566e-05,
+      "loss": 0.6426,
+      "step": 5202
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.351472471190781e-05,
+      "loss": 0.6255,
+      "step": 5203
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.348271446862996e-05,
+      "loss": 0.6057,
+      "step": 5204
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.345070422535211e-05,
+      "loss": 0.6264,
+      "step": 5205
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3418693982074265e-05,
+      "loss": 0.688,
+      "step": 5206
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.338668373879642e-05,
+      "loss": 0.6414,
+      "step": 5207
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.335467349551857e-05,
+      "loss": 0.7045,
+      "step": 5208
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3322663252240724e-05,
+      "loss": 0.6398,
+      "step": 5209
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.329065300896287e-05,
+      "loss": 0.6006,
+      "step": 5210
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.325864276568502e-05,
+      "loss": 0.801,
+      "step": 5211
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.322663252240717e-05,
+      "loss": 0.6635,
+      "step": 5212
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.319462227912932e-05,
+      "loss": 0.6328,
+      "step": 5213
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3162612035851475e-05,
+      "loss": 0.6821,
+      "step": 5214
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.313060179257362e-05,
+      "loss": 0.598,
+      "step": 5215
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3098591549295775e-05,
+      "loss": 0.6764,
+      "step": 5216
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.306658130601793e-05,
+      "loss": 0.6303,
+      "step": 5217
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.303457106274008e-05,
+      "loss": 0.72,
+      "step": 5218
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.3002560819462234e-05,
+      "loss": 0.6282,
+      "step": 5219
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.297055057618438e-05,
+      "loss": 0.7607,
+      "step": 5220
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.293854033290653e-05,
+      "loss": 0.559,
+      "step": 5221
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2906530089628686e-05,
+      "loss": 0.7558,
+      "step": 5222
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.287451984635083e-05,
+      "loss": 0.6373,
+      "step": 5223
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2842509603072985e-05,
+      "loss": 0.6427,
+      "step": 5224
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.281049935979514e-05,
+      "loss": 0.7338,
+      "step": 5225
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2778489116517284e-05,
+      "loss": 0.7189,
+      "step": 5226
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.274647887323944e-05,
+      "loss": 0.6977,
+      "step": 5227
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2714468629961584e-05,
+      "loss": 0.739,
+      "step": 5228
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2682458386683737e-05,
+      "loss": 0.6425,
+      "step": 5229
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.265044814340589e-05,
+      "loss": 0.6482,
+      "step": 5230
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.261843790012804e-05,
+      "loss": 0.6759,
+      "step": 5231
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2586427656850196e-05,
+      "loss": 0.6177,
+      "step": 5232
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.255441741357235e-05,
+      "loss": 0.7699,
+      "step": 5233
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2522407170294495e-05,
+      "loss": 0.5653,
+      "step": 5234
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.249039692701665e-05,
+      "loss": 0.725,
+      "step": 5235
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.24583866837388e-05,
+      "loss": 0.6683,
+      "step": 5236
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.242637644046095e-05,
+      "loss": 0.6399,
+      "step": 5237
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.23943661971831e-05,
+      "loss": 0.6915,
+      "step": 5238
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2362355953905246e-05,
+      "loss": 0.6994,
+      "step": 5239
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.23303457106274e-05,
+      "loss": 0.6228,
+      "step": 5240
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.229833546734955e-05,
+      "loss": 0.668,
+      "step": 5241
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2266325224071705e-05,
+      "loss": 0.6997,
+      "step": 5242
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.223431498079386e-05,
+      "loss": 0.6011,
+      "step": 5243
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.220230473751601e-05,
+      "loss": 0.7141,
+      "step": 5244
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.217029449423816e-05,
+      "loss": 0.6852,
+      "step": 5245
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.213828425096031e-05,
+      "loss": 0.7747,
+      "step": 5246
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.210627400768246e-05,
+      "loss": 0.672,
+      "step": 5247
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.207426376440461e-05,
+      "loss": 0.6178,
+      "step": 5248
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.204225352112676e-05,
+      "loss": 0.6686,
+      "step": 5249
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.201024327784891e-05,
+      "loss": 0.7281,
+      "step": 5250
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.197823303457106e-05,
+      "loss": 0.5741,
+      "step": 5251
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1946222791293215e-05,
+      "loss": 0.5512,
+      "step": 5252
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.191421254801537e-05,
+      "loss": 0.7622,
+      "step": 5253
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.188220230473752e-05,
+      "loss": 0.6493,
+      "step": 5254
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.185019206145967e-05,
+      "loss": 0.605,
+      "step": 5255
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.181818181818182e-05,
+      "loss": 0.6595,
+      "step": 5256
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.178617157490397e-05,
+      "loss": 0.6204,
+      "step": 5257
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.175416133162612e-05,
+      "loss": 0.623,
+      "step": 5258
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.172215108834827e-05,
+      "loss": 0.5982,
+      "step": 5259
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1690140845070426e-05,
+      "loss": 0.6751,
+      "step": 5260
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.165813060179257e-05,
+      "loss": 0.5153,
+      "step": 5261
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1626120358514725e-05,
+      "loss": 0.63,
+      "step": 5262
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.159411011523688e-05,
+      "loss": 0.65,
+      "step": 5263
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.156209987195903e-05,
+      "loss": 0.6807,
+      "step": 5264
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1530089628681184e-05,
+      "loss": 0.6238,
+      "step": 5265
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.149807938540333e-05,
+      "loss": 0.6202,
+      "step": 5266
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.146606914212548e-05,
+      "loss": 0.6171,
+      "step": 5267
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1434058898847636e-05,
+      "loss": 0.6467,
+      "step": 5268
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.140204865556978e-05,
+      "loss": 0.6194,
+      "step": 5269
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1370038412291935e-05,
+      "loss": 0.7074,
+      "step": 5270
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.133802816901409e-05,
+      "loss": 0.7263,
+      "step": 5271
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1306017925736235e-05,
+      "loss": 0.6102,
+      "step": 5272
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.127400768245839e-05,
+      "loss": 0.7727,
+      "step": 5273
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1241997439180534e-05,
+      "loss": 0.6285,
+      "step": 5274
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.120998719590269e-05,
+      "loss": 0.7517,
+      "step": 5275
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.117797695262485e-05,
+      "loss": 0.7618,
+      "step": 5276
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.114596670934699e-05,
+      "loss": 0.5989,
+      "step": 5277
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1113956466069146e-05,
+      "loss": 0.6422,
+      "step": 5278
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.10819462227913e-05,
+      "loss": 0.6629,
+      "step": 5279
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1049935979513445e-05,
+      "loss": 0.6521,
+      "step": 5280
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.10179257362356e-05,
+      "loss": 0.7657,
+      "step": 5281
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0985915492957744e-05,
+      "loss": 0.5988,
+      "step": 5282
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.09539052496799e-05,
+      "loss": 0.753,
+      "step": 5283
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.092189500640205e-05,
+      "loss": 0.655,
+      "step": 5284
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0889884763124197e-05,
+      "loss": 0.5489,
+      "step": 5285
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.085787451984635e-05,
+      "loss": 0.7285,
+      "step": 5286
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.08258642765685e-05,
+      "loss": 0.6484,
+      "step": 5287
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0793854033290656e-05,
+      "loss": 0.5359,
+      "step": 5288
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.076184379001281e-05,
+      "loss": 0.6728,
+      "step": 5289
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0729833546734955e-05,
+      "loss": 0.6176,
+      "step": 5290
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.069782330345711e-05,
+      "loss": 0.5841,
+      "step": 5291
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.066581306017926e-05,
+      "loss": 0.5999,
+      "step": 5292
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.063380281690141e-05,
+      "loss": 0.5406,
+      "step": 5293
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.060179257362356e-05,
+      "loss": 0.6608,
+      "step": 5294
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.056978233034571e-05,
+      "loss": 0.6487,
+      "step": 5295
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.053777208706786e-05,
+      "loss": 0.5892,
+      "step": 5296
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0505761843790016e-05,
+      "loss": 0.6522,
+      "step": 5297
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0473751600512162e-05,
+      "loss": 0.6607,
+      "step": 5298
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0441741357234315e-05,
+      "loss": 0.6999,
+      "step": 5299
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0409731113956468e-05,
+      "loss": 0.6472,
+      "step": 5300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0377720870678618e-05,
+      "loss": 0.6596,
+      "step": 5301
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.034571062740077e-05,
+      "loss": 0.5675,
+      "step": 5302
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0313700384122924e-05,
+      "loss": 0.5999,
+      "step": 5303
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.028169014084507e-05,
+      "loss": 0.7481,
+      "step": 5304
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0249679897567223e-05,
+      "loss": 0.7533,
+      "step": 5305
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0217669654289376e-05,
+      "loss": 0.5903,
+      "step": 5306
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0185659411011526e-05,
+      "loss": 0.6855,
+      "step": 5307
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.015364916773368e-05,
+      "loss": 0.7491,
+      "step": 5308
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0121638924455825e-05,
+      "loss": 0.6039,
+      "step": 5309
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0089628681177978e-05,
+      "loss": 0.6281,
+      "step": 5310
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.005761843790013e-05,
+      "loss": 0.6453,
+      "step": 5311
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.002560819462228e-05,
+      "loss": 0.6134,
+      "step": 5312
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9993597951344433e-05,
+      "loss": 0.647,
+      "step": 5313
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9961587708066586e-05,
+      "loss": 0.7176,
+      "step": 5314
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9929577464788733e-05,
+      "loss": 0.5648,
+      "step": 5315
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9897567221510886e-05,
+      "loss": 0.688,
+      "step": 5316
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9865556978233032e-05,
+      "loss": 0.6736,
+      "step": 5317
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9833546734955188e-05,
+      "loss": 0.7056,
+      "step": 5318
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.980153649167734e-05,
+      "loss": 0.7305,
+      "step": 5319
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9769526248399488e-05,
+      "loss": 0.5886,
+      "step": 5320
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.973751600512164e-05,
+      "loss": 0.6955,
+      "step": 5321
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9705505761843794e-05,
+      "loss": 0.6499,
+      "step": 5322
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.967349551856594e-05,
+      "loss": 0.6834,
+      "step": 5323
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9641485275288096e-05,
+      "loss": 0.6474,
+      "step": 5324
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9609475032010242e-05,
+      "loss": 0.5837,
+      "step": 5325
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9577464788732395e-05,
+      "loss": 0.6491,
+      "step": 5326
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.954545454545455e-05,
+      "loss": 0.621,
+      "step": 5327
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9513444302176695e-05,
+      "loss": 0.7181,
+      "step": 5328
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9481434058898848e-05,
+      "loss": 0.6588,
+      "step": 5329
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9449423815621e-05,
+      "loss": 0.7559,
+      "step": 5330
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.941741357234315e-05,
+      "loss": 0.7544,
+      "step": 5331
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9385403329065303e-05,
+      "loss": 0.5908,
+      "step": 5332
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.935339308578745e-05,
+      "loss": 0.7422,
+      "step": 5333
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9321382842509603e-05,
+      "loss": 0.6433,
+      "step": 5334
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9289372599231756e-05,
+      "loss": 0.6804,
+      "step": 5335
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9257362355953905e-05,
+      "loss": 0.7066,
+      "step": 5336
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9225352112676058e-05,
+      "loss": 0.6117,
+      "step": 5337
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.919334186939821e-05,
+      "loss": 0.7083,
+      "step": 5338
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9161331626120357e-05,
+      "loss": 0.6518,
+      "step": 5339
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.912932138284251e-05,
+      "loss": 0.5896,
+      "step": 5340
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9097311139564663e-05,
+      "loss": 0.613,
+      "step": 5341
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9065300896286813e-05,
+      "loss": 0.7438,
+      "step": 5342
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9033290653008966e-05,
+      "loss": 0.6183,
+      "step": 5343
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.9001280409731112e-05,
+      "loss": 0.6559,
+      "step": 5344
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8969270166453265e-05,
+      "loss": 0.5662,
+      "step": 5345
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8937259923175418e-05,
+      "loss": 0.6956,
+      "step": 5346
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8905249679897568e-05,
+      "loss": 0.6096,
+      "step": 5347
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.887323943661972e-05,
+      "loss": 0.6928,
+      "step": 5348
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8841229193341874e-05,
+      "loss": 0.526,
+      "step": 5349
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.880921895006402e-05,
+      "loss": 0.6964,
+      "step": 5350
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8777208706786173e-05,
+      "loss": 0.7463,
+      "step": 5351
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8745198463508323e-05,
+      "loss": 0.7229,
+      "step": 5352
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8713188220230476e-05,
+      "loss": 0.7143,
+      "step": 5353
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.868117797695263e-05,
+      "loss": 0.6582,
+      "step": 5354
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8649167733674775e-05,
+      "loss": 0.6198,
+      "step": 5355
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8617157490396928e-05,
+      "loss": 0.5727,
+      "step": 5356
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.858514724711908e-05,
+      "loss": 0.6859,
+      "step": 5357
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855313700384123e-05,
+      "loss": 0.6665,
+      "step": 5358
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8521126760563384e-05,
+      "loss": 0.5731,
+      "step": 5359
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.848911651728553e-05,
+      "loss": 0.7578,
+      "step": 5360
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8457106274007683e-05,
+      "loss": 0.6387,
+      "step": 5361
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8425096030729836e-05,
+      "loss": 0.6575,
+      "step": 5362
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8393085787451986e-05,
+      "loss": 0.621,
+      "step": 5363
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.836107554417414e-05,
+      "loss": 0.7882,
+      "step": 5364
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.832906530089629e-05,
+      "loss": 0.7239,
+      "step": 5365
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8297055057618438e-05,
+      "loss": 0.5955,
+      "step": 5366
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.826504481434059e-05,
+      "loss": 0.8315,
+      "step": 5367
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8233034571062744e-05,
+      "loss": 0.7171,
+      "step": 5368
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8201024327784893e-05,
+      "loss": 0.603,
+      "step": 5369
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8169014084507046e-05,
+      "loss": 0.6795,
+      "step": 5370
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8137003841229193e-05,
+      "loss": 0.5604,
+      "step": 5371
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8104993597951346e-05,
+      "loss": 0.6585,
+      "step": 5372
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.80729833546735e-05,
+      "loss": 0.6368,
+      "step": 5373
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8040973111395645e-05,
+      "loss": 0.6519,
+      "step": 5374
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8008962868117798e-05,
+      "loss": 0.7036,
+      "step": 5375
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7976952624839954e-05,
+      "loss": 0.7992,
+      "step": 5376
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.79449423815621e-05,
+      "loss": 0.5695,
+      "step": 5377
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7912932138284254e-05,
+      "loss": 0.7711,
+      "step": 5378
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.78809218950064e-05,
+      "loss": 0.6557,
+      "step": 5379
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7848911651728553e-05,
+      "loss": 0.7017,
+      "step": 5380
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7816901408450706e-05,
+      "loss": 0.6468,
+      "step": 5381
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7784891165172855e-05,
+      "loss": 0.6789,
+      "step": 5382
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.775288092189501e-05,
+      "loss": 0.5784,
+      "step": 5383
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.772087067861716e-05,
+      "loss": 0.6075,
+      "step": 5384
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7688860435339308e-05,
+      "loss": 0.6885,
+      "step": 5385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.765685019206146e-05,
+      "loss": 0.6508,
+      "step": 5386
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.762483994878361e-05,
+      "loss": 0.6331,
+      "step": 5387
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7592829705505763e-05,
+      "loss": 0.5895,
+      "step": 5388
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7560819462227916e-05,
+      "loss": 0.7672,
+      "step": 5389
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7528809218950063e-05,
+      "loss": 0.6117,
+      "step": 5390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7496798975672216e-05,
+      "loss": 0.7022,
+      "step": 5391
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.746478873239437e-05,
+      "loss": 0.7279,
+      "step": 5392
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7432778489116518e-05,
+      "loss": 0.7343,
+      "step": 5393
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.740076824583867e-05,
+      "loss": 0.6588,
+      "step": 5394
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7368758002560817e-05,
+      "loss": 0.5883,
+      "step": 5395
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.733674775928297e-05,
+      "loss": 0.6211,
+      "step": 5396
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7304737516005123e-05,
+      "loss": 0.6284,
+      "step": 5397
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.6185,
+      "step": 5398
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7240717029449426e-05,
+      "loss": 0.712,
+      "step": 5399
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.720870678617158e-05,
+      "loss": 0.7412,
+      "step": 5400
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7176696542893725e-05,
+      "loss": 0.612,
+      "step": 5401
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.714468629961588e-05,
+      "loss": 0.5962,
+      "step": 5402
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.711267605633803e-05,
+      "loss": 0.562,
+      "step": 5403
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.708066581306018e-05,
+      "loss": 0.7743,
+      "step": 5404
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7048655569782334e-05,
+      "loss": 0.5637,
+      "step": 5405
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.701664532650448e-05,
+      "loss": 0.6334,
+      "step": 5406
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6984635083226633e-05,
+      "loss": 0.7589,
+      "step": 5407
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6952624839948786e-05,
+      "loss": 0.6662,
+      "step": 5408
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6920614596670936e-05,
+      "loss": 0.5661,
+      "step": 5409
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.688860435339309e-05,
+      "loss": 0.5992,
+      "step": 5410
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6856594110115242e-05,
+      "loss": 0.5246,
+      "step": 5411
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6824583866837388e-05,
+      "loss": 0.702,
+      "step": 5412
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.679257362355954e-05,
+      "loss": 0.7051,
+      "step": 5413
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.676056338028169e-05,
+      "loss": 0.6433,
+      "step": 5414
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6728553137003844e-05,
+      "loss": 0.7156,
+      "step": 5415
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6696542893725997e-05,
+      "loss": 0.6018,
+      "step": 5416
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6664532650448143e-05,
+      "loss": 0.6194,
+      "step": 5417
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6632522407170296e-05,
+      "loss": 0.7008,
+      "step": 5418
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.660051216389245e-05,
+      "loss": 0.6147,
+      "step": 5419
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6568501920614595e-05,
+      "loss": 0.6932,
+      "step": 5420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.653649167733675e-05,
+      "loss": 0.6572,
+      "step": 5421
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6504481434058898e-05,
+      "loss": 0.6893,
+      "step": 5422
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.647247119078105e-05,
+      "loss": 0.5799,
+      "step": 5423
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6440460947503204e-05,
+      "loss": 0.7251,
+      "step": 5424
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.640845070422535e-05,
+      "loss": 0.6485,
+      "step": 5425
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6376440460947503e-05,
+      "loss": 0.7895,
+      "step": 5426
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6344430217669656e-05,
+      "loss": 0.7611,
+      "step": 5427
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6312419974391806e-05,
+      "loss": 0.5717,
+      "step": 5428
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.628040973111396e-05,
+      "loss": 0.6753,
+      "step": 5429
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6248399487836105e-05,
+      "loss": 0.651,
+      "step": 5430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6216389244558258e-05,
+      "loss": 0.758,
+      "step": 5431
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.618437900128041e-05,
+      "loss": 0.6365,
+      "step": 5432
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.615236875800256e-05,
+      "loss": 0.6632,
+      "step": 5433
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6120358514724714e-05,
+      "loss": 0.7146,
+      "step": 5434
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6088348271446867e-05,
+      "loss": 0.7343,
+      "step": 5435
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6056338028169013e-05,
+      "loss": 0.6725,
+      "step": 5436
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6024327784891166e-05,
+      "loss": 0.6568,
+      "step": 5437
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.599231754161332e-05,
+      "loss": 0.6921,
+      "step": 5438
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.596030729833547e-05,
+      "loss": 0.6488,
+      "step": 5439
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.592829705505762e-05,
+      "loss": 0.6519,
+      "step": 5440
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5896286811779768e-05,
+      "loss": 0.5882,
+      "step": 5441
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.586427656850192e-05,
+      "loss": 0.6288,
+      "step": 5442
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5832266325224074e-05,
+      "loss": 0.6036,
+      "step": 5443
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5800256081946223e-05,
+      "loss": 0.7171,
+      "step": 5444
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5768245838668376e-05,
+      "loss": 0.6476,
+      "step": 5445
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.573623559539053e-05,
+      "loss": 0.6696,
+      "step": 5446
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5704225352112676e-05,
+      "loss": 0.6514,
+      "step": 5447
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.567221510883483e-05,
+      "loss": 0.6642,
+      "step": 5448
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5640204865556978e-05,
+      "loss": 0.7138,
+      "step": 5449
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.560819462227913e-05,
+      "loss": 0.7086,
+      "step": 5450
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5576184379001284e-05,
+      "loss": 0.6912,
+      "step": 5451
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.554417413572343e-05,
+      "loss": 0.612,
+      "step": 5452
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5512163892445584e-05,
+      "loss": 0.6721,
+      "step": 5453
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5480153649167737e-05,
+      "loss": 0.6158,
+      "step": 5454
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5448143405889886e-05,
+      "loss": 0.6948,
+      "step": 5455
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.541613316261204e-05,
+      "loss": 0.5848,
+      "step": 5456
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5384122919334185e-05,
+      "loss": 0.6856,
+      "step": 5457
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.535211267605634e-05,
+      "loss": 0.6318,
+      "step": 5458
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.532010243277849e-05,
+      "loss": 0.649,
+      "step": 5459
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.528809218950064e-05,
+      "loss": 0.7003,
+      "step": 5460
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5256081946222794e-05,
+      "loss": 0.685,
+      "step": 5461
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5224071702944947e-05,
+      "loss": 0.6378,
+      "step": 5462
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5192061459667093e-05,
+      "loss": 0.7063,
+      "step": 5463
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5160051216389246e-05,
+      "loss": 0.6583,
+      "step": 5464
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5128040973111393e-05,
+      "loss": 0.6343,
+      "step": 5465
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.509603072983355e-05,
+      "loss": 0.6641,
+      "step": 5466
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5064020486555702e-05,
+      "loss": 0.7507,
+      "step": 5467
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5032010243277848e-05,
+      "loss": 0.5453,
+      "step": 5468
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6481,
+      "step": 5469
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.496798975672215e-05,
+      "loss": 0.6354,
+      "step": 5470
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4935979513444304e-05,
+      "loss": 0.6671,
+      "step": 5471
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4903969270166453e-05,
+      "loss": 0.7768,
+      "step": 5472
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4871959026888606e-05,
+      "loss": 0.542,
+      "step": 5473
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4839948783610756e-05,
+      "loss": 0.6329,
+      "step": 5474
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.480793854033291e-05,
+      "loss": 0.6544,
+      "step": 5475
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.477592829705506e-05,
+      "loss": 0.6491,
+      "step": 5476
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.474391805377721e-05,
+      "loss": 0.8764,
+      "step": 5477
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.471190781049936e-05,
+      "loss": 0.7096,
+      "step": 5478
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4679897567221514e-05,
+      "loss": 0.588,
+      "step": 5479
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4647887323943664e-05,
+      "loss": 0.6934,
+      "step": 5480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4615877080665814e-05,
+      "loss": 0.6525,
+      "step": 5481
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4583866837387963e-05,
+      "loss": 0.6507,
+      "step": 5482
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4551856594110116e-05,
+      "loss": 0.6907,
+      "step": 5483
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.451984635083227e-05,
+      "loss": 0.7025,
+      "step": 5484
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.448783610755442e-05,
+      "loss": 0.5338,
+      "step": 5485
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.445582586427657e-05,
+      "loss": 0.7106,
+      "step": 5486
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.442381562099872e-05,
+      "loss": 0.6269,
+      "step": 5487
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.439180537772087e-05,
+      "loss": 0.6741,
+      "step": 5488
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4359795134443024e-05,
+      "loss": 0.7562,
+      "step": 5489
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4327784891165174e-05,
+      "loss": 0.7614,
+      "step": 5490
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4295774647887327e-05,
+      "loss": 0.6973,
+      "step": 5491
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4263764404609476e-05,
+      "loss": 0.6533,
+      "step": 5492
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4231754161331626e-05,
+      "loss": 0.7386,
+      "step": 5493
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4199743918053776e-05,
+      "loss": 0.6729,
+      "step": 5494
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4167733674775932e-05,
+      "loss": 0.5701,
+      "step": 5495
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.413572343149808e-05,
+      "loss": 0.6856,
+      "step": 5496
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.410371318822023e-05,
+      "loss": 0.6357,
+      "step": 5497
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.407170294494238e-05,
+      "loss": 0.7307,
+      "step": 5498
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4039692701664534e-05,
+      "loss": 0.6799,
+      "step": 5499
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4007682458386683e-05,
+      "loss": 0.5935,
+      "step": 5500
+    }
+  ],
+  "max_steps": 6250,
+  "num_train_epochs": 5,
+  "total_flos": 2.156691989232599e+17,
+  "trial_name": null,
+  "trial_params": null
+}