diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,78033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999230828397816,
+  "eval_steps": 500,
+  "global_step": 13000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 7.691716021844474e-05,
+      "learning_rate": 0.0029999999562067186,
+      "loss": 1.9092,
+      "step": 1
+    },
+    {
+      "epoch": 0.00015383432043688947,
+      "learning_rate": 0.0029999998248268762,
+      "loss": 1.9814,
+      "step": 2
+    },
+    {
+      "epoch": 0.0002307514806553342,
+      "learning_rate": 0.002999999605860481,
+      "loss": 1.7023,
+      "step": 3
+    },
+    {
+      "epoch": 0.00030766864087377894,
+      "learning_rate": 0.0029999992993075455,
+      "loss": 1.9262,
+      "step": 4
+    },
+    {
+      "epoch": 0.00038458580109222365,
+      "learning_rate": 0.002999998905168087,
+      "loss": 2.1237,
+      "step": 5
+    },
+    {
+      "epoch": 0.0004615029613106684,
+      "learning_rate": 0.0029999984234421297,
+      "loss": 1.6514,
+      "step": 6
+    },
+    {
+      "epoch": 0.0005384201215291131,
+      "learning_rate": 0.0029999978541297013,
+      "loss": 2.2443,
+      "step": 7
+    },
+    {
+      "epoch": 0.0006153372817475579,
+      "learning_rate": 0.0029999971972308344,
+      "loss": 1.872,
+      "step": 8
+    },
+    {
+      "epoch": 0.0006922544419660026,
+      "learning_rate": 0.0029999964527455685,
+      "loss": 1.8875,
+      "step": 9
+    },
+    {
+      "epoch": 0.0007691716021844473,
+      "learning_rate": 0.002999995620673947,
+      "loss": 2.0612,
+      "step": 10
+    },
+    {
+      "epoch": 0.0008460887624028921,
+      "learning_rate": 0.0029999947010160167,
+      "loss": 1.9884,
+      "step": 11
+    },
+    {
+      "epoch": 0.0009230059226213368,
+      "learning_rate": 0.0029999936937718332,
+      "loss": 2.2815,
+      "step": 12
+    },
+    {
+      "epoch": 0.0009999230828397815,
+      "learning_rate": 0.0029999925989414545,
+      "loss": 1.6098,
+      "step": 13
+    },
+    {
+      "epoch": 0.0010768402430582262,
+      "learning_rate": 0.0029999914165249448,
+      "loss": 1.8513,
+      "step": 14
+    },
+    {
+      "epoch": 0.001153757403276671,
+      "learning_rate": 0.002999990146522373,
+      "loss": 1.9658,
+      "step": 15
+    },
+    {
+      "epoch": 0.0012306745634951158,
+      "learning_rate": 0.002999988788933813,
+      "loss": 1.6379,
+      "step": 16
+    },
+    {
+      "epoch": 0.0013075917237135605,
+      "learning_rate": 0.0029999873437593447,
+      "loss": 1.7758,
+      "step": 17
+    },
+    {
+      "epoch": 0.0013845088839320053,
+      "learning_rate": 0.0029999858109990524,
+      "loss": 1.7943,
+      "step": 18
+    },
+    {
+      "epoch": 0.00146142604415045,
+      "learning_rate": 0.002999984190653025,
+      "loss": 1.3816,
+      "step": 19
+    },
+    {
+      "epoch": 0.0015383432043688946,
+      "learning_rate": 0.0029999824827213572,
+      "loss": 1.7113,
+      "step": 20
+    },
+    {
+      "epoch": 0.0016152603645873394,
+      "learning_rate": 0.0029999806872041497,
+      "loss": 1.6993,
+      "step": 21
+    },
+    {
+      "epoch": 0.0016921775248057841,
+      "learning_rate": 0.0029999788041015062,
+      "loss": 1.3472,
+      "step": 22
+    },
+    {
+      "epoch": 0.0017690946850242289,
+      "learning_rate": 0.002999976833413537,
+      "loss": 1.7283,
+      "step": 23
+    },
+    {
+      "epoch": 0.0018460118452426736,
+      "learning_rate": 0.002999974775140357,
+      "loss": 1.0342,
+      "step": 24
+    },
+    {
+      "epoch": 0.0019229290054611184,
+      "learning_rate": 0.0029999726292820874,
+      "loss": 1.3413,
+      "step": 25
+    },
+    {
+      "epoch": 0.001999846165679563,
+      "learning_rate": 0.0029999703958388524,
+      "loss": 1.3192,
+      "step": 26
+    },
+    {
+      "epoch": 0.002076763325898008,
+      "learning_rate": 0.0029999680748107826,
+      "loss": 1.4627,
+      "step": 27
+    },
+    {
+      "epoch": 0.0021536804861164525,
+      "learning_rate": 0.002999965666198014,
+      "loss": 1.2075,
+      "step": 28
+    },
+    {
+      "epoch": 0.0022305976463348975,
+      "learning_rate": 0.0029999631700006863,
+      "loss": 1.3928,
+      "step": 29
+    },
+    {
+      "epoch": 0.002307514806553342,
+      "learning_rate": 0.0029999605862189465,
+      "loss": 1.3366,
+      "step": 30
+    },
+    {
+      "epoch": 0.002384431966771787,
+      "learning_rate": 0.002999957914852945,
+      "loss": 1.2492,
+      "step": 31
+    },
+    {
+      "epoch": 0.0024613491269902315,
+      "learning_rate": 0.002999955155902837,
+      "loss": 1.5065,
+      "step": 32
+    },
+    {
+      "epoch": 0.002538266287208676,
+      "learning_rate": 0.0029999523093687847,
+      "loss": 1.3048,
+      "step": 33
+    },
+    {
+      "epoch": 0.002615183447427121,
+      "learning_rate": 0.002999949375250953,
+      "loss": 1.209,
+      "step": 34
+    },
+    {
+      "epoch": 0.0026921006076455656,
+      "learning_rate": 0.002999946353549515,
+      "loss": 1.457,
+      "step": 35
+    },
+    {
+      "epoch": 0.0027690177678640106,
+      "learning_rate": 0.002999943244264646,
+      "loss": 1.0777,
+      "step": 36
+    },
+    {
+      "epoch": 0.002845934928082455,
+      "learning_rate": 0.0029999400473965277,
+      "loss": 1.1821,
+      "step": 37
+    },
+    {
+      "epoch": 0.0029228520883009,
+      "learning_rate": 0.0029999367629453466,
+      "loss": 0.8748,
+      "step": 38
+    },
+    {
+      "epoch": 0.0029997692485193446,
+      "learning_rate": 0.0029999333909112953,
+      "loss": 1.3999,
+      "step": 39
+    },
+    {
+      "epoch": 0.003076686408737789,
+      "learning_rate": 0.0029999299312945694,
+      "loss": 1.1431,
+      "step": 40
+    },
+    {
+      "epoch": 0.003153603568956234,
+      "learning_rate": 0.0029999263840953722,
+      "loss": 1.2118,
+      "step": 41
+    },
+    {
+      "epoch": 0.0032305207291746787,
+      "learning_rate": 0.00299992274931391,
+      "loss": 1.1694,
+      "step": 42
+    },
+    {
+      "epoch": 0.0033074378893931237,
+      "learning_rate": 0.0029999190269503952,
+      "loss": 1.5292,
+      "step": 43
+    },
+    {
+      "epoch": 0.0033843550496115682,
+      "learning_rate": 0.0029999152170050456,
+      "loss": 0.9456,
+      "step": 44
+    },
+    {
+      "epoch": 0.0034612722098300132,
+      "learning_rate": 0.002999911319478083,
+      "loss": 1.3324,
+      "step": 45
+    },
+    {
+      "epoch": 0.0035381893700484578,
+      "learning_rate": 0.0029999073343697355,
+      "loss": 1.2328,
+      "step": 46
+    },
+    {
+      "epoch": 0.0036151065302669027,
+      "learning_rate": 0.002999903261680235,
+      "loss": 0.9419,
+      "step": 47
+    },
+    {
+      "epoch": 0.0036920236904853473,
+      "learning_rate": 0.0029998991014098205,
+      "loss": 0.8992,
+      "step": 48
+    },
+    {
+      "epoch": 0.003768940850703792,
+      "learning_rate": 0.0029998948535587345,
+      "loss": 1.1284,
+      "step": 49
+    },
+    {
+      "epoch": 0.003845858010922237,
+      "learning_rate": 0.0029998905181272237,
+      "loss": 1.3362,
+      "step": 50
+    },
+    {
+      "epoch": 0.003922775171140681,
+      "learning_rate": 0.0029998860951155437,
+      "loss": 1.3353,
+      "step": 51
+    },
+    {
+      "epoch": 0.003999692331359126,
+      "learning_rate": 0.0029998815845239508,
+      "loss": 1.1782,
+      "step": 52
+    },
+    {
+      "epoch": 0.004076609491577571,
+      "learning_rate": 0.002999876986352709,
+      "loss": 1.4899,
+      "step": 53
+    },
+    {
+      "epoch": 0.004153526651796016,
+      "learning_rate": 0.002999872300602087,
+      "loss": 1.2823,
+      "step": 54
+    },
+    {
+      "epoch": 0.00423044381201446,
+      "learning_rate": 0.0029998675272723584,
+      "loss": 1.1416,
+      "step": 55
+    },
+    {
+      "epoch": 0.004307360972232905,
+      "learning_rate": 0.0029998626663638017,
+      "loss": 1.4207,
+      "step": 56
+    },
+    {
+      "epoch": 0.0043842781324513495,
+      "learning_rate": 0.0029998577178767007,
+      "loss": 1.2175,
+      "step": 57
+    },
+    {
+      "epoch": 0.004461195292669795,
+      "learning_rate": 0.0029998526818113443,
+      "loss": 1.1006,
+      "step": 58
+    },
+    {
+      "epoch": 0.0045381124528882395,
+      "learning_rate": 0.002999847558168027,
+      "loss": 1.2269,
+      "step": 59
+    },
+    {
+      "epoch": 0.004615029613106684,
+      "learning_rate": 0.002999842346947047,
+      "loss": 1.5156,
+      "step": 60
+    },
+    {
+      "epoch": 0.0046919467733251286,
+      "learning_rate": 0.00299983704814871,
+      "loss": 1.0262,
+      "step": 61
+    },
+    {
+      "epoch": 0.004768863933543574,
+      "learning_rate": 0.0029998316617733246,
+      "loss": 1.5848,
+      "step": 62
+    },
+    {
+      "epoch": 0.0048457810937620185,
+      "learning_rate": 0.002999826187821205,
+      "loss": 1.5708,
+      "step": 63
+    },
+    {
+      "epoch": 0.004922698253980463,
+      "learning_rate": 0.002999820626292671,
+      "loss": 1.1905,
+      "step": 64
+    },
+    {
+      "epoch": 0.004999615414198908,
+      "learning_rate": 0.0029998149771880483,
+      "loss": 1.4874,
+      "step": 65
+    },
+    {
+      "epoch": 0.005076532574417352,
+      "learning_rate": 0.002999809240507666,
+      "loss": 1.3132,
+      "step": 66
+    },
+    {
+      "epoch": 0.0051534497346357976,
+      "learning_rate": 0.0029998034162518586,
+      "loss": 1.2571,
+      "step": 67
+    },
+    {
+      "epoch": 0.005230366894854242,
+      "learning_rate": 0.0029997975044209666,
+      "loss": 1.3234,
+      "step": 68
+    },
+    {
+      "epoch": 0.005307284055072687,
+      "learning_rate": 0.002999791505015336,
+      "loss": 1.2275,
+      "step": 69
+    },
+    {
+      "epoch": 0.005384201215291131,
+      "learning_rate": 0.0029997854180353156,
+      "loss": 1.2061,
+      "step": 70
+    },
+    {
+      "epoch": 0.005461118375509577,
+      "learning_rate": 0.002999779243481262,
+      "loss": 1.1499,
+      "step": 71
+    },
+    {
+      "epoch": 0.005538035535728021,
+      "learning_rate": 0.0029997729813535353,
+      "loss": 1.4395,
+      "step": 72
+    },
+    {
+      "epoch": 0.005614952695946466,
+      "learning_rate": 0.0029997666316525008,
+      "loss": 1.416,
+      "step": 73
+    },
+    {
+      "epoch": 0.00569186985616491,
+      "learning_rate": 0.00299976019437853,
+      "loss": 1.0341,
+      "step": 74
+    },
+    {
+      "epoch": 0.005768787016383355,
+      "learning_rate": 0.002999753669531999,
+      "loss": 1.5975,
+      "step": 75
+    },
+    {
+      "epoch": 0.0058457041766018,
+      "learning_rate": 0.0029997470571132873,
+      "loss": 0.9277,
+      "step": 76
+    },
+    {
+      "epoch": 0.005922621336820245,
+      "learning_rate": 0.002999740357122782,
+      "loss": 0.894,
+      "step": 77
+    },
+    {
+      "epoch": 0.005999538497038689,
+      "learning_rate": 0.0029997335695608747,
+      "loss": 1.4544,
+      "step": 78
+    },
+    {
+      "epoch": 0.006076455657257134,
+      "learning_rate": 0.002999726694427961,
+      "loss": 0.9724,
+      "step": 79
+    },
+    {
+      "epoch": 0.006153372817475578,
+      "learning_rate": 0.0029997197317244433,
+      "loss": 1.4934,
+      "step": 80
+    },
+    {
+      "epoch": 0.006230289977694024,
+      "learning_rate": 0.0029997126814507264,
+      "loss": 1.193,
+      "step": 81
+    },
+    {
+      "epoch": 0.006307207137912468,
+      "learning_rate": 0.0029997055436072236,
+      "loss": 0.8429,
+      "step": 82
+    },
+    {
+      "epoch": 0.006384124298130913,
+      "learning_rate": 0.0029996983181943514,
+      "loss": 1.2726,
+      "step": 83
+    },
+    {
+      "epoch": 0.006461041458349357,
+      "learning_rate": 0.0029996910052125314,
+      "loss": 1.218,
+      "step": 84
+    },
+    {
+      "epoch": 0.006537958618567803,
+      "learning_rate": 0.00299968360466219,
+      "loss": 0.9128,
+      "step": 85
+    },
+    {
+      "epoch": 0.006614875778786247,
+      "learning_rate": 0.002999676116543761,
+      "loss": 1.2736,
+      "step": 86
+    },
+    {
+      "epoch": 0.006691792939004692,
+      "learning_rate": 0.00299966854085768,
+      "loss": 1.0608,
+      "step": 87
+    },
+    {
+      "epoch": 0.0067687100992231365,
+      "learning_rate": 0.00299966087760439,
+      "loss": 1.0455,
+      "step": 88
+    },
+    {
+      "epoch": 0.006845627259441581,
+      "learning_rate": 0.0029996531267843384,
+      "loss": 1.3783,
+      "step": 89
+    },
+    {
+      "epoch": 0.0069225444196600264,
+      "learning_rate": 0.002999645288397978,
+      "loss": 1.0357,
+      "step": 90
+    },
+    {
+      "epoch": 0.006999461579878471,
+      "learning_rate": 0.0029996373624457664,
+      "loss": 1.2053,
+      "step": 91
+    },
+    {
+      "epoch": 0.0070763787400969155,
+      "learning_rate": 0.002999629348928166,
+      "loss": 1.098,
+      "step": 92
+    },
+    {
+      "epoch": 0.00715329590031536,
+      "learning_rate": 0.0029996212478456455,
+      "loss": 1.2888,
+      "step": 93
+    },
+    {
+      "epoch": 0.0072302130605338055,
+      "learning_rate": 0.002999613059198677,
+      "loss": 1.1579,
+      "step": 94
+    },
+    {
+      "epoch": 0.00730713022075225,
+      "learning_rate": 0.002999604782987739,
+      "loss": 1.2183,
+      "step": 95
+    },
+    {
+      "epoch": 0.007384047380970695,
+      "learning_rate": 0.0029995964192133157,
+      "loss": 0.9659,
+      "step": 96
+    },
+    {
+      "epoch": 0.007460964541189139,
+      "learning_rate": 0.002999587967875894,
+      "loss": 0.9437,
+      "step": 97
+    },
+    {
+      "epoch": 0.007537881701407584,
+      "learning_rate": 0.002999579428975968,
+      "loss": 1.0836,
+      "step": 98
+    },
+    {
+      "epoch": 0.007614798861626029,
+      "learning_rate": 0.002999570802514037,
+      "loss": 1.0699,
+      "step": 99
+    },
+    {
+      "epoch": 0.007691716021844474,
+      "learning_rate": 0.002999562088490603,
+      "loss": 0.7358,
+      "step": 100
+    },
+    {
+      "epoch": 0.007768633182062918,
+      "learning_rate": 0.0029995532869061766,
+      "loss": 1.0536,
+      "step": 101
+    },
+    {
+      "epoch": 0.007845550342281363,
+      "learning_rate": 0.002999544397761271,
+      "loss": 0.7006,
+      "step": 102
+    },
+    {
+      "epoch": 0.007922467502499808,
+      "learning_rate": 0.002999535421056405,
+      "loss": 0.8425,
+      "step": 103
+    },
+    {
+      "epoch": 0.007999384662718252,
+      "learning_rate": 0.0029995263567921026,
+      "loss": 1.2017,
+      "step": 104
+    },
+    {
+      "epoch": 0.008076301822936697,
+      "learning_rate": 0.0029995172049688936,
+      "loss": 0.9685,
+      "step": 105
+    },
+    {
+      "epoch": 0.008153218983155143,
+      "learning_rate": 0.0029995079655873126,
+      "loss": 1.0908,
+      "step": 106
+    },
+    {
+      "epoch": 0.008230136143373586,
+      "learning_rate": 0.0029994986386478986,
+      "loss": 0.7359,
+      "step": 107
+    },
+    {
+      "epoch": 0.008307053303592032,
+      "learning_rate": 0.0029994892241511965,
+      "loss": 1.1188,
+      "step": 108
+    },
+    {
+      "epoch": 0.008383970463810475,
+      "learning_rate": 0.0029994797220977555,
+      "loss": 1.5312,
+      "step": 109
+    },
+    {
+      "epoch": 0.00846088762402892,
+      "learning_rate": 0.002999470132488131,
+      "loss": 0.888,
+      "step": 110
+    },
+    {
+      "epoch": 0.008537804784247366,
+      "learning_rate": 0.002999460455322883,
+      "loss": 0.8843,
+      "step": 111
+    },
+    {
+      "epoch": 0.00861472194446581,
+      "learning_rate": 0.0029994506906025763,
+      "loss": 1.1051,
+      "step": 112
+    },
+    {
+      "epoch": 0.008691639104684255,
+      "learning_rate": 0.0029994408383277804,
+      "loss": 1.2141,
+      "step": 113
+    },
+    {
+      "epoch": 0.008768556264902699,
+      "learning_rate": 0.002999430898499072,
+      "loss": 1.2885,
+      "step": 114
+    },
+    {
+      "epoch": 0.008845473425121144,
+      "learning_rate": 0.0029994208711170305,
+      "loss": 1.3617,
+      "step": 115
+    },
+    {
+      "epoch": 0.00892239058533959,
+      "learning_rate": 0.0029994107561822415,
+      "loss": 1.0363,
+      "step": 116
+    },
+    {
+      "epoch": 0.008999307745558034,
+      "learning_rate": 0.002999400553695296,
+      "loss": 1.1184,
+      "step": 117
+    },
+    {
+      "epoch": 0.009076224905776479,
+      "learning_rate": 0.0029993902636567898,
+      "loss": 0.8976,
+      "step": 118
+    },
+    {
+      "epoch": 0.009153142065994924,
+      "learning_rate": 0.002999379886067323,
+      "loss": 1.283,
+      "step": 119
+    },
+    {
+      "epoch": 0.009230059226213368,
+      "learning_rate": 0.0029993694209275025,
+      "loss": 1.3905,
+      "step": 120
+    },
+    {
+      "epoch": 0.009306976386431813,
+      "learning_rate": 0.0029993588682379383,
+      "loss": 1.2637,
+      "step": 121
+    },
+    {
+      "epoch": 0.009383893546650257,
+      "learning_rate": 0.0029993482279992477,
+      "loss": 1.0531,
+      "step": 122
+    },
+    {
+      "epoch": 0.009460810706868703,
+      "learning_rate": 0.002999337500212051,
+      "loss": 1.2387,
+      "step": 123
+    },
+    {
+      "epoch": 0.009537727867087148,
+      "learning_rate": 0.0029993266848769755,
+      "loss": 1.2568,
+      "step": 124
+    },
+    {
+      "epoch": 0.009614645027305592,
+      "learning_rate": 0.0029993157819946524,
+      "loss": 1.0881,
+      "step": 125
+    },
+    {
+      "epoch": 0.009691562187524037,
+      "learning_rate": 0.002999304791565718,
+      "loss": 1.0521,
+      "step": 126
+    },
+    {
+      "epoch": 0.00976847934774248,
+      "learning_rate": 0.002999293713590814,
+      "loss": 1.1498,
+      "step": 127
+    },
+    {
+      "epoch": 0.009845396507960926,
+      "learning_rate": 0.0029992825480705883,
+      "loss": 0.8994,
+      "step": 128
+    },
+    {
+      "epoch": 0.009922313668179372,
+      "learning_rate": 0.0029992712950056917,
+      "loss": 1.3625,
+      "step": 129
+    },
+    {
+      "epoch": 0.009999230828397815,
+      "learning_rate": 0.0029992599543967814,
+      "loss": 1.3445,
+      "step": 130
+    },
+    {
+      "epoch": 0.01007614798861626,
+      "learning_rate": 0.00299924852624452,
+      "loss": 1.3456,
+      "step": 131
+    },
+    {
+      "epoch": 0.010153065148834704,
+      "learning_rate": 0.002999237010549575,
+      "loss": 1.2097,
+      "step": 132
+    },
+    {
+      "epoch": 0.01022998230905315,
+      "learning_rate": 0.0029992254073126184,
+      "loss": 1.2986,
+      "step": 133
+    },
+    {
+      "epoch": 0.010306899469271595,
+      "learning_rate": 0.0029992137165343277,
+      "loss": 1.3062,
+      "step": 134
+    },
+    {
+      "epoch": 0.010383816629490039,
+      "learning_rate": 0.0029992019382153855,
+      "loss": 1.0305,
+      "step": 135
+    },
+    {
+      "epoch": 0.010460733789708484,
+      "learning_rate": 0.00299919007235648,
+      "loss": 1.1904,
+      "step": 136
+    },
+    {
+      "epoch": 0.010537650949926928,
+      "learning_rate": 0.0029991781189583036,
+      "loss": 1.1767,
+      "step": 137
+    },
+    {
+      "epoch": 0.010614568110145373,
+      "learning_rate": 0.0029991660780215542,
+      "loss": 1.3267,
+      "step": 138
+    },
+    {
+      "epoch": 0.010691485270363819,
+      "learning_rate": 0.0029991539495469355,
+      "loss": 1.2611,
+      "step": 139
+    },
+    {
+      "epoch": 0.010768402430582262,
+      "learning_rate": 0.0029991417335351552,
+      "loss": 1.0256,
+      "step": 140
+    },
+    {
+      "epoch": 0.010845319590800708,
+      "learning_rate": 0.002999129429986927,
+      "loss": 1.1199,
+      "step": 141
+    },
+    {
+      "epoch": 0.010922236751019153,
+      "learning_rate": 0.002999117038902968,
+      "loss": 1.138,
+      "step": 142
+    },
+    {
+      "epoch": 0.010999153911237597,
+      "learning_rate": 0.002999104560284004,
+      "loss": 1.232,
+      "step": 143
+    },
+    {
+      "epoch": 0.011076071071456042,
+      "learning_rate": 0.0029990919941307614,
+      "loss": 1.1461,
+      "step": 144
+    },
+    {
+      "epoch": 0.011152988231674486,
+      "learning_rate": 0.0029990793404439755,
+      "loss": 1.1026,
+      "step": 145
+    },
+    {
+      "epoch": 0.011229905391892931,
+      "learning_rate": 0.0029990665992243844,
+      "loss": 0.9884,
+      "step": 146
+    },
+    {
+      "epoch": 0.011306822552111377,
+      "learning_rate": 0.0029990537704727327,
+      "loss": 1.3524,
+      "step": 147
+    },
+    {
+      "epoch": 0.01138373971232982,
+      "learning_rate": 0.0029990408541897685,
+      "loss": 1.2749,
+      "step": 148
+    },
+    {
+      "epoch": 0.011460656872548266,
+      "learning_rate": 0.002999027850376247,
+      "loss": 1.2295,
+      "step": 149
+    },
+    {
+      "epoch": 0.01153757403276671,
+      "learning_rate": 0.0029990147590329267,
+      "loss": 1.1709,
+      "step": 150
+    },
+    {
+      "epoch": 0.011614491192985155,
+      "learning_rate": 0.0029990015801605726,
+      "loss": 0.7045,
+      "step": 151
+    },
+    {
+      "epoch": 0.0116914083532036,
+      "learning_rate": 0.002998988313759954,
+      "loss": 1.0454,
+      "step": 152
+    },
+    {
+      "epoch": 0.011768325513422044,
+      "learning_rate": 0.0029989749598318453,
+      "loss": 1.2119,
+      "step": 153
+    },
+    {
+      "epoch": 0.01184524267364049,
+      "learning_rate": 0.0029989615183770267,
+      "loss": 1.4649,
+      "step": 154
+    },
+    {
+      "epoch": 0.011922159833858933,
+      "learning_rate": 0.002998947989396283,
+      "loss": 1.183,
+      "step": 155
+    },
+    {
+      "epoch": 0.011999076994077379,
+      "learning_rate": 0.002998934372890404,
+      "loss": 1.307,
+      "step": 156
+    },
+    {
+      "epoch": 0.012075994154295824,
+      "learning_rate": 0.0029989206688601847,
+      "loss": 0.9439,
+      "step": 157
+    },
+    {
+      "epoch": 0.012152911314514268,
+      "learning_rate": 0.0029989068773064254,
+      "loss": 1.4585,
+      "step": 158
+    },
+    {
+      "epoch": 0.012229828474732713,
+      "learning_rate": 0.0029988929982299314,
+      "loss": 1.0362,
+      "step": 159
+    },
+    {
+      "epoch": 0.012306745634951157,
+      "learning_rate": 0.0029988790316315132,
+      "loss": 0.9236,
+      "step": 160
+    },
+    {
+      "epoch": 0.012383662795169602,
+      "learning_rate": 0.0029988649775119862,
+      "loss": 1.1296,
+      "step": 161
+    },
+    {
+      "epoch": 0.012460579955388048,
+      "learning_rate": 0.002998850835872171,
+      "loss": 0.7674,
+      "step": 162
+    },
+    {
+      "epoch": 0.012537497115606491,
+      "learning_rate": 0.0029988366067128934,
+      "loss": 1.3065,
+      "step": 163
+    },
+    {
+      "epoch": 0.012614414275824937,
+      "learning_rate": 0.0029988222900349847,
+      "loss": 0.9165,
+      "step": 164
+    },
+    {
+      "epoch": 0.012691331436043382,
+      "learning_rate": 0.00299880788583928,
+      "loss": 1.0541,
+      "step": 165
+    },
+    {
+      "epoch": 0.012768248596261826,
+      "learning_rate": 0.002998793394126621,
+      "loss": 1.2432,
+      "step": 166
+    },
+    {
+      "epoch": 0.012845165756480271,
+      "learning_rate": 0.0029987788148978537,
+      "loss": 1.2695,
+      "step": 167
+    },
+    {
+      "epoch": 0.012922082916698715,
+      "learning_rate": 0.002998764148153829,
+      "loss": 1.6526,
+      "step": 168
+    },
+    {
+      "epoch": 0.01299900007691716,
+      "learning_rate": 0.0029987493938954048,
+      "loss": 1.4513,
+      "step": 169
+    },
+    {
+      "epoch": 0.013075917237135606,
+      "learning_rate": 0.0029987345521234406,
+      "loss": 1.1943,
+      "step": 170
+    },
+    {
+      "epoch": 0.01315283439735405,
+      "learning_rate": 0.002998719622838804,
+      "loss": 1.2793,
+      "step": 171
+    },
+    {
+      "epoch": 0.013229751557572495,
+      "learning_rate": 0.002998704606042367,
+      "loss": 1.2412,
+      "step": 172
+    },
+    {
+      "epoch": 0.013306668717790938,
+      "learning_rate": 0.002998689501735006,
+      "loss": 1.5079,
+      "step": 173
+    },
+    {
+      "epoch": 0.013383585878009384,
+      "learning_rate": 0.0029986743099176035,
+      "loss": 1.124,
+      "step": 174
+    },
+    {
+      "epoch": 0.01346050303822783,
+      "learning_rate": 0.0029986590305910456,
+      "loss": 1.4621,
+      "step": 175
+    },
+    {
+      "epoch": 0.013537420198446273,
+      "learning_rate": 0.0029986436637562254,
+      "loss": 0.8467,
+      "step": 176
+    },
+    {
+      "epoch": 0.013614337358664718,
+      "learning_rate": 0.0029986282094140396,
+      "loss": 1.0726,
+      "step": 177
+    },
+    {
+      "epoch": 0.013691254518883162,
+      "learning_rate": 0.002998612667565391,
+      "loss": 0.8901,
+      "step": 178
+    },
+    {
+      "epoch": 0.013768171679101607,
+      "learning_rate": 0.002998597038211187,
+      "loss": 1.2866,
+      "step": 179
+    },
+    {
+      "epoch": 0.013845088839320053,
+      "learning_rate": 0.0029985813213523394,
+      "loss": 1.2543,
+      "step": 180
+    },
+    {
+      "epoch": 0.013922005999538497,
+      "learning_rate": 0.0029985655169897673,
+      "loss": 0.9437,
+      "step": 181
+    },
+    {
+      "epoch": 0.013998923159756942,
+      "learning_rate": 0.002998549625124393,
+      "loss": 1.0591,
+      "step": 182
+    },
+    {
+      "epoch": 0.014075840319975386,
+      "learning_rate": 0.0029985336457571438,
+      "loss": 1.4138,
+      "step": 183
+    },
+    {
+      "epoch": 0.014152757480193831,
+      "learning_rate": 0.002998517578888954,
+      "loss": 1.6227,
+      "step": 184
+    },
+    {
+      "epoch": 0.014229674640412276,
+      "learning_rate": 0.0029985014245207608,
+      "loss": 1.4723,
+      "step": 185
+    },
+    {
+      "epoch": 0.01430659180063072,
+      "learning_rate": 0.0029984851826535073,
+      "loss": 1.3542,
+      "step": 186
+    },
+    {
+      "epoch": 0.014383508960849166,
+      "learning_rate": 0.0029984688532881426,
+      "loss": 1.76,
+      "step": 187
+    },
+    {
+      "epoch": 0.014460426121067611,
+      "learning_rate": 0.00299845243642562,
+      "loss": 0.858,
+      "step": 188
+    },
+    {
+      "epoch": 0.014537343281286055,
+      "learning_rate": 0.002998435932066898,
+      "loss": 1.4137,
+      "step": 189
+    },
+    {
+      "epoch": 0.0146142604415045,
+      "learning_rate": 0.00299841934021294,
+      "loss": 1.4808,
+      "step": 190
+    },
+    {
+      "epoch": 0.014691177601722944,
+      "learning_rate": 0.0029984026608647157,
+      "loss": 1.1505,
+      "step": 191
+    },
+    {
+      "epoch": 0.01476809476194139,
+      "learning_rate": 0.0029983858940231976,
+      "loss": 1.2201,
+      "step": 192
+    },
+    {
+      "epoch": 0.014845011922159835,
+      "learning_rate": 0.0029983690396893662,
+      "loss": 1.1977,
+      "step": 193
+    },
+    {
+      "epoch": 0.014921929082378278,
+      "learning_rate": 0.0029983520978642047,
+      "loss": 1.2986,
+      "step": 194
+    },
+    {
+      "epoch": 0.014998846242596724,
+      "learning_rate": 0.0029983350685487027,
+      "loss": 0.8975,
+      "step": 195
+    },
+    {
+      "epoch": 0.015075763402815167,
+      "learning_rate": 0.002998317951743854,
+      "loss": 1.3165,
+      "step": 196
+    },
+    {
+      "epoch": 0.015152680563033613,
+      "learning_rate": 0.0029983007474506597,
+      "loss": 1.2725,
+      "step": 197
+    },
+    {
+      "epoch": 0.015229597723252058,
+      "learning_rate": 0.0029982834556701228,
+      "loss": 1.3455,
+      "step": 198
+    },
+    {
+      "epoch": 0.015306514883470502,
+      "learning_rate": 0.0029982660764032534,
+      "loss": 1.0963,
+      "step": 199
+    },
+    {
+      "epoch": 0.015383432043688947,
+      "learning_rate": 0.0029982486096510664,
+      "loss": 1.0346,
+      "step": 200
+    },
+    {
+      "epoch": 0.015460349203907391,
+      "learning_rate": 0.0029982310554145814,
+      "loss": 1.281,
+      "step": 201
+    },
+    {
+      "epoch": 0.015537266364125836,
+      "learning_rate": 0.0029982134136948236,
+      "loss": 1.6145,
+      "step": 202
+    },
+    {
+      "epoch": 0.015614183524344282,
+      "learning_rate": 0.0029981956844928235,
+      "loss": 0.9711,
+      "step": 203
+    },
+    {
+      "epoch": 0.015691100684562725,
+      "learning_rate": 0.0029981778678096157,
+      "loss": 0.9053,
+      "step": 204
+    },
+    {
+      "epoch": 0.01576801784478117,
+      "learning_rate": 0.002998159963646241,
+      "loss": 1.4662,
+      "step": 205
+    },
+    {
+      "epoch": 0.015844935004999616,
+      "learning_rate": 0.002998141972003745,
+      "loss": 1.0082,
+      "step": 206
+    },
+    {
+      "epoch": 0.01592185216521806,
+      "learning_rate": 0.0029981238928831777,
+      "loss": 1.4221,
+      "step": 207
+    },
+    {
+      "epoch": 0.015998769325436504,
+      "learning_rate": 0.002998105726285595,
+      "loss": 0.7058,
+      "step": 208
+    },
+    {
+      "epoch": 0.01607568648565495,
+      "learning_rate": 0.0029980874722120576,
+      "loss": 1.0257,
+      "step": 209
+    },
+    {
+      "epoch": 0.016152603645873394,
+      "learning_rate": 0.0029980691306636313,
+      "loss": 1.1321,
+      "step": 210
+    },
+    {
+      "epoch": 0.01622952080609184,
+      "learning_rate": 0.0029980507016413877,
+      "loss": 1.0979,
+      "step": 211
+    },
+    {
+      "epoch": 0.016306437966310285,
+      "learning_rate": 0.0029980321851464015,
+      "loss": 1.3522,
+      "step": 212
+    },
+    {
+      "epoch": 0.016383355126528727,
+      "learning_rate": 0.0029980135811797557,
+      "loss": 0.8686,
+      "step": 213
+    },
+    {
+      "epoch": 0.016460272286747173,
+      "learning_rate": 0.002997994889742535,
+      "loss": 1.5654,
+      "step": 214
+    },
+    {
+      "epoch": 0.016537189446965618,
+      "learning_rate": 0.002997976110835832,
+      "loss": 0.8115,
+      "step": 215
+    },
+    {
+      "epoch": 0.016614106607184063,
+      "learning_rate": 0.0029979572444607428,
+      "loss": 1.0477,
+      "step": 216
+    },
+    {
+      "epoch": 0.01669102376740251,
+      "learning_rate": 0.0029979382906183687,
+      "loss": 0.9833,
+      "step": 217
+    },
+    {
+      "epoch": 0.01676794092762095,
+      "learning_rate": 0.002997919249309816,
+      "loss": 1.4087,
+      "step": 218
+    },
+    {
+      "epoch": 0.016844858087839396,
+      "learning_rate": 0.0029979001205361986,
+      "loss": 0.9413,
+      "step": 219
+    },
+    {
+      "epoch": 0.01692177524805784,
+      "learning_rate": 0.0029978809042986313,
+      "loss": 1.3196,
+      "step": 220
+    },
+    {
+      "epoch": 0.016998692408276287,
+      "learning_rate": 0.0029978616005982375,
+      "loss": 1.3242,
+      "step": 221
+    },
+    {
+      "epoch": 0.017075609568494732,
+      "learning_rate": 0.0029978422094361435,
+      "loss": 1.1892,
+      "step": 222
+    },
+    {
+      "epoch": 0.017152526728713174,
+      "learning_rate": 0.0029978227308134816,
+      "loss": 1.1711,
+      "step": 223
+    },
+    {
+      "epoch": 0.01722944388893162,
+      "learning_rate": 0.0029978031647313894,
+      "loss": 1.1237,
+      "step": 224
+    },
+    {
+      "epoch": 0.017306361049150065,
+      "learning_rate": 0.0029977835111910096,
+      "loss": 1.2171,
+      "step": 225
+    },
+    {
+      "epoch": 0.01738327820936851,
+      "learning_rate": 0.0029977637701934902,
+      "loss": 1.4833,
+      "step": 226
+    },
+    {
+      "epoch": 0.017460195369586956,
+      "learning_rate": 0.002997743941739982,
+      "loss": 1.383,
+      "step": 227
+    },
+    {
+      "epoch": 0.017537112529805398,
+      "learning_rate": 0.0029977240258316456,
+      "loss": 1.2232,
+      "step": 228
+    },
+    {
+      "epoch": 0.017614029690023843,
+      "learning_rate": 0.0029977040224696415,
+      "loss": 1.2125,
+      "step": 229
+    },
+    {
+      "epoch": 0.01769094685024229,
+      "learning_rate": 0.002997683931655139,
+      "loss": 1.0057,
+      "step": 230
+    },
+    {
+      "epoch": 0.017767864010460734,
+      "learning_rate": 0.0029976637533893112,
+      "loss": 1.2638,
+      "step": 231
+    },
+    {
+      "epoch": 0.01784478117067918,
+      "learning_rate": 0.0029976434876733352,
+      "loss": 1.3071,
+      "step": 232
+    },
+    {
+      "epoch": 0.01792169833089762,
+      "learning_rate": 0.0029976231345083958,
+      "loss": 1.1912,
+      "step": 233
+    },
+    {
+      "epoch": 0.017998615491116067,
+      "learning_rate": 0.0029976026938956808,
+      "loss": 1.2945,
+      "step": 234
+    },
+    {
+      "epoch": 0.018075532651334512,
+      "learning_rate": 0.0029975821658363836,
+      "loss": 1.1906,
+      "step": 235
+    },
+    {
+      "epoch": 0.018152449811552958,
+      "learning_rate": 0.0029975615503317027,
+      "loss": 1.329,
+      "step": 236
+    },
+    {
+      "epoch": 0.018229366971771403,
+      "learning_rate": 0.0029975408473828426,
+      "loss": 1.1752,
+      "step": 237
+    },
+    {
+      "epoch": 0.01830628413198985,
+      "learning_rate": 0.0029975200569910113,
+      "loss": 1.157,
+      "step": 238
+    },
+    {
+      "epoch": 0.01838320129220829,
+      "learning_rate": 0.0029974991791574235,
+      "loss": 1.0557,
+      "step": 239
+    },
+    {
+      "epoch": 0.018460118452426736,
+      "learning_rate": 0.002997478213883298,
+      "loss": 0.9176,
+      "step": 240
+    },
+    {
+      "epoch": 0.01853703561264518,
+      "learning_rate": 0.0029974571611698585,
+      "loss": 1.3192,
+      "step": 241
+    },
+    {
+      "epoch": 0.018613952772863627,
+      "learning_rate": 0.002997436021018335,
+      "loss": 1.2865,
+      "step": 242
+    },
+    {
+      "epoch": 0.018690869933082072,
+      "learning_rate": 0.0029974147934299623,
+      "loss": 1.2836,
+      "step": 243
+    },
+    {
+      "epoch": 0.018767787093300514,
+      "learning_rate": 0.002997393478405978,
+      "loss": 1.161,
+      "step": 244
+    },
+    {
+      "epoch": 0.01884470425351896,
+      "learning_rate": 0.002997372075947629,
+      "loss": 0.9845,
+      "step": 245
+    },
+    {
+      "epoch": 0.018921621413737405,
+      "learning_rate": 0.0029973505860561634,
+      "loss": 1.3582,
+      "step": 246
+    },
+    {
+      "epoch": 0.01899853857395585,
+      "learning_rate": 0.002997329008732837,
+      "loss": 1.3781,
+      "step": 247
+    },
+    {
+      "epoch": 0.019075455734174296,
+      "learning_rate": 0.0029973073439789086,
+      "loss": 1.0758,
+      "step": 248
+    },
+    {
+      "epoch": 0.019152372894392738,
+      "learning_rate": 0.0029972855917956447,
+      "loss": 1.1509,
+      "step": 249
+    },
+    {
+      "epoch": 0.019229290054611183,
+      "learning_rate": 0.0029972637521843144,
+      "loss": 1.2729,
+      "step": 250
+    },
+    {
+      "epoch": 0.01930620721482963,
+      "learning_rate": 0.002997241825146193,
+      "loss": 1.3322,
+      "step": 251
+    },
+    {
+      "epoch": 0.019383124375048074,
+      "learning_rate": 0.0029972198106825616,
+      "loss": 1.2791,
+      "step": 252
+    },
+    {
+      "epoch": 0.01946004153526652,
+      "learning_rate": 0.0029971977087947044,
+      "loss": 1.4703,
+      "step": 253
+    },
+    {
+      "epoch": 0.01953695869548496,
+      "learning_rate": 0.002997175519483913,
+      "loss": 1.2515,
+      "step": 254
+    },
+    {
+      "epoch": 0.019613875855703407,
+      "learning_rate": 0.0029971532427514828,
+      "loss": 1.1393,
+      "step": 255
+    },
+    {
+      "epoch": 0.019690793015921852,
+      "learning_rate": 0.002997130878598714,
+      "loss": 1.3116,
+      "step": 256
+    },
+    {
+      "epoch": 0.019767710176140298,
+      "learning_rate": 0.0029971084270269135,
+      "loss": 1.0154,
+      "step": 257
+    },
+    {
+      "epoch": 0.019844627336358743,
+      "learning_rate": 0.002997085888037391,
+      "loss": 1.0423,
+      "step": 258
+    },
+    {
+      "epoch": 0.019921544496577185,
+      "learning_rate": 0.002997063261631464,
+      "loss": 1.0003,
+      "step": 259
+    },
+    {
+      "epoch": 0.01999846165679563,
+      "learning_rate": 0.0029970405478104522,
+      "loss": 1.2568,
+      "step": 260
+    },
+    {
+      "epoch": 0.020075378817014076,
+      "learning_rate": 0.0029970177465756835,
+      "loss": 1.1609,
+      "step": 261
+    },
+    {
+      "epoch": 0.02015229597723252,
+      "learning_rate": 0.002996994857928488,
+      "loss": 1.55,
+      "step": 262
+    },
+    {
+      "epoch": 0.020229213137450967,
+      "learning_rate": 0.0029969718818702025,
+      "loss": 1.1516,
+      "step": 263
+    },
+    {
+      "epoch": 0.02030613029766941,
+      "learning_rate": 0.0029969488184021687,
+      "loss": 1.3809,
+      "step": 264
+    },
+    {
+      "epoch": 0.020383047457887854,
+      "learning_rate": 0.002996925667525733,
+      "loss": 1.4538,
+      "step": 265
+    },
+    {
+      "epoch": 0.0204599646181063,
+      "learning_rate": 0.0029969024292422486,
+      "loss": 0.8137,
+      "step": 266
+    },
+    {
+      "epoch": 0.020536881778324745,
+      "learning_rate": 0.0029968791035530702,
+      "loss": 1.0519,
+      "step": 267
+    },
+    {
+      "epoch": 0.02061379893854319,
+      "learning_rate": 0.0029968556904595616,
+      "loss": 1.4587,
+      "step": 268
+    },
+    {
+      "epoch": 0.020690716098761632,
+      "learning_rate": 0.0029968321899630887,
+      "loss": 1.2783,
+      "step": 269
+    },
+    {
+      "epoch": 0.020767633258980078,
+      "learning_rate": 0.0029968086020650247,
+      "loss": 1.0779,
+      "step": 270
+    },
+    {
+      "epoch": 0.020844550419198523,
+      "learning_rate": 0.0029967849267667464,
+      "loss": 1.0817,
+      "step": 271
+    },
+    {
+      "epoch": 0.02092146757941697,
+      "learning_rate": 0.0029967611640696365,
+      "loss": 1.3202,
+      "step": 272
+    },
+    {
+      "epoch": 0.020998384739635414,
+      "learning_rate": 0.0029967373139750822,
+      "loss": 1.3649,
+      "step": 273
+    },
+    {
+      "epoch": 0.021075301899853856,
+      "learning_rate": 0.002996713376484476,
+      "loss": 1.101,
+      "step": 274
+    },
+    {
+      "epoch": 0.0211522190600723,
+      "learning_rate": 0.002996689351599216,
+      "loss": 1.0024,
+      "step": 275
+    },
+    {
+      "epoch": 0.021229136220290747,
+      "learning_rate": 0.0029966652393207055,
+      "loss": 1.2042,
+      "step": 276
+    },
+    {
+      "epoch": 0.021306053380509192,
+      "learning_rate": 0.002996641039650351,
+      "loss": 0.9547,
+      "step": 277
+    },
+    {
+      "epoch": 0.021382970540727637,
+      "learning_rate": 0.002996616752589567,
+      "loss": 1.2709,
+      "step": 278
+    },
+    {
+      "epoch": 0.02145988770094608,
+      "learning_rate": 0.0029965923781397714,
+      "loss": 1.166,
+      "step": 279
+    },
+    {
+      "epoch": 0.021536804861164525,
+      "learning_rate": 0.0029965679163023863,
+      "loss": 1.1005,
+      "step": 280
+    },
+    {
+      "epoch": 0.02161372202138297,
+      "learning_rate": 0.0029965433670788417,
+      "loss": 1.1991,
+      "step": 281
+    },
+    {
+      "epoch": 0.021690639181601416,
+      "learning_rate": 0.0029965187304705695,
+      "loss": 1.2865,
+      "step": 282
+    },
+    {
+      "epoch": 0.02176755634181986,
+      "learning_rate": 0.0029964940064790098,
+      "loss": 1.2444,
+      "step": 283
+    },
+    {
+      "epoch": 0.021844473502038306,
+      "learning_rate": 0.0029964691951056043,
+      "loss": 1.1044,
+      "step": 284
+    },
+    {
+      "epoch": 0.02192139066225675,
+      "learning_rate": 0.0029964442963518042,
+      "loss": 1.3341,
+      "step": 285
+    },
+    {
+      "epoch": 0.021998307822475194,
+      "learning_rate": 0.002996419310219061,
+      "loss": 0.9102,
+      "step": 286
+    },
+    {
+      "epoch": 0.02207522498269364,
+      "learning_rate": 0.0029963942367088356,
+      "loss": 1.6223,
+      "step": 287
+    },
+    {
+      "epoch": 0.022152142142912085,
+      "learning_rate": 0.002996369075822591,
+      "loss": 1.2945,
+      "step": 288
+    },
+    {
+      "epoch": 0.02222905930313053,
+      "learning_rate": 0.0029963438275617963,
+      "loss": 1.1997,
+      "step": 289
+    },
+    {
+      "epoch": 0.022305976463348972,
+      "learning_rate": 0.002996318491927926,
+      "loss": 1.1263,
+      "step": 290
+    },
+    {
+      "epoch": 0.022382893623567417,
+      "learning_rate": 0.00299629306892246,
+      "loss": 1.1537,
+      "step": 291
+    },
+    {
+      "epoch": 0.022459810783785863,
+      "learning_rate": 0.002996267558546882,
+      "loss": 1.0203,
+      "step": 292
+    },
+    {
+      "epoch": 0.022536727944004308,
+      "learning_rate": 0.0029962419608026815,
+      "loss": 1.1749,
+      "step": 293
+    },
+    {
+      "epoch": 0.022613645104222754,
+      "learning_rate": 0.0029962162756913538,
+      "loss": 0.9919,
+      "step": 294
+    },
+    {
+      "epoch": 0.022690562264441196,
+      "learning_rate": 0.002996190503214399,
+      "loss": 1.3889,
+      "step": 295
+    },
+    {
+      "epoch": 0.02276747942465964,
+      "learning_rate": 0.0029961646433733207,
+      "loss": 1.2516,
+      "step": 296
+    },
+    {
+      "epoch": 0.022844396584878086,
+      "learning_rate": 0.00299613869616963,
+      "loss": 1.1091,
+      "step": 297
+    },
+    {
+      "epoch": 0.022921313745096532,
+      "learning_rate": 0.0029961126616048417,
+      "loss": 0.9692,
+      "step": 298
+    },
+    {
+      "epoch": 0.022998230905314977,
+      "learning_rate": 0.0029960865396804755,
+      "loss": 1.6793,
+      "step": 299
+    },
+    {
+      "epoch": 0.02307514806553342,
+      "learning_rate": 0.002996060330398057,
+      "loss": 1.5065,
+      "step": 300
+    },
+    {
+      "epoch": 0.023152065225751865,
+      "learning_rate": 0.0029960340337591173,
+      "loss": 1.0508,
+      "step": 301
+    },
+    {
+      "epoch": 0.02322898238597031,
+      "learning_rate": 0.0029960076497651906,
+      "loss": 1.2244,
+      "step": 302
+    },
+    {
+      "epoch": 0.023305899546188755,
+      "learning_rate": 0.0029959811784178187,
+      "loss": 1.2929,
+      "step": 303
+    },
+    {
+      "epoch": 0.0233828167064072,
+      "learning_rate": 0.0029959546197185463,
+      "loss": 1.0615,
+      "step": 304
+    },
+    {
+      "epoch": 0.023459733866625643,
+      "learning_rate": 0.002995927973668925,
+      "loss": 1.2689,
+      "step": 305
+    },
+    {
+      "epoch": 0.023536651026844088,
+      "learning_rate": 0.0029959012402705097,
+      "loss": 1.1925,
+      "step": 306
+    },
+    {
+      "epoch": 0.023613568187062534,
+      "learning_rate": 0.0029958744195248623,
+      "loss": 1.0969,
+      "step": 307
+    },
+    {
+      "epoch": 0.02369048534728098,
+      "learning_rate": 0.0029958475114335486,
+      "loss": 1.2156,
+      "step": 308
+    },
+    {
+      "epoch": 0.023767402507499424,
+      "learning_rate": 0.0029958205159981398,
+      "loss": 1.3017,
+      "step": 309
+    },
+    {
+      "epoch": 0.023844319667717866,
+      "learning_rate": 0.002995793433220212,
+      "loss": 1.0581,
+      "step": 310
+    },
+    {
+      "epoch": 0.023921236827936312,
+      "learning_rate": 0.002995766263101347,
+      "loss": 1.2094,
+      "step": 311
+    },
+    {
+      "epoch": 0.023998153988154757,
+      "learning_rate": 0.002995739005643131,
+      "loss": 0.9861,
+      "step": 312
+    },
+    {
+      "epoch": 0.024075071148373203,
+      "learning_rate": 0.0029957116608471555,
+      "loss": 0.9938,
+      "step": 313
+    },
+    {
+      "epoch": 0.024151988308591648,
+      "learning_rate": 0.002995684228715017,
+      "loss": 1.0069,
+      "step": 314
+    },
+    {
+      "epoch": 0.02422890546881009,
+      "learning_rate": 0.0029956567092483182,
+      "loss": 1.0306,
+      "step": 315
+    },
+    {
+      "epoch": 0.024305822629028535,
+      "learning_rate": 0.0029956291024486655,
+      "loss": 1.2565,
+      "step": 316
+    },
+    {
+      "epoch": 0.02438273978924698,
+      "learning_rate": 0.0029956014083176706,
+      "loss": 1.054,
+      "step": 317
+    },
+    {
+      "epoch": 0.024459656949465426,
+      "learning_rate": 0.0029955736268569505,
+      "loss": 0.9882,
+      "step": 318
+    },
+    {
+      "epoch": 0.02453657410968387,
+      "learning_rate": 0.002995545758068128,
+      "loss": 1.4163,
+      "step": 319
+    },
+    {
+      "epoch": 0.024613491269902314,
+      "learning_rate": 0.0029955178019528297,
+      "loss": 0.9077,
+      "step": 320
+    },
+    {
+      "epoch": 0.02469040843012076,
+      "learning_rate": 0.002995489758512689,
+      "loss": 1.0501,
+      "step": 321
+    },
+    {
+      "epoch": 0.024767325590339204,
+      "learning_rate": 0.002995461627749342,
+      "loss": 1.2069,
+      "step": 322
+    },
+    {
+      "epoch": 0.02484424275055765,
+      "learning_rate": 0.0029954334096644326,
+      "loss": 1.3046,
+      "step": 323
+    },
+    {
+      "epoch": 0.024921159910776095,
+      "learning_rate": 0.0029954051042596075,
+      "loss": 1.3116,
+      "step": 324
+    },
+    {
+      "epoch": 0.02499807707099454,
+      "learning_rate": 0.0029953767115365204,
+      "loss": 1.2408,
+      "step": 325
+    },
+    {
+      "epoch": 0.025074994231212983,
+      "learning_rate": 0.002995348231496828,
+      "loss": 0.8897,
+      "step": 326
+    },
+    {
+      "epoch": 0.025151911391431428,
+      "learning_rate": 0.0029953196641421946,
+      "loss": 1.9307,
+      "step": 327
+    },
+    {
+      "epoch": 0.025228828551649873,
+      "learning_rate": 0.002995291009474287,
+      "loss": 1.0036,
+      "step": 328
+    },
+    {
+      "epoch": 0.02530574571186832,
+      "learning_rate": 0.0029952622674947796,
+      "loss": 1.1598,
+      "step": 329
+    },
+    {
+      "epoch": 0.025382662872086764,
+      "learning_rate": 0.0029952334382053495,
+      "loss": 0.8737,
+      "step": 330
+    },
+    {
+      "epoch": 0.025459580032305206,
+      "learning_rate": 0.0029952045216076807,
+      "loss": 1.431,
+      "step": 331
+    },
+    {
+      "epoch": 0.02553649719252365,
+      "learning_rate": 0.0029951755177034622,
+      "loss": 1.4716,
+      "step": 332
+    },
+    {
+      "epoch": 0.025613414352742097,
+      "learning_rate": 0.0029951464264943864,
+      "loss": 0.9831,
+      "step": 333
+    },
+    {
+      "epoch": 0.025690331512960542,
+      "learning_rate": 0.0029951172479821533,
+      "loss": 1.1124,
+      "step": 334
+    },
+    {
+      "epoch": 0.025767248673178988,
+      "learning_rate": 0.002995087982168465,
+      "loss": 1.0785,
+      "step": 335
+    },
+    {
+      "epoch": 0.02584416583339743,
+      "learning_rate": 0.0029950586290550317,
+      "loss": 1.1973,
+      "step": 336
+    },
+    {
+      "epoch": 0.025921082993615875,
+      "learning_rate": 0.0029950291886435667,
+      "loss": 1.3794,
+      "step": 337
+    },
+    {
+      "epoch": 0.02599800015383432,
+      "learning_rate": 0.00299499966093579,
+      "loss": 1.0184,
+      "step": 338
+    },
+    {
+      "epoch": 0.026074917314052766,
+      "learning_rate": 0.002994970045933424,
+      "loss": 1.3075,
+      "step": 339
+    },
+    {
+      "epoch": 0.02615183447427121,
+      "learning_rate": 0.0029949403436382,
+      "loss": 1.416,
+      "step": 340
+    },
+    {
+      "epoch": 0.026228751634489653,
+      "learning_rate": 0.0029949105540518508,
+      "loss": 1.4782,
+      "step": 341
+    },
+    {
+      "epoch": 0.0263056687947081,
+      "learning_rate": 0.0029948806771761165,
+      "loss": 1.1191,
+      "step": 342
+    },
+    {
+      "epoch": 0.026382585954926544,
+      "learning_rate": 0.0029948507130127416,
+      "loss": 0.8146,
+      "step": 343
+    },
+    {
+      "epoch": 0.02645950311514499,
+      "learning_rate": 0.0029948206615634755,
+      "loss": 2.0228,
+      "step": 344
+    },
+    {
+      "epoch": 0.026536420275363435,
+      "learning_rate": 0.0029947905228300733,
+      "loss": 1.329,
+      "step": 345
+    },
+    {
+      "epoch": 0.026613337435581877,
+      "learning_rate": 0.0029947602968142945,
+      "loss": 1.0044,
+      "step": 346
+    },
+    {
+      "epoch": 0.026690254595800322,
+      "learning_rate": 0.002994729983517904,
+      "loss": 1.1451,
+      "step": 347
+    },
+    {
+      "epoch": 0.026767171756018768,
+      "learning_rate": 0.002994699582942673,
+      "loss": 1.4733,
+      "step": 348
+    },
+    {
+      "epoch": 0.026844088916237213,
+      "learning_rate": 0.002994669095090374,
+      "loss": 0.8037,
+      "step": 349
+    },
+    {
+      "epoch": 0.02692100607645566,
+      "learning_rate": 0.00299463851996279,
+      "loss": 1.0593,
+      "step": 350
+    },
+    {
+      "epoch": 0.0269979232366741,
+      "learning_rate": 0.0029946078575617046,
+      "loss": 1.276,
+      "step": 351
+    },
+    {
+      "epoch": 0.027074840396892546,
+      "learning_rate": 0.002994577107888909,
+      "loss": 1.3483,
+      "step": 352
+    },
+    {
+      "epoch": 0.02715175755711099,
+      "learning_rate": 0.0029945462709461985,
+      "loss": 0.9461,
+      "step": 353
+    },
+    {
+      "epoch": 0.027228674717329437,
+      "learning_rate": 0.0029945153467353736,
+      "loss": 1.2791,
+      "step": 354
+    },
+    {
+      "epoch": 0.027305591877547882,
+      "learning_rate": 0.00299448433525824,
+      "loss": 0.7401,
+      "step": 355
+    },
+    {
+      "epoch": 0.027382509037766324,
+      "learning_rate": 0.0029944532365166086,
+      "loss": 1.539,
+      "step": 356
+    },
+    {
+      "epoch": 0.02745942619798477,
+      "learning_rate": 0.002994422050512295,
+      "loss": 1.0192,
+      "step": 357
+    },
+    {
+      "epoch": 0.027536343358203215,
+      "learning_rate": 0.002994390777247121,
+      "loss": 1.0759,
+      "step": 358
+    },
+    {
+      "epoch": 0.02761326051842166,
+      "learning_rate": 0.0029943594167229116,
+      "loss": 0.9437,
+      "step": 359
+    },
+    {
+      "epoch": 0.027690177678640106,
+      "learning_rate": 0.0029943279689414984,
+      "loss": 1.3718,
+      "step": 360
+    },
+    {
+      "epoch": 0.027767094838858548,
+      "learning_rate": 0.002994296433904718,
+      "loss": 1.1736,
+      "step": 361
+    },
+    {
+      "epoch": 0.027844011999076993,
+      "learning_rate": 0.0029942648116144115,
+      "loss": 1.116,
+      "step": 362
+    },
+    {
+      "epoch": 0.02792092915929544,
+      "learning_rate": 0.002994233102072425,
+      "loss": 1.3106,
+      "step": 363
+    },
+    {
+      "epoch": 0.027997846319513884,
+      "learning_rate": 0.002994201305280611,
+      "loss": 1.15,
+      "step": 364
+    },
+    {
+      "epoch": 0.02807476347973233,
+      "learning_rate": 0.0029941694212408253,
+      "loss": 1.177,
+      "step": 365
+    },
+    {
+      "epoch": 0.02815168063995077,
+      "learning_rate": 0.0029941374499549297,
+      "loss": 1.2402,
+      "step": 366
+    },
+    {
+      "epoch": 0.028228597800169217,
+      "learning_rate": 0.002994105391424792,
+      "loss": 1.4125,
+      "step": 367
+    },
+    {
+      "epoch": 0.028305514960387662,
+      "learning_rate": 0.0029940732456522822,
+      "loss": 1.2672,
+      "step": 368
+    },
+    {
+      "epoch": 0.028382432120606108,
+      "learning_rate": 0.0029940410126392796,
+      "loss": 1.5322,
+      "step": 369
+    },
+    {
+      "epoch": 0.028459349280824553,
+      "learning_rate": 0.0029940086923876644,
+      "loss": 1.2714,
+      "step": 370
+    },
+    {
+      "epoch": 0.028536266441043,
+      "learning_rate": 0.0029939762848993252,
+      "loss": 1.2146,
+      "step": 371
+    },
+    {
+      "epoch": 0.02861318360126144,
+      "learning_rate": 0.0029939437901761535,
+      "loss": 1.4489,
+      "step": 372
+    },
+    {
+      "epoch": 0.028690100761479886,
+      "learning_rate": 0.002993911208220047,
+      "loss": 0.878,
+      "step": 373
+    },
+    {
+      "epoch": 0.02876701792169833,
+      "learning_rate": 0.002993878539032908,
+      "loss": 0.9767,
+      "step": 374
+    },
+    {
+      "epoch": 0.028843935081916777,
+      "learning_rate": 0.0029938457826166444,
+      "loss": 1.1914,
+      "step": 375
+    },
+    {
+      "epoch": 0.028920852242135222,
+      "learning_rate": 0.0029938129389731685,
+      "loss": 1.1226,
+      "step": 376
+    },
+    {
+      "epoch": 0.028997769402353664,
+      "learning_rate": 0.0029937800081043984,
+      "loss": 1.3933,
+      "step": 377
+    },
+    {
+      "epoch": 0.02907468656257211,
+      "learning_rate": 0.002993746990012257,
+      "loss": 1.2843,
+      "step": 378
+    },
+    {
+      "epoch": 0.029151603722790555,
+      "learning_rate": 0.002993713884698672,
+      "loss": 1.5428,
+      "step": 379
+    },
+    {
+      "epoch": 0.029228520883009,
+      "learning_rate": 0.0029936806921655765,
+      "loss": 1.1047,
+      "step": 380
+    },
+    {
+      "epoch": 0.029305438043227446,
+      "learning_rate": 0.002993647412414909,
+      "loss": 1.5107,
+      "step": 381
+    },
+    {
+      "epoch": 0.029382355203445887,
+      "learning_rate": 0.002993614045448612,
+      "loss": 1.3802,
+      "step": 382
+    },
+    {
+      "epoch": 0.029459272363664333,
+      "learning_rate": 0.0029935805912686343,
+      "loss": 1.1889,
+      "step": 383
+    },
+    {
+      "epoch": 0.02953618952388278,
+      "learning_rate": 0.0029935470498769295,
+      "loss": 1.0533,
+      "step": 384
+    },
+    {
+      "epoch": 0.029613106684101224,
+      "learning_rate": 0.0029935134212754557,
+      "loss": 1.1599,
+      "step": 385
+    },
+    {
+      "epoch": 0.02969002384431967,
+      "learning_rate": 0.0029934797054661775,
+      "loss": 0.9938,
+      "step": 386
+    },
+    {
+      "epoch": 0.02976694100453811,
+      "learning_rate": 0.0029934459024510618,
+      "loss": 1.2057,
+      "step": 387
+    },
+    {
+      "epoch": 0.029843858164756557,
+      "learning_rate": 0.002993412012232084,
+      "loss": 0.9993,
+      "step": 388
+    },
+    {
+      "epoch": 0.029920775324975002,
+      "learning_rate": 0.002993378034811222,
+      "loss": 1.4387,
+      "step": 389
+    },
+    {
+      "epoch": 0.029997692485193447,
+      "learning_rate": 0.002993343970190461,
+      "loss": 1.2548,
+      "step": 390
+    },
+    {
+      "epoch": 0.030074609645411893,
+      "learning_rate": 0.0029933098183717883,
+      "loss": 1.747,
+      "step": 391
+    },
+    {
+      "epoch": 0.030151526805630335,
+      "learning_rate": 0.0029932755793571995,
+      "loss": 1.5008,
+      "step": 392
+    },
+    {
+      "epoch": 0.03022844396584878,
+      "learning_rate": 0.002993241253148693,
+      "loss": 1.0202,
+      "step": 393
+    },
+    {
+      "epoch": 0.030305361126067226,
+      "learning_rate": 0.002993206839748274,
+      "loss": 1.0718,
+      "step": 394
+    },
+    {
+      "epoch": 0.03038227828628567,
+      "learning_rate": 0.0029931723391579514,
+      "loss": 1.2228,
+      "step": 395
+    },
+    {
+      "epoch": 0.030459195446504116,
+      "learning_rate": 0.0029931377513797396,
+      "loss": 1.4326,
+      "step": 396
+    },
+    {
+      "epoch": 0.03053611260672256,
+      "learning_rate": 0.0029931030764156585,
+      "loss": 1.293,
+      "step": 397
+    },
+    {
+      "epoch": 0.030613029766941004,
+      "learning_rate": 0.0029930683142677323,
+      "loss": 1.514,
+      "step": 398
+    },
+    {
+      "epoch": 0.03068994692715945,
+      "learning_rate": 0.0029930334649379917,
+      "loss": 1.1175,
+      "step": 399
+    },
+    {
+      "epoch": 0.030766864087377895,
+      "learning_rate": 0.002992998528428471,
+      "loss": 1.0261,
+      "step": 400
+    },
+    {
+      "epoch": 0.03084378124759634,
+      "learning_rate": 0.0029929635047412097,
+      "loss": 1.2312,
+      "step": 401
+    },
+    {
+      "epoch": 0.030920698407814782,
+      "learning_rate": 0.002992928393878254,
+      "loss": 1.1255,
+      "step": 402
+    },
+    {
+      "epoch": 0.030997615568033227,
+      "learning_rate": 0.0029928931958416536,
+      "loss": 1.2353,
+      "step": 403
+    },
+    {
+      "epoch": 0.031074532728251673,
+      "learning_rate": 0.002992857910633463,
+      "loss": 1.3461,
+      "step": 404
+    },
+    {
+      "epoch": 0.031151449888470118,
+      "learning_rate": 0.002992822538255744,
+      "loss": 1.3139,
+      "step": 405
+    },
+    {
+      "epoch": 0.031228367048688564,
+      "learning_rate": 0.0029927870787105604,
+      "loss": 1.2285,
+      "step": 406
+    },
+    {
+      "epoch": 0.031305284208907005,
+      "learning_rate": 0.002992751531999984,
+      "loss": 1.4545,
+      "step": 407
+    },
+    {
+      "epoch": 0.03138220136912545,
+      "learning_rate": 0.00299271589812609,
+      "loss": 1.2949,
+      "step": 408
+    },
+    {
+      "epoch": 0.031459118529343896,
+      "learning_rate": 0.002992680177090959,
+      "loss": 1.0288,
+      "step": 409
+    },
+    {
+      "epoch": 0.03153603568956234,
+      "learning_rate": 0.0029926443688966763,
+      "loss": 0.8704,
+      "step": 410
+    },
+    {
+      "epoch": 0.03161295284978079,
+      "learning_rate": 0.0029926084735453335,
+      "loss": 1.3113,
+      "step": 411
+    },
+    {
+      "epoch": 0.03168987000999923,
+      "learning_rate": 0.002992572491039027,
+      "loss": 0.9521,
+      "step": 412
+    },
+    {
+      "epoch": 0.03176678717021768,
+      "learning_rate": 0.002992536421379857,
+      "loss": 1.1472,
+      "step": 413
+    },
+    {
+      "epoch": 0.03184370433043612,
+      "learning_rate": 0.0029925002645699297,
+      "loss": 1.3,
+      "step": 414
+    },
+    {
+      "epoch": 0.03192062149065456,
+      "learning_rate": 0.002992464020611356,
+      "loss": 1.5984,
+      "step": 415
+    },
+    {
+      "epoch": 0.03199753865087301,
+      "learning_rate": 0.0029924276895062535,
+      "loss": 0.7235,
+      "step": 416
+    },
+    {
+      "epoch": 0.03207445581109145,
+      "learning_rate": 0.0029923912712567424,
+      "loss": 1.4046,
+      "step": 417
+    },
+    {
+      "epoch": 0.0321513729713099,
+      "learning_rate": 0.00299235476586495,
+      "loss": 1.1582,
+      "step": 418
+    },
+    {
+      "epoch": 0.032228290131528343,
+      "learning_rate": 0.002992318173333007,
+      "loss": 1.2819,
+      "step": 419
+    },
+    {
+      "epoch": 0.03230520729174679,
+      "learning_rate": 0.002992281493663051,
+      "loss": 1.0412,
+      "step": 420
+    },
+    {
+      "epoch": 0.032382124451965234,
+      "learning_rate": 0.0029922447268572235,
+      "loss": 1.2739,
+      "step": 421
+    },
+    {
+      "epoch": 0.03245904161218368,
+      "learning_rate": 0.0029922078729176706,
+      "loss": 1.2815,
+      "step": 422
+    },
+    {
+      "epoch": 0.032535958772402125,
+      "learning_rate": 0.0029921709318465454,
+      "loss": 1.4791,
+      "step": 423
+    },
+    {
+      "epoch": 0.03261287593262057,
+      "learning_rate": 0.002992133903646004,
+      "loss": 1.1519,
+      "step": 424
+    },
+    {
+      "epoch": 0.03268979309283901,
+      "learning_rate": 0.0029920967883182094,
+      "loss": 1.0397,
+      "step": 425
+    },
+    {
+      "epoch": 0.032766710253057454,
+      "learning_rate": 0.002992059585865328,
+      "loss": 0.7153,
+      "step": 426
+    },
+    {
+      "epoch": 0.0328436274132759,
+      "learning_rate": 0.0029920222962895324,
+      "loss": 1.1698,
+      "step": 427
+    },
+    {
+      "epoch": 0.032920544573494345,
+      "learning_rate": 0.0029919849195930003,
+      "loss": 1.0705,
+      "step": 428
+    },
+    {
+      "epoch": 0.03299746173371279,
+      "learning_rate": 0.0029919474557779137,
+      "loss": 1.2043,
+      "step": 429
+    },
+    {
+      "epoch": 0.033074378893931236,
+      "learning_rate": 0.00299190990484646,
+      "loss": 1.1393,
+      "step": 430
+    },
+    {
+      "epoch": 0.03315129605414968,
+      "learning_rate": 0.0029918722668008323,
+      "loss": 1.285,
+      "step": 431
+    },
+    {
+      "epoch": 0.03322821321436813,
+      "learning_rate": 0.002991834541643228,
+      "loss": 1.0071,
+      "step": 432
+    },
+    {
+      "epoch": 0.03330513037458657,
+      "learning_rate": 0.00299179672937585,
+      "loss": 1.197,
+      "step": 433
+    },
+    {
+      "epoch": 0.03338204753480502,
+      "learning_rate": 0.0029917588300009063,
+      "loss": 1.0339,
+      "step": 434
+    },
+    {
+      "epoch": 0.03345896469502346,
+      "learning_rate": 0.00299172084352061,
+      "loss": 1.3188,
+      "step": 435
+    },
+    {
+      "epoch": 0.0335358818552419,
+      "learning_rate": 0.002991682769937179,
+      "loss": 1.0145,
+      "step": 436
+    },
+    {
+      "epoch": 0.03361279901546035,
+      "learning_rate": 0.0029916446092528364,
+      "loss": 0.8821,
+      "step": 437
+    },
+    {
+      "epoch": 0.03368971617567879,
+      "learning_rate": 0.0029916063614698106,
+      "loss": 0.9893,
+      "step": 438
+    },
+    {
+      "epoch": 0.03376663333589724,
+      "learning_rate": 0.002991568026590335,
+      "loss": 1.1368,
+      "step": 439
+    },
+    {
+      "epoch": 0.03384355049611568,
+      "learning_rate": 0.0029915296046166477,
+      "loss": 0.9755,
+      "step": 440
+    },
+    {
+      "epoch": 0.03392046765633413,
+      "learning_rate": 0.0029914910955509922,
+      "loss": 1.1123,
+      "step": 441
+    },
+    {
+      "epoch": 0.033997384816552574,
+      "learning_rate": 0.0029914524993956177,
+      "loss": 1.1482,
+      "step": 442
+    },
+    {
+      "epoch": 0.03407430197677102,
+      "learning_rate": 0.002991413816152777,
+      "loss": 0.9763,
+      "step": 443
+    },
+    {
+      "epoch": 0.034151219136989465,
+      "learning_rate": 0.0029913750458247287,
+      "loss": 0.8,
+      "step": 444
+    },
+    {
+      "epoch": 0.03422813629720791,
+      "learning_rate": 0.002991336188413738,
+      "loss": 1.1179,
+      "step": 445
+    },
+    {
+      "epoch": 0.03430505345742635,
+      "learning_rate": 0.002991297243922073,
+      "loss": 1.185,
+      "step": 446
+    },
+    {
+      "epoch": 0.034381970617644794,
+      "learning_rate": 0.0029912582123520076,
+      "loss": 1.2302,
+      "step": 447
+    },
+    {
+      "epoch": 0.03445888777786324,
+      "learning_rate": 0.0029912190937058207,
+      "loss": 1.2877,
+      "step": 448
+    },
+    {
+      "epoch": 0.034535804938081685,
+      "learning_rate": 0.0029911798879857974,
+      "loss": 0.9684,
+      "step": 449
+    },
+    {
+      "epoch": 0.03461272209830013,
+      "learning_rate": 0.0029911405951942257,
+      "loss": 1.5201,
+      "step": 450
+    },
+    {
+      "epoch": 0.034689639258518576,
+      "learning_rate": 0.002991101215333401,
+      "loss": 1.19,
+      "step": 451
+    },
+    {
+      "epoch": 0.03476655641873702,
+      "learning_rate": 0.002991061748405623,
+      "loss": 1.2372,
+      "step": 452
+    },
+    {
+      "epoch": 0.03484347357895547,
+      "learning_rate": 0.0029910221944131946,
+      "loss": 1.1227,
+      "step": 453
+    },
+    {
+      "epoch": 0.03492039073917391,
+      "learning_rate": 0.0029909825533584265,
+      "loss": 0.8888,
+      "step": 454
+    },
+    {
+      "epoch": 0.03499730789939236,
+      "learning_rate": 0.0029909428252436338,
+      "loss": 1.14,
+      "step": 455
+    },
+    {
+      "epoch": 0.035074225059610796,
+      "learning_rate": 0.002990903010071135,
+      "loss": 1.1346,
+      "step": 456
+    },
+    {
+      "epoch": 0.03515114221982924,
+      "learning_rate": 0.0029908631078432563,
+      "loss": 1.4103,
+      "step": 457
+    },
+    {
+      "epoch": 0.03522805938004769,
+      "learning_rate": 0.0029908231185623263,
+      "loss": 1.0122,
+      "step": 458
+    },
+    {
+      "epoch": 0.03530497654026613,
+      "learning_rate": 0.0029907830422306815,
+      "loss": 1.1954,
+      "step": 459
+    },
+    {
+      "epoch": 0.03538189370048458,
+      "learning_rate": 0.0029907428788506604,
+      "loss": 1.2236,
+      "step": 460
+    },
+    {
+      "epoch": 0.03545881086070302,
+      "learning_rate": 0.0029907026284246097,
+      "loss": 0.8695,
+      "step": 461
+    },
+    {
+      "epoch": 0.03553572802092147,
+      "learning_rate": 0.0029906622909548785,
+      "loss": 0.7571,
+      "step": 462
+    },
+    {
+      "epoch": 0.035612645181139914,
+      "learning_rate": 0.0029906218664438227,
+      "loss": 1.1938,
+      "step": 463
+    },
+    {
+      "epoch": 0.03568956234135836,
+      "learning_rate": 0.0029905813548938027,
+      "loss": 0.971,
+      "step": 464
+    },
+    {
+      "epoch": 0.035766479501576805,
+      "learning_rate": 0.002990540756307184,
+      "loss": 1.3198,
+      "step": 465
+    },
+    {
+      "epoch": 0.03584339666179524,
+      "learning_rate": 0.002990500070686337,
+      "loss": 0.9848,
+      "step": 466
+    },
+    {
+      "epoch": 0.03592031382201369,
+      "learning_rate": 0.0029904592980336373,
+      "loss": 1.342,
+      "step": 467
+    },
+    {
+      "epoch": 0.035997230982232134,
+      "learning_rate": 0.002990418438351466,
+      "loss": 1.4538,
+      "step": 468
+    },
+    {
+      "epoch": 0.03607414814245058,
+      "learning_rate": 0.0029903774916422094,
+      "loss": 1.2134,
+      "step": 469
+    },
+    {
+      "epoch": 0.036151065302669025,
+      "learning_rate": 0.002990336457908257,
+      "loss": 1.1878,
+      "step": 470
+    },
+    {
+      "epoch": 0.03622798246288747,
+      "learning_rate": 0.002990295337152006,
+      "loss": 1.277,
+      "step": 471
+    },
+    {
+      "epoch": 0.036304899623105916,
+      "learning_rate": 0.0029902541293758576,
+      "loss": 1.2576,
+      "step": 472
+    },
+    {
+      "epoch": 0.03638181678332436,
+      "learning_rate": 0.0029902128345822166,
+      "loss": 0.758,
+      "step": 473
+    },
+    {
+      "epoch": 0.036458733943542806,
+      "learning_rate": 0.0029901714527734956,
+      "loss": 1.2159,
+      "step": 474
+    },
+    {
+      "epoch": 0.03653565110376125,
+      "learning_rate": 0.00299012998395211,
+      "loss": 1.1898,
+      "step": 475
+    },
+    {
+      "epoch": 0.0366125682639797,
+      "learning_rate": 0.0029900884281204823,
+      "loss": 1.3354,
+      "step": 476
+    },
+    {
+      "epoch": 0.036689485424198136,
+      "learning_rate": 0.0029900467852810382,
+      "loss": 1.1051,
+      "step": 477
+    },
+    {
+      "epoch": 0.03676640258441658,
+      "learning_rate": 0.002990005055436209,
+      "loss": 1.3771,
+      "step": 478
+    },
+    {
+      "epoch": 0.03684331974463503,
+      "learning_rate": 0.0029899632385884324,
+      "loss": 1.0696,
+      "step": 479
+    },
+    {
+      "epoch": 0.03692023690485347,
+      "learning_rate": 0.002989921334740149,
+      "loss": 1.15,
+      "step": 480
+    },
+    {
+      "epoch": 0.03699715406507192,
+      "learning_rate": 0.002989879343893806,
+      "loss": 1.3266,
+      "step": 481
+    },
+    {
+      "epoch": 0.03707407122529036,
+      "learning_rate": 0.002989837266051856,
+      "loss": 1.2287,
+      "step": 482
+    },
+    {
+      "epoch": 0.03715098838550881,
+      "learning_rate": 0.002989795101216755,
+      "loss": 1.2021,
+      "step": 483
+    },
+    {
+      "epoch": 0.037227905545727254,
+      "learning_rate": 0.0029897528493909656,
+      "loss": 1.4613,
+      "step": 484
+    },
+    {
+      "epoch": 0.0373048227059457,
+      "learning_rate": 0.0029897105105769544,
+      "loss": 0.9157,
+      "step": 485
+    },
+    {
+      "epoch": 0.037381739866164145,
+      "learning_rate": 0.002989668084777194,
+      "loss": 1.4113,
+      "step": 486
+    },
+    {
+      "epoch": 0.03745865702638258,
+      "learning_rate": 0.0029896255719941618,
+      "loss": 1.1766,
+      "step": 487
+    },
+    {
+      "epoch": 0.03753557418660103,
+      "learning_rate": 0.00298958297223034,
+      "loss": 1.5105,
+      "step": 488
+    },
+    {
+      "epoch": 0.037612491346819474,
+      "learning_rate": 0.0029895402854882157,
+      "loss": 1.1155,
+      "step": 489
+    },
+    {
+      "epoch": 0.03768940850703792,
+      "learning_rate": 0.002989497511770282,
+      "loss": 1.1206,
+      "step": 490
+    },
+    {
+      "epoch": 0.037766325667256365,
+      "learning_rate": 0.002989454651079037,
+      "loss": 0.8207,
+      "step": 491
+    },
+    {
+      "epoch": 0.03784324282747481,
+      "learning_rate": 0.0029894117034169813,
+      "loss": 1.1771,
+      "step": 492
+    },
+    {
+      "epoch": 0.037920159987693255,
+      "learning_rate": 0.0029893686687866247,
+      "loss": 1.0857,
+      "step": 493
+    },
+    {
+      "epoch": 0.0379970771479117,
+      "learning_rate": 0.0029893255471904797,
+      "loss": 1.0518,
+      "step": 494
+    },
+    {
+      "epoch": 0.038073994308130146,
+      "learning_rate": 0.002989282338631063,
+      "loss": 1.1607,
+      "step": 495
+    },
+    {
+      "epoch": 0.03815091146834859,
+      "learning_rate": 0.0029892390431108997,
+      "loss": 0.9763,
+      "step": 496
+    },
+    {
+      "epoch": 0.03822782862856703,
+      "learning_rate": 0.0029891956606325158,
+      "loss": 1.2348,
+      "step": 497
+    },
+    {
+      "epoch": 0.038304745788785476,
+      "learning_rate": 0.002989152191198445,
+      "loss": 1.3312,
+      "step": 498
+    },
+    {
+      "epoch": 0.03838166294900392,
+      "learning_rate": 0.0029891086348112265,
+      "loss": 1.1876,
+      "step": 499
+    },
+    {
+      "epoch": 0.038458580109222366,
+      "learning_rate": 0.002989064991473402,
+      "loss": 1.3925,
+      "step": 500
+    },
+    {
+      "epoch": 0.03853549726944081,
+      "learning_rate": 0.002989021261187522,
+      "loss": 1.2061,
+      "step": 501
+    },
+    {
+      "epoch": 0.03861241442965926,
+      "learning_rate": 0.0029889774439561378,
+      "loss": 1.2061,
+      "step": 502
+    },
+    {
+      "epoch": 0.0386893315898777,
+      "learning_rate": 0.002988933539781809,
+      "loss": 1.178,
+      "step": 503
+    },
+    {
+      "epoch": 0.03876624875009615,
+      "learning_rate": 0.0029888895486670994,
+      "loss": 1.0782,
+      "step": 504
+    },
+    {
+      "epoch": 0.038843165910314593,
+      "learning_rate": 0.002988845470614577,
+      "loss": 1.1933,
+      "step": 505
+    },
+    {
+      "epoch": 0.03892008307053304,
+      "learning_rate": 0.002988801305626816,
+      "loss": 1.0163,
+      "step": 506
+    },
+    {
+      "epoch": 0.03899700023075148,
+      "learning_rate": 0.0029887570537063955,
+      "loss": 1.1219,
+      "step": 507
+    },
+    {
+      "epoch": 0.03907391739096992,
+      "learning_rate": 0.002988712714855899,
+      "loss": 1.0401,
+      "step": 508
+    },
+    {
+      "epoch": 0.03915083455118837,
+      "learning_rate": 0.0029886682890779156,
+      "loss": 1.1159,
+      "step": 509
+    },
+    {
+      "epoch": 0.039227751711406814,
+      "learning_rate": 0.002988623776375039,
+      "loss": 0.8955,
+      "step": 510
+    },
+    {
+      "epoch": 0.03930466887162526,
+      "learning_rate": 0.002988579176749869,
+      "loss": 1.1181,
+      "step": 511
+    },
+    {
+      "epoch": 0.039381586031843704,
+      "learning_rate": 0.0029885344902050096,
+      "loss": 0.998,
+      "step": 512
+    },
+    {
+      "epoch": 0.03945850319206215,
+      "learning_rate": 0.00298848971674307,
+      "loss": 0.9474,
+      "step": 513
+    },
+    {
+      "epoch": 0.039535420352280595,
+      "learning_rate": 0.0029884448563666645,
+      "loss": 1.3869,
+      "step": 514
+    },
+    {
+      "epoch": 0.03961233751249904,
+      "learning_rate": 0.0029883999090784125,
+      "loss": 1.0953,
+      "step": 515
+    },
+    {
+      "epoch": 0.039689254672717486,
+      "learning_rate": 0.0029883548748809385,
+      "loss": 1.061,
+      "step": 516
+    },
+    {
+      "epoch": 0.039766171832935925,
+      "learning_rate": 0.0029883097537768723,
+      "loss": 1.1206,
+      "step": 517
+    },
+    {
+      "epoch": 0.03984308899315437,
+      "learning_rate": 0.0029882645457688485,
+      "loss": 1.0127,
+      "step": 518
+    },
+    {
+      "epoch": 0.039920006153372815,
+      "learning_rate": 0.002988219250859507,
+      "loss": 1.5546,
+      "step": 519
+    },
+    {
+      "epoch": 0.03999692331359126,
+      "learning_rate": 0.0029881738690514924,
+      "loss": 1.1249,
+      "step": 520
+    },
+    {
+      "epoch": 0.040073840473809706,
+      "learning_rate": 0.0029881284003474545,
+      "loss": 1.5855,
+      "step": 521
+    },
+    {
+      "epoch": 0.04015075763402815,
+      "learning_rate": 0.002988082844750049,
+      "loss": 1.1175,
+      "step": 522
+    },
+    {
+      "epoch": 0.0402276747942466,
+      "learning_rate": 0.0029880372022619344,
+      "loss": 1.0745,
+      "step": 523
+    },
+    {
+      "epoch": 0.04030459195446504,
+      "learning_rate": 0.002987991472885777,
+      "loss": 1.1992,
+      "step": 524
+    },
+    {
+      "epoch": 0.04038150911468349,
+      "learning_rate": 0.0029879456566242473,
+      "loss": 1.2495,
+      "step": 525
+    },
+    {
+      "epoch": 0.04045842627490193,
+      "learning_rate": 0.0029878997534800197,
+      "loss": 1.0105,
+      "step": 526
+    },
+    {
+      "epoch": 0.04053534343512038,
+      "learning_rate": 0.0029878537634557747,
+      "loss": 1.2723,
+      "step": 527
+    },
+    {
+      "epoch": 0.04061226059533882,
+      "learning_rate": 0.0029878076865541984,
+      "loss": 1.4241,
+      "step": 528
+    },
+    {
+      "epoch": 0.04068917775555726,
+      "learning_rate": 0.00298776152277798,
+      "loss": 0.9104,
+      "step": 529
+    },
+    {
+      "epoch": 0.04076609491577571,
+      "learning_rate": 0.0029877152721298162,
+      "loss": 1.346,
+      "step": 530
+    },
+    {
+      "epoch": 0.04084301207599415,
+      "learning_rate": 0.002987668934612407,
+      "loss": 1.1694,
+      "step": 531
+    },
+    {
+      "epoch": 0.0409199292362126,
+      "learning_rate": 0.0029876225102284585,
+      "loss": 0.8494,
+      "step": 532
+    },
+    {
+      "epoch": 0.040996846396431044,
+      "learning_rate": 0.0029875759989806813,
+      "loss": 0.9455,
+      "step": 533
+    },
+    {
+      "epoch": 0.04107376355664949,
+      "learning_rate": 0.002987529400871791,
+      "loss": 1.3369,
+      "step": 534
+    },
+    {
+      "epoch": 0.041150680716867935,
+      "learning_rate": 0.002987482715904509,
+      "loss": 1.1992,
+      "step": 535
+    },
+    {
+      "epoch": 0.04122759787708638,
+      "learning_rate": 0.0029874359440815604,
+      "loss": 1.2606,
+      "step": 536
+    },
+    {
+      "epoch": 0.041304515037304826,
+      "learning_rate": 0.0029873890854056778,
+      "loss": 1.5976,
+      "step": 537
+    },
+    {
+      "epoch": 0.041381432197523264,
+      "learning_rate": 0.002987342139879596,
+      "loss": 1.2291,
+      "step": 538
+    },
+    {
+      "epoch": 0.04145834935774171,
+      "learning_rate": 0.002987295107506056,
+      "loss": 0.8531,
+      "step": 539
+    },
+    {
+      "epoch": 0.041535266517960155,
+      "learning_rate": 0.002987247988287805,
+      "loss": 1.3273,
+      "step": 540
+    },
+    {
+      "epoch": 0.0416121836781786,
+      "learning_rate": 0.002987200782227594,
+      "loss": 1.0323,
+      "step": 541
+    },
+    {
+      "epoch": 0.041689100838397046,
+      "learning_rate": 0.0029871534893281797,
+      "loss": 0.9876,
+      "step": 542
+    },
+    {
+      "epoch": 0.04176601799861549,
+      "learning_rate": 0.002987106109592323,
+      "loss": 1.7456,
+      "step": 543
+    },
+    {
+      "epoch": 0.04184293515883394,
+      "learning_rate": 0.0029870586430227907,
+      "loss": 0.9066,
+      "step": 544
+    },
+    {
+      "epoch": 0.04191985231905238,
+      "learning_rate": 0.0029870110896223546,
+      "loss": 1.4167,
+      "step": 545
+    },
+    {
+      "epoch": 0.04199676947927083,
+      "learning_rate": 0.0029869634493937913,
+      "loss": 1.4998,
+      "step": 546
+    },
+    {
+      "epoch": 0.04207368663948927,
+      "learning_rate": 0.002986915722339882,
+      "loss": 1.2597,
+      "step": 547
+    },
+    {
+      "epoch": 0.04215060379970771,
+      "learning_rate": 0.0029868679084634146,
+      "loss": 1.2115,
+      "step": 548
+    },
+    {
+      "epoch": 0.04222752095992616,
+      "learning_rate": 0.0029868200077671804,
+      "loss": 1.027,
+      "step": 549
+    },
+    {
+      "epoch": 0.0423044381201446,
+      "learning_rate": 0.0029867720202539764,
+      "loss": 1.1086,
+      "step": 550
+    },
+    {
+      "epoch": 0.04238135528036305,
+      "learning_rate": 0.0029867239459266046,
+      "loss": 1.2673,
+      "step": 551
+    },
+    {
+      "epoch": 0.04245827244058149,
+      "learning_rate": 0.002986675784787872,
+      "loss": 1.4465,
+      "step": 552
+    },
+    {
+      "epoch": 0.04253518960079994,
+      "learning_rate": 0.0029866275368405915,
+      "loss": 1.1503,
+      "step": 553
+    },
+    {
+      "epoch": 0.042612106761018384,
+      "learning_rate": 0.0029865792020875794,
+      "loss": 1.2056,
+      "step": 554
+    },
+    {
+      "epoch": 0.04268902392123683,
+      "learning_rate": 0.0029865307805316582,
+      "loss": 0.8459,
+      "step": 555
+    },
+    {
+      "epoch": 0.042765941081455275,
+      "learning_rate": 0.0029864822721756557,
+      "loss": 0.896,
+      "step": 556
+    },
+    {
+      "epoch": 0.04284285824167372,
+      "learning_rate": 0.0029864336770224044,
+      "loss": 1.189,
+      "step": 557
+    },
+    {
+      "epoch": 0.04291977540189216,
+      "learning_rate": 0.0029863849950747413,
+      "loss": 1.0664,
+      "step": 558
+    },
+    {
+      "epoch": 0.042996692562110604,
+      "learning_rate": 0.002986336226335509,
+      "loss": 1.0806,
+      "step": 559
+    },
+    {
+      "epoch": 0.04307360972232905,
+      "learning_rate": 0.002986287370807556,
+      "loss": 0.7105,
+      "step": 560
+    },
+    {
+      "epoch": 0.043150526882547495,
+      "learning_rate": 0.002986238428493734,
+      "loss": 1.3178,
+      "step": 561
+    },
+    {
+      "epoch": 0.04322744404276594,
+      "learning_rate": 0.0029861893993969016,
+      "loss": 1.0987,
+      "step": 562
+    },
+    {
+      "epoch": 0.043304361202984386,
+      "learning_rate": 0.002986140283519921,
+      "loss": 1.0808,
+      "step": 563
+    },
+    {
+      "epoch": 0.04338127836320283,
+      "learning_rate": 0.0029860910808656607,
+      "loss": 1.2433,
+      "step": 564
+    },
+    {
+      "epoch": 0.04345819552342128,
+      "learning_rate": 0.0029860417914369932,
+      "loss": 1.2721,
+      "step": 565
+    },
+    {
+      "epoch": 0.04353511268363972,
+      "learning_rate": 0.0029859924152367966,
+      "loss": 1.0626,
+      "step": 566
+    },
+    {
+      "epoch": 0.04361202984385817,
+      "learning_rate": 0.0029859429522679547,
+      "loss": 1.2338,
+      "step": 567
+    },
+    {
+      "epoch": 0.04368894700407661,
+      "learning_rate": 0.002985893402533355,
+      "loss": 1.0487,
+      "step": 568
+    },
+    {
+      "epoch": 0.04376586416429505,
+      "learning_rate": 0.002985843766035891,
+      "loss": 1.0549,
+      "step": 569
+    },
+    {
+      "epoch": 0.0438427813245135,
+      "learning_rate": 0.002985794042778461,
+      "loss": 0.9051,
+      "step": 570
+    },
+    {
+      "epoch": 0.04391969848473194,
+      "learning_rate": 0.002985744232763968,
+      "loss": 0.9448,
+      "step": 571
+    },
+    {
+      "epoch": 0.04399661564495039,
+      "learning_rate": 0.002985694335995321,
+      "loss": 1.314,
+      "step": 572
+    },
+    {
+      "epoch": 0.04407353280516883,
+      "learning_rate": 0.002985644352475434,
+      "loss": 1.0038,
+      "step": 573
+    },
+    {
+      "epoch": 0.04415044996538728,
+      "learning_rate": 0.0029855942822072244,
+      "loss": 1.2588,
+      "step": 574
+    },
+    {
+      "epoch": 0.044227367125605724,
+      "learning_rate": 0.0029855441251936165,
+      "loss": 1.1826,
+      "step": 575
+    },
+    {
+      "epoch": 0.04430428428582417,
+      "learning_rate": 0.0029854938814375385,
+      "loss": 1.1935,
+      "step": 576
+    },
+    {
+      "epoch": 0.044381201446042615,
+      "learning_rate": 0.0029854435509419253,
+      "loss": 1.3885,
+      "step": 577
+    },
+    {
+      "epoch": 0.04445811860626106,
+      "learning_rate": 0.0029853931337097146,
+      "loss": 1.5294,
+      "step": 578
+    },
+    {
+      "epoch": 0.0445350357664795,
+      "learning_rate": 0.0029853426297438515,
+      "loss": 1.0981,
+      "step": 579
+    },
+    {
+      "epoch": 0.044611952926697944,
+      "learning_rate": 0.0029852920390472833,
+      "loss": 1.5897,
+      "step": 580
+    },
+    {
+      "epoch": 0.04468887008691639,
+      "learning_rate": 0.0029852413616229655,
+      "loss": 1.2229,
+      "step": 581
+    },
+    {
+      "epoch": 0.044765787247134835,
+      "learning_rate": 0.002985190597473857,
+      "loss": 1.0862,
+      "step": 582
+    },
+    {
+      "epoch": 0.04484270440735328,
+      "learning_rate": 0.002985139746602921,
+      "loss": 1.0314,
+      "step": 583
+    },
+    {
+      "epoch": 0.044919621567571726,
+      "learning_rate": 0.0029850888090131276,
+      "loss": 1.0007,
+      "step": 584
+    },
+    {
+      "epoch": 0.04499653872779017,
+      "learning_rate": 0.002985037784707451,
+      "loss": 1.0844,
+      "step": 585
+    },
+    {
+      "epoch": 0.045073455888008616,
+      "learning_rate": 0.002984986673688871,
+      "loss": 0.95,
+      "step": 586
+    },
+    {
+      "epoch": 0.04515037304822706,
+      "learning_rate": 0.002984935475960371,
+      "loss": 0.7464,
+      "step": 587
+    },
+    {
+      "epoch": 0.04522729020844551,
+      "learning_rate": 0.002984884191524941,
+      "loss": 1.3654,
+      "step": 588
+    },
+    {
+      "epoch": 0.045304207368663946,
+      "learning_rate": 0.0029848328203855756,
+      "loss": 1.3091,
+      "step": 589
+    },
+    {
+      "epoch": 0.04538112452888239,
+      "learning_rate": 0.0029847813625452743,
+      "loss": 1.201,
+      "step": 590
+    },
+    {
+      "epoch": 0.04545804168910084,
+      "learning_rate": 0.002984729818007042,
+      "loss": 0.9744,
+      "step": 591
+    },
+    {
+      "epoch": 0.04553495884931928,
+      "learning_rate": 0.002984678186773888,
+      "loss": 1.5352,
+      "step": 592
+    },
+    {
+      "epoch": 0.04561187600953773,
+      "learning_rate": 0.002984626468848828,
+      "loss": 1.1746,
+      "step": 593
+    },
+    {
+      "epoch": 0.04568879316975617,
+      "learning_rate": 0.0029845746642348805,
+      "loss": 1.044,
+      "step": 594
+    },
+    {
+      "epoch": 0.04576571032997462,
+      "learning_rate": 0.0029845227729350717,
+      "loss": 0.9655,
+      "step": 595
+    },
+    {
+      "epoch": 0.045842627490193064,
+      "learning_rate": 0.0029844707949524306,
+      "loss": 1.1739,
+      "step": 596
+    },
+    {
+      "epoch": 0.04591954465041151,
+      "learning_rate": 0.0029844187302899934,
+      "loss": 1.2033,
+      "step": 597
+    },
+    {
+      "epoch": 0.045996461810629954,
+      "learning_rate": 0.002984366578950799,
+      "loss": 1.0512,
+      "step": 598
+    },
+    {
+      "epoch": 0.04607337897084839,
+      "learning_rate": 0.002984314340937893,
+      "loss": 1.1152,
+      "step": 599
+    },
+    {
+      "epoch": 0.04615029613106684,
+      "learning_rate": 0.002984262016254326,
+      "loss": 1.351,
+      "step": 600
+    },
+    {
+      "epoch": 0.046227213291285284,
+      "learning_rate": 0.0029842096049031526,
+      "loss": 1.1306,
+      "step": 601
+    },
+    {
+      "epoch": 0.04630413045150373,
+      "learning_rate": 0.002984157106887434,
+      "loss": 1.548,
+      "step": 602
+    },
+    {
+      "epoch": 0.046381047611722175,
+      "learning_rate": 0.002984104522210235,
+      "loss": 1.4748,
+      "step": 603
+    },
+    {
+      "epoch": 0.04645796477194062,
+      "learning_rate": 0.0029840518508746266,
+      "loss": 1.2217,
+      "step": 604
+    },
+    {
+      "epoch": 0.046534881932159065,
+      "learning_rate": 0.0029839990928836836,
+      "loss": 1.1253,
+      "step": 605
+    },
+    {
+      "epoch": 0.04661179909237751,
+      "learning_rate": 0.002983946248240487,
+      "loss": 0.9298,
+      "step": 606
+    },
+    {
+      "epoch": 0.046688716252595956,
+      "learning_rate": 0.0029838933169481227,
+      "loss": 0.8226,
+      "step": 607
+    },
+    {
+      "epoch": 0.0467656334128144,
+      "learning_rate": 0.0029838402990096805,
+      "loss": 0.8816,
+      "step": 608
+    },
+    {
+      "epoch": 0.04684255057303285,
+      "learning_rate": 0.0029837871944282577,
+      "loss": 1.0678,
+      "step": 609
+    },
+    {
+      "epoch": 0.046919467733251286,
+      "learning_rate": 0.0029837340032069534,
+      "loss": 0.8852,
+      "step": 610
+    },
+    {
+      "epoch": 0.04699638489346973,
+      "learning_rate": 0.0029836807253488746,
+      "loss": 1.4652,
+      "step": 611
+    },
+    {
+      "epoch": 0.047073302053688176,
+      "learning_rate": 0.0029836273608571324,
+      "loss": 1.445,
+      "step": 612
+    },
+    {
+      "epoch": 0.04715021921390662,
+      "learning_rate": 0.0029835739097348424,
+      "loss": 1.1648,
+      "step": 613
+    },
+    {
+      "epoch": 0.04722713637412507,
+      "learning_rate": 0.002983520371985125,
+      "loss": 1.0133,
+      "step": 614
+    },
+    {
+      "epoch": 0.04730405353434351,
+      "learning_rate": 0.0029834667476111073,
+      "loss": 1.1974,
+      "step": 615
+    },
+    {
+      "epoch": 0.04738097069456196,
+      "learning_rate": 0.0029834130366159202,
+      "loss": 1.2651,
+      "step": 616
+    },
+    {
+      "epoch": 0.0474578878547804,
+      "learning_rate": 0.0029833592390027,
+      "loss": 0.8177,
+      "step": 617
+    },
+    {
+      "epoch": 0.04753480501499885,
+      "learning_rate": 0.0029833053547745885,
+      "loss": 1.5305,
+      "step": 618
+    },
+    {
+      "epoch": 0.047611722175217294,
+      "learning_rate": 0.0029832513839347305,
+      "loss": 1.1941,
+      "step": 619
+    },
+    {
+      "epoch": 0.04768863933543573,
+      "learning_rate": 0.002983197326486279,
+      "loss": 1.0049,
+      "step": 620
+    },
+    {
+      "epoch": 0.04776555649565418,
+      "learning_rate": 0.0029831431824323892,
+      "loss": 1.4915,
+      "step": 621
+    },
+    {
+      "epoch": 0.047842473655872624,
+      "learning_rate": 0.002983088951776224,
+      "loss": 0.9474,
+      "step": 622
+    },
+    {
+      "epoch": 0.04791939081609107,
+      "learning_rate": 0.0029830346345209487,
+      "loss": 1.2237,
+      "step": 623
+    },
+    {
+      "epoch": 0.047996307976309514,
+      "learning_rate": 0.0029829802306697356,
+      "loss": 1.4277,
+      "step": 624
+    },
+    {
+      "epoch": 0.04807322513652796,
+      "learning_rate": 0.0029829257402257613,
+      "loss": 1.0073,
+      "step": 625
+    },
+    {
+      "epoch": 0.048150142296746405,
+      "learning_rate": 0.0029828711631922082,
+      "loss": 0.9298,
+      "step": 626
+    },
+    {
+      "epoch": 0.04822705945696485,
+      "learning_rate": 0.0029828164995722617,
+      "loss": 1.5419,
+      "step": 627
+    },
+    {
+      "epoch": 0.048303976617183296,
+      "learning_rate": 0.002982761749369115,
+      "loss": 1.1724,
+      "step": 628
+    },
+    {
+      "epoch": 0.04838089377740174,
+      "learning_rate": 0.002982706912585964,
+      "loss": 1.0517,
+      "step": 629
+    },
+    {
+      "epoch": 0.04845781093762018,
+      "learning_rate": 0.0029826519892260117,
+      "loss": 1.3097,
+      "step": 630
+    },
+    {
+      "epoch": 0.048534728097838625,
+      "learning_rate": 0.002982596979292464,
+      "loss": 1.1801,
+      "step": 631
+    },
+    {
+      "epoch": 0.04861164525805707,
+      "learning_rate": 0.0029825418827885337,
+      "loss": 1.2174,
+      "step": 632
+    },
+    {
+      "epoch": 0.048688562418275516,
+      "learning_rate": 0.002982486699717438,
+      "loss": 0.9513,
+      "step": 633
+    },
+    {
+      "epoch": 0.04876547957849396,
+      "learning_rate": 0.0029824314300823987,
+      "loss": 1.3247,
+      "step": 634
+    },
+    {
+      "epoch": 0.04884239673871241,
+      "learning_rate": 0.002982376073886643,
+      "loss": 1.0623,
+      "step": 635
+    },
+    {
+      "epoch": 0.04891931389893085,
+      "learning_rate": 0.002982320631133404,
+      "loss": 1.3911,
+      "step": 636
+    },
+    {
+      "epoch": 0.0489962310591493,
+      "learning_rate": 0.0029822651018259184,
+      "loss": 1.2937,
+      "step": 637
+    },
+    {
+      "epoch": 0.04907314821936774,
+      "learning_rate": 0.002982209485967428,
+      "loss": 1.1114,
+      "step": 638
+    },
+    {
+      "epoch": 0.04915006537958619,
+      "learning_rate": 0.002982153783561182,
+      "loss": 1.2588,
+      "step": 639
+    },
+    {
+      "epoch": 0.04922698253980463,
+      "learning_rate": 0.0029820979946104313,
+      "loss": 1.2962,
+      "step": 640
+    },
+    {
+      "epoch": 0.04930389970002307,
+      "learning_rate": 0.0029820421191184346,
+      "loss": 1.272,
+      "step": 641
+    },
+    {
+      "epoch": 0.04938081686024152,
+      "learning_rate": 0.002981986157088453,
+      "loss": 1.178,
+      "step": 642
+    },
+    {
+      "epoch": 0.04945773402045996,
+      "learning_rate": 0.002981930108523756,
+      "loss": 1.3278,
+      "step": 643
+    },
+    {
+      "epoch": 0.04953465118067841,
+      "learning_rate": 0.0029818739734276153,
+      "loss": 0.9838,
+      "step": 644
+    },
+    {
+      "epoch": 0.049611568340896854,
+      "learning_rate": 0.0029818177518033096,
+      "loss": 1.325,
+      "step": 645
+    },
+    {
+      "epoch": 0.0496884855011153,
+      "learning_rate": 0.0029817614436541203,
+      "loss": 1.1888,
+      "step": 646
+    },
+    {
+      "epoch": 0.049765402661333745,
+      "learning_rate": 0.0029817050489833363,
+      "loss": 1.5704,
+      "step": 647
+    },
+    {
+      "epoch": 0.04984231982155219,
+      "learning_rate": 0.00298164856779425,
+      "loss": 0.7675,
+      "step": 648
+    },
+    {
+      "epoch": 0.049919236981770636,
+      "learning_rate": 0.00298159200009016,
+      "loss": 1.1223,
+      "step": 649
+    },
+    {
+      "epoch": 0.04999615414198908,
+      "learning_rate": 0.0029815353458743686,
+      "loss": 1.1525,
+      "step": 650
+    },
+    {
+      "epoch": 0.05007307130220752,
+      "learning_rate": 0.002981478605150185,
+      "loss": 1.3927,
+      "step": 651
+    },
+    {
+      "epoch": 0.050149988462425965,
+      "learning_rate": 0.0029814217779209214,
+      "loss": 1.1732,
+      "step": 652
+    },
+    {
+      "epoch": 0.05022690562264441,
+      "learning_rate": 0.0029813648641898965,
+      "loss": 1.3517,
+      "step": 653
+    },
+    {
+      "epoch": 0.050303822782862856,
+      "learning_rate": 0.002981307863960433,
+      "loss": 0.8532,
+      "step": 654
+    },
+    {
+      "epoch": 0.0503807399430813,
+      "learning_rate": 0.0029812507772358594,
+      "loss": 1.2538,
+      "step": 655
+    },
+    {
+      "epoch": 0.05045765710329975,
+      "learning_rate": 0.0029811936040195094,
+      "loss": 1.4504,
+      "step": 656
+    },
+    {
+      "epoch": 0.05053457426351819,
+      "learning_rate": 0.0029811363443147214,
+      "loss": 1.0522,
+      "step": 657
+    },
+    {
+      "epoch": 0.05061149142373664,
+      "learning_rate": 0.0029810789981248387,
+      "loss": 1.2034,
+      "step": 658
+    },
+    {
+      "epoch": 0.05068840858395508,
+      "learning_rate": 0.0029810215654532094,
+      "loss": 1.0663,
+      "step": 659
+    },
+    {
+      "epoch": 0.05076532574417353,
+      "learning_rate": 0.002980964046303188,
+      "loss": 1.4473,
+      "step": 660
+    },
+    {
+      "epoch": 0.05084224290439197,
+      "learning_rate": 0.0029809064406781318,
+      "loss": 1.0033,
+      "step": 661
+    },
+    {
+      "epoch": 0.05091916006461041,
+      "learning_rate": 0.002980848748581405,
+      "loss": 1.2814,
+      "step": 662
+    },
+    {
+      "epoch": 0.05099607722482886,
+      "learning_rate": 0.002980790970016377,
+      "loss": 1.2509,
+      "step": 663
+    },
+    {
+      "epoch": 0.0510729943850473,
+      "learning_rate": 0.002980733104986421,
+      "loss": 1.04,
+      "step": 664
+    },
+    {
+      "epoch": 0.05114991154526575,
+      "learning_rate": 0.002980675153494916,
+      "loss": 0.9383,
+      "step": 665
+    },
+    {
+      "epoch": 0.051226828705484194,
+      "learning_rate": 0.0029806171155452453,
+      "loss": 1.2609,
+      "step": 666
+    },
+    {
+      "epoch": 0.05130374586570264,
+      "learning_rate": 0.0029805589911407982,
+      "loss": 1.2434,
+      "step": 667
+    },
+    {
+      "epoch": 0.051380663025921085,
+      "learning_rate": 0.0029805007802849687,
+      "loss": 1.0247,
+      "step": 668
+    },
+    {
+      "epoch": 0.05145758018613953,
+      "learning_rate": 0.0029804424829811558,
+      "loss": 1.3384,
+      "step": 669
+    },
+    {
+      "epoch": 0.051534497346357976,
+      "learning_rate": 0.0029803840992327635,
+      "loss": 0.8823,
+      "step": 670
+    },
+    {
+      "epoch": 0.051611414506576414,
+      "learning_rate": 0.0029803256290432,
+      "loss": 0.763,
+      "step": 671
+    },
+    {
+      "epoch": 0.05168833166679486,
+      "learning_rate": 0.0029802670724158814,
+      "loss": 1.2358,
+      "step": 672
+    },
+    {
+      "epoch": 0.051765248827013305,
+      "learning_rate": 0.0029802084293542247,
+      "loss": 1.425,
+      "step": 673
+    },
+    {
+      "epoch": 0.05184216598723175,
+      "learning_rate": 0.002980149699861656,
+      "loss": 1.3391,
+      "step": 674
+    },
+    {
+      "epoch": 0.051919083147450196,
+      "learning_rate": 0.0029800908839416034,
+      "loss": 1.4196,
+      "step": 675
+    },
+    {
+      "epoch": 0.05199600030766864,
+      "learning_rate": 0.002980031981597502,
+      "loss": 1.3749,
+      "step": 676
+    },
+    {
+      "epoch": 0.052072917467887087,
+      "learning_rate": 0.0029799729928327904,
+      "loss": 1.0857,
+      "step": 677
+    },
+    {
+      "epoch": 0.05214983462810553,
+      "learning_rate": 0.002979913917650913,
+      "loss": 0.8182,
+      "step": 678
+    },
+    {
+      "epoch": 0.05222675178832398,
+      "learning_rate": 0.00297985475605532,
+      "loss": 1.1418,
+      "step": 679
+    },
+    {
+      "epoch": 0.05230366894854242,
+      "learning_rate": 0.0029797955080494653,
+      "loss": 1.7806,
+      "step": 680
+    },
+    {
+      "epoch": 0.05238058610876086,
+      "learning_rate": 0.002979736173636809,
+      "loss": 0.9753,
+      "step": 681
+    },
+    {
+      "epoch": 0.05245750326897931,
+      "learning_rate": 0.0029796767528208153,
+      "loss": 1.2917,
+      "step": 682
+    },
+    {
+      "epoch": 0.05253442042919775,
+      "learning_rate": 0.002979617245604954,
+      "loss": 0.8755,
+      "step": 683
+    },
+    {
+      "epoch": 0.0526113375894162,
+      "learning_rate": 0.0029795576519926993,
+      "loss": 1.1086,
+      "step": 684
+    },
+    {
+      "epoch": 0.05268825474963464,
+      "learning_rate": 0.0029794979719875315,
+      "loss": 0.9721,
+      "step": 685
+    },
+    {
+      "epoch": 0.05276517190985309,
+      "learning_rate": 0.0029794382055929355,
+      "loss": 0.8475,
+      "step": 686
+    },
+    {
+      "epoch": 0.052842089070071534,
+      "learning_rate": 0.0029793783528124004,
+      "loss": 1.1931,
+      "step": 687
+    },
+    {
+      "epoch": 0.05291900623028998,
+      "learning_rate": 0.0029793184136494223,
+      "loss": 1.8459,
+      "step": 688
+    },
+    {
+      "epoch": 0.052995923390508425,
+      "learning_rate": 0.0029792583881074995,
+      "loss": 1.4024,
+      "step": 689
+    },
+    {
+      "epoch": 0.05307284055072687,
+      "learning_rate": 0.002979198276190138,
+      "loss": 1.2084,
+      "step": 690
+    },
+    {
+      "epoch": 0.05314975771094531,
+      "learning_rate": 0.0029791380779008478,
+      "loss": 1.2549,
+      "step": 691
+    },
+    {
+      "epoch": 0.053226674871163754,
+      "learning_rate": 0.0029790777932431436,
+      "loss": 1.1673,
+      "step": 692
+    },
+    {
+      "epoch": 0.0533035920313822,
+      "learning_rate": 0.0029790174222205455,
+      "loss": 0.9272,
+      "step": 693
+    },
+    {
+      "epoch": 0.053380509191600645,
+      "learning_rate": 0.0029789569648365787,
+      "loss": 0.9013,
+      "step": 694
+    },
+    {
+      "epoch": 0.05345742635181909,
+      "learning_rate": 0.0029788964210947736,
+      "loss": 0.8581,
+      "step": 695
+    },
+    {
+      "epoch": 0.053534343512037535,
+      "learning_rate": 0.002978835790998665,
+      "loss": 0.8097,
+      "step": 696
+    },
+    {
+      "epoch": 0.05361126067225598,
+      "learning_rate": 0.0029787750745517935,
+      "loss": 1.4727,
+      "step": 697
+    },
+    {
+      "epoch": 0.053688177832474426,
+      "learning_rate": 0.002978714271757704,
+      "loss": 1.0355,
+      "step": 698
+    },
+    {
+      "epoch": 0.05376509499269287,
+      "learning_rate": 0.0029786533826199474,
+      "loss": 0.8615,
+      "step": 699
+    },
+    {
+      "epoch": 0.05384201215291132,
+      "learning_rate": 0.0029785924071420787,
+      "loss": 1.2511,
+      "step": 700
+    },
+    {
+      "epoch": 0.05391892931312976,
+      "learning_rate": 0.002978531345327658,
+      "loss": 1.4,
+      "step": 701
+    },
+    {
+      "epoch": 0.0539958464733482,
+      "learning_rate": 0.0029784701971802517,
+      "loss": 0.9655,
+      "step": 702
+    },
+    {
+      "epoch": 0.054072763633566646,
+      "learning_rate": 0.0029784089627034297,
+      "loss": 1.2346,
+      "step": 703
+    },
+    {
+      "epoch": 0.05414968079378509,
+      "learning_rate": 0.002978347641900767,
+      "loss": 1.3287,
+      "step": 704
+    },
+    {
+      "epoch": 0.05422659795400354,
+      "learning_rate": 0.002978286234775845,
+      "loss": 1.1252,
+      "step": 705
+    },
+    {
+      "epoch": 0.05430351511422198,
+      "learning_rate": 0.00297822474133225,
+      "loss": 0.9429,
+      "step": 706
+    },
+    {
+      "epoch": 0.05438043227444043,
+      "learning_rate": 0.002978163161573571,
+      "loss": 1.5489,
+      "step": 707
+    },
+    {
+      "epoch": 0.054457349434658874,
+      "learning_rate": 0.0029781014955034047,
+      "loss": 1.1839,
+      "step": 708
+    },
+    {
+      "epoch": 0.05453426659487732,
+      "learning_rate": 0.0029780397431253517,
+      "loss": 1.1866,
+      "step": 709
+    },
+    {
+      "epoch": 0.054611183755095764,
+      "learning_rate": 0.0029779779044430176,
+      "loss": 1.4301,
+      "step": 710
+    },
+    {
+      "epoch": 0.05468810091531421,
+      "learning_rate": 0.002977915979460013,
+      "loss": 1.018,
+      "step": 711
+    },
+    {
+      "epoch": 0.05476501807553265,
+      "learning_rate": 0.002977853968179955,
+      "loss": 1.1193,
+      "step": 712
+    },
+    {
+      "epoch": 0.054841935235751094,
+      "learning_rate": 0.002977791870606463,
+      "loss": 1.4326,
+      "step": 713
+    },
+    {
+      "epoch": 0.05491885239596954,
+      "learning_rate": 0.0029777296867431638,
+      "loss": 1.0494,
+      "step": 714
+    },
+    {
+      "epoch": 0.054995769556187984,
+      "learning_rate": 0.0029776674165936873,
+      "loss": 0.8348,
+      "step": 715
+    },
+    {
+      "epoch": 0.05507268671640643,
+      "learning_rate": 0.0029776050601616716,
+      "loss": 1.4864,
+      "step": 716
+    },
+    {
+      "epoch": 0.055149603876624875,
+      "learning_rate": 0.0029775426174507556,
+      "loss": 1.1649,
+      "step": 717
+    },
+    {
+      "epoch": 0.05522652103684332,
+      "learning_rate": 0.0029774800884645866,
+      "loss": 1.0731,
+      "step": 718
+    },
+    {
+      "epoch": 0.055303438197061766,
+      "learning_rate": 0.0029774174732068157,
+      "loss": 1.1337,
+      "step": 719
+    },
+    {
+      "epoch": 0.05538035535728021,
+      "learning_rate": 0.0029773547716810986,
+      "loss": 1.435,
+      "step": 720
+    },
+    {
+      "epoch": 0.05545727251749866,
+      "learning_rate": 0.0029772919838910966,
+      "loss": 1.1594,
+      "step": 721
+    },
+    {
+      "epoch": 0.055534189677717095,
+      "learning_rate": 0.002977229109840476,
+      "loss": 0.7081,
+      "step": 722
+    },
+    {
+      "epoch": 0.05561110683793554,
+      "learning_rate": 0.0029771661495329085,
+      "loss": 0.9894,
+      "step": 723
+    },
+    {
+      "epoch": 0.055688023998153986,
+      "learning_rate": 0.0029771031029720697,
+      "loss": 1.0422,
+      "step": 724
+    },
+    {
+      "epoch": 0.05576494115837243,
+      "learning_rate": 0.002977039970161642,
+      "loss": 1.4302,
+      "step": 725
+    },
+    {
+      "epoch": 0.05584185831859088,
+      "learning_rate": 0.0029769767511053104,
+      "loss": 1.08,
+      "step": 726
+    },
+    {
+      "epoch": 0.05591877547880932,
+      "learning_rate": 0.0029769134458067672,
+      "loss": 1.3204,
+      "step": 727
+    },
+    {
+      "epoch": 0.05599569263902777,
+      "learning_rate": 0.002976850054269709,
+      "loss": 1.1759,
+      "step": 728
+    },
+    {
+      "epoch": 0.05607260979924621,
+      "learning_rate": 0.002976786576497837,
+      "loss": 1.091,
+      "step": 729
+    },
+    {
+      "epoch": 0.05614952695946466,
+      "learning_rate": 0.002976723012494857,
+      "loss": 1.3387,
+      "step": 730
+    },
+    {
+      "epoch": 0.056226444119683104,
+      "learning_rate": 0.002976659362264482,
+      "loss": 1.2249,
+      "step": 731
+    },
+    {
+      "epoch": 0.05630336127990154,
+      "learning_rate": 0.0029765956258104276,
+      "loss": 1.3977,
+      "step": 732
+    },
+    {
+      "epoch": 0.05638027844011999,
+      "learning_rate": 0.0029765318031364156,
+      "loss": 1.3686,
+      "step": 733
+    },
+    {
+      "epoch": 0.05645719560033843,
+      "learning_rate": 0.002976467894246173,
+      "loss": 1.0985,
+      "step": 734
+    },
+    {
+      "epoch": 0.05653411276055688,
+      "learning_rate": 0.0029764038991434314,
+      "loss": 0.868,
+      "step": 735
+    },
+    {
+      "epoch": 0.056611029920775324,
+      "learning_rate": 0.0029763398178319278,
+      "loss": 1.3035,
+      "step": 736
+    },
+    {
+      "epoch": 0.05668794708099377,
+      "learning_rate": 0.002976275650315403,
+      "loss": 1.1757,
+      "step": 737
+    },
+    {
+      "epoch": 0.056764864241212215,
+      "learning_rate": 0.0029762113965976045,
+      "loss": 0.9565,
+      "step": 738
+    },
+    {
+      "epoch": 0.05684178140143066,
+      "learning_rate": 0.0029761470566822845,
+      "loss": 1.1045,
+      "step": 739
+    },
+    {
+      "epoch": 0.056918698561649106,
+      "learning_rate": 0.0029760826305731992,
+      "loss": 1.0485,
+      "step": 740
+    },
+    {
+      "epoch": 0.05699561572186755,
+      "learning_rate": 0.0029760181182741105,
+      "loss": 0.9465,
+      "step": 741
+    },
+    {
+      "epoch": 0.057072532882086,
+      "learning_rate": 0.002975953519788786,
+      "loss": 1.6061,
+      "step": 742
+    },
+    {
+      "epoch": 0.057149450042304435,
+      "learning_rate": 0.0029758888351209974,
+      "loss": 1.2262,
+      "step": 743
+    },
+    {
+      "epoch": 0.05722636720252288,
+      "learning_rate": 0.002975824064274521,
+      "loss": 1.2773,
+      "step": 744
+    },
+    {
+      "epoch": 0.057303284362741326,
+      "learning_rate": 0.00297575920725314,
+      "loss": 1.3701,
+      "step": 745
+    },
+    {
+      "epoch": 0.05738020152295977,
+      "learning_rate": 0.0029756942640606406,
+      "loss": 1.3194,
+      "step": 746
+    },
+    {
+      "epoch": 0.05745711868317822,
+      "learning_rate": 0.0029756292347008152,
+      "loss": 1.332,
+      "step": 747
+    },
+    {
+      "epoch": 0.05753403584339666,
+      "learning_rate": 0.002975564119177461,
+      "loss": 1.3512,
+      "step": 748
+    },
+    {
+      "epoch": 0.05761095300361511,
+      "learning_rate": 0.00297549891749438,
+      "loss": 1.2467,
+      "step": 749
+    },
+    {
+      "epoch": 0.05768787016383355,
+      "learning_rate": 0.0029754336296553795,
+      "loss": 1.335,
+      "step": 750
+    },
+    {
+      "epoch": 0.057764787324052,
+      "learning_rate": 0.002975368255664272,
+      "loss": 1.0811,
+      "step": 751
+    },
+    {
+      "epoch": 0.057841704484270444,
+      "learning_rate": 0.0029753027955248743,
+      "loss": 1.0731,
+      "step": 752
+    },
+    {
+      "epoch": 0.05791862164448888,
+      "learning_rate": 0.0029752372492410084,
+      "loss": 0.8023,
+      "step": 753
+    },
+    {
+      "epoch": 0.05799553880470733,
+      "learning_rate": 0.002975171616816503,
+      "loss": 1.5936,
+      "step": 754
+    },
+    {
+      "epoch": 0.05807245596492577,
+      "learning_rate": 0.0029751058982551886,
+      "loss": 1.0659,
+      "step": 755
+    },
+    {
+      "epoch": 0.05814937312514422,
+      "learning_rate": 0.0029750400935609034,
+      "loss": 1.4928,
+      "step": 756
+    },
+    {
+      "epoch": 0.058226290285362664,
+      "learning_rate": 0.0029749742027374904,
+      "loss": 1.2801,
+      "step": 757
+    },
+    {
+      "epoch": 0.05830320744558111,
+      "learning_rate": 0.002974908225788796,
+      "loss": 1.7553,
+      "step": 758
+    },
+    {
+      "epoch": 0.058380124605799555,
+      "learning_rate": 0.0029748421627186737,
+      "loss": 1.2631,
+      "step": 759
+    },
+    {
+      "epoch": 0.058457041766018,
+      "learning_rate": 0.0029747760135309803,
+      "loss": 1.1292,
+      "step": 760
+    },
+    {
+      "epoch": 0.058533958926236446,
+      "learning_rate": 0.0029747097782295787,
+      "loss": 1.4035,
+      "step": 761
+    },
+    {
+      "epoch": 0.05861087608645489,
+      "learning_rate": 0.002974643456818336,
+      "loss": 0.9446,
+      "step": 762
+    },
+    {
+      "epoch": 0.05868779324667333,
+      "learning_rate": 0.0029745770493011256,
+      "loss": 0.7152,
+      "step": 763
+    },
+    {
+      "epoch": 0.058764710406891775,
+      "learning_rate": 0.002974510555681824,
+      "loss": 1.262,
+      "step": 764
+    },
+    {
+      "epoch": 0.05884162756711022,
+      "learning_rate": 0.0029744439759643144,
+      "loss": 1.2302,
+      "step": 765
+    },
+    {
+      "epoch": 0.058918544727328666,
+      "learning_rate": 0.002974377310152484,
+      "loss": 1.1512,
+      "step": 766
+    },
+    {
+      "epoch": 0.05899546188754711,
+      "learning_rate": 0.0029743105582502264,
+      "loss": 1.1798,
+      "step": 767
+    },
+    {
+      "epoch": 0.05907237904776556,
+      "learning_rate": 0.0029742437202614387,
+      "loss": 1.1354,
+      "step": 768
+    },
+    {
+      "epoch": 0.059149296207984,
+      "learning_rate": 0.0029741767961900243,
+      "loss": 1.3438,
+      "step": 769
+    },
+    {
+      "epoch": 0.05922621336820245,
+      "learning_rate": 0.0029741097860398894,
+      "loss": 1.4128,
+      "step": 770
+    },
+    {
+      "epoch": 0.05930313052842089,
+      "learning_rate": 0.0029740426898149487,
+      "loss": 1.2916,
+      "step": 771
+    },
+    {
+      "epoch": 0.05938004768863934,
+      "learning_rate": 0.0029739755075191185,
+      "loss": 0.9568,
+      "step": 772
+    },
+    {
+      "epoch": 0.05945696484885778,
+      "learning_rate": 0.0029739082391563223,
+      "loss": 1.5856,
+      "step": 773
+    },
+    {
+      "epoch": 0.05953388200907622,
+      "learning_rate": 0.0029738408847304883,
+      "loss": 1.0162,
+      "step": 774
+    },
+    {
+      "epoch": 0.05961079916929467,
+      "learning_rate": 0.002973773444245549,
+      "loss": 1.1085,
+      "step": 775
+    },
+    {
+      "epoch": 0.05968771632951311,
+      "learning_rate": 0.002973705917705442,
+      "loss": 1.1943,
+      "step": 776
+    },
+    {
+      "epoch": 0.05976463348973156,
+      "learning_rate": 0.002973638305114111,
+      "loss": 1.1486,
+      "step": 777
+    },
+    {
+      "epoch": 0.059841550649950004,
+      "learning_rate": 0.0029735706064755035,
+      "loss": 1.1634,
+      "step": 778
+    },
+    {
+      "epoch": 0.05991846781016845,
+      "learning_rate": 0.0029735028217935723,
+      "loss": 1.5149,
+      "step": 779
+    },
+    {
+      "epoch": 0.059995384970386895,
+      "learning_rate": 0.002973434951072276,
+      "loss": 0.8127,
+      "step": 780
+    },
+    {
+      "epoch": 0.06007230213060534,
+      "learning_rate": 0.0029733669943155776,
+      "loss": 0.9215,
+      "step": 781
+    },
+    {
+      "epoch": 0.060149219290823785,
+      "learning_rate": 0.002973298951527444,
+      "loss": 1.2661,
+      "step": 782
+    },
+    {
+      "epoch": 0.06022613645104223,
+      "learning_rate": 0.00297323082271185,
+      "loss": 1.2779,
+      "step": 783
+    },
+    {
+      "epoch": 0.06030305361126067,
+      "learning_rate": 0.0029731626078727726,
+      "loss": 1.2186,
+      "step": 784
+    },
+    {
+      "epoch": 0.060379970771479115,
+      "learning_rate": 0.0029730943070141952,
+      "loss": 1.2627,
+      "step": 785
+    },
+    {
+      "epoch": 0.06045688793169756,
+      "learning_rate": 0.002973025920140106,
+      "loss": 1.4769,
+      "step": 786
+    },
+    {
+      "epoch": 0.060533805091916006,
+      "learning_rate": 0.0029729574472544987,
+      "loss": 1.2963,
+      "step": 787
+    },
+    {
+      "epoch": 0.06061072225213445,
+      "learning_rate": 0.00297288888836137,
+      "loss": 1.6462,
+      "step": 788
+    },
+    {
+      "epoch": 0.060687639412352896,
+      "learning_rate": 0.002972820243464725,
+      "loss": 0.9503,
+      "step": 789
+    },
+    {
+      "epoch": 0.06076455657257134,
+      "learning_rate": 0.0029727515125685705,
+      "loss": 1.1223,
+      "step": 790
+    },
+    {
+      "epoch": 0.06084147373278979,
+      "learning_rate": 0.00297268269567692,
+      "loss": 0.9932,
+      "step": 791
+    },
+    {
+      "epoch": 0.06091839089300823,
+      "learning_rate": 0.0029726137927937927,
+      "loss": 1.1108,
+      "step": 792
+    },
+    {
+      "epoch": 0.06099530805322668,
+      "learning_rate": 0.002972544803923211,
+      "loss": 1.1645,
+      "step": 793
+    },
+    {
+      "epoch": 0.06107222521344512,
+      "learning_rate": 0.002972475729069204,
+      "loss": 1.0648,
+      "step": 794
+    },
+    {
+      "epoch": 0.06114914237366356,
+      "learning_rate": 0.0029724065682358043,
+      "loss": 1.1004,
+      "step": 795
+    },
+    {
+      "epoch": 0.06122605953388201,
+      "learning_rate": 0.0029723373214270506,
+      "loss": 1.2627,
+      "step": 796
+    },
+    {
+      "epoch": 0.06130297669410045,
+      "learning_rate": 0.002972267988646986,
+      "loss": 1.2579,
+      "step": 797
+    },
+    {
+      "epoch": 0.0613798938543189,
+      "learning_rate": 0.0029721985698996597,
+      "loss": 1.2242,
+      "step": 798
+    },
+    {
+      "epoch": 0.061456811014537344,
+      "learning_rate": 0.0029721290651891245,
+      "loss": 1.0346,
+      "step": 799
+    },
+    {
+      "epoch": 0.06153372817475579,
+      "learning_rate": 0.002972059474519439,
+      "loss": 1.258,
+      "step": 800
+    },
+    {
+      "epoch": 0.061610645334974234,
+      "learning_rate": 0.002971989797894666,
+      "loss": 1.3715,
+      "step": 801
+    },
+    {
+      "epoch": 0.06168756249519268,
+      "learning_rate": 0.0029719200353188753,
+      "loss": 1.5083,
+      "step": 802
+    },
+    {
+      "epoch": 0.061764479655411125,
+      "learning_rate": 0.0029718501867961396,
+      "loss": 0.7944,
+      "step": 803
+    },
+    {
+      "epoch": 0.061841396815629564,
+      "learning_rate": 0.0029717802523305374,
+      "loss": 1.3351,
+      "step": 804
+    },
+    {
+      "epoch": 0.06191831397584801,
+      "learning_rate": 0.0029717102319261526,
+      "loss": 1.1212,
+      "step": 805
+    },
+    {
+      "epoch": 0.061995231136066455,
+      "learning_rate": 0.0029716401255870735,
+      "loss": 1.1144,
+      "step": 806
+    },
+    {
+      "epoch": 0.0620721482962849,
+      "learning_rate": 0.0029715699333173937,
+      "loss": 1.0605,
+      "step": 807
+    },
+    {
+      "epoch": 0.062149065456503345,
+      "learning_rate": 0.002971499655121212,
+      "loss": 1.3306,
+      "step": 808
+    },
+    {
+      "epoch": 0.06222598261672179,
+      "learning_rate": 0.002971429291002632,
+      "loss": 0.7732,
+      "step": 809
+    },
+    {
+      "epoch": 0.062302899776940236,
+      "learning_rate": 0.0029713588409657618,
+      "loss": 1.21,
+      "step": 810
+    },
+    {
+      "epoch": 0.06237981693715868,
+      "learning_rate": 0.0029712883050147156,
+      "loss": 1.0372,
+      "step": 811
+    },
+    {
+      "epoch": 0.06245673409737713,
+      "learning_rate": 0.0029712176831536118,
+      "loss": 1.3111,
+      "step": 812
+    },
+    {
+      "epoch": 0.06253365125759557,
+      "learning_rate": 0.0029711469753865745,
+      "loss": 0.9122,
+      "step": 813
+    },
+    {
+      "epoch": 0.06261056841781401,
+      "learning_rate": 0.002971076181717732,
+      "loss": 1.1221,
+      "step": 814
+    },
+    {
+      "epoch": 0.06268748557803246,
+      "learning_rate": 0.002971005302151218,
+      "loss": 1.3591,
+      "step": 815
+    },
+    {
+      "epoch": 0.0627644027382509,
+      "learning_rate": 0.0029709343366911716,
+      "loss": 1.5147,
+      "step": 816
+    },
+    {
+      "epoch": 0.06284131989846935,
+      "learning_rate": 0.002970863285341736,
+      "loss": 1.1363,
+      "step": 817
+    },
+    {
+      "epoch": 0.06291823705868779,
+      "learning_rate": 0.0029707921481070608,
+      "loss": 0.8343,
+      "step": 818
+    },
+    {
+      "epoch": 0.06299515421890624,
+      "learning_rate": 0.002970720924991299,
+      "loss": 1.0197,
+      "step": 819
+    },
+    {
+      "epoch": 0.06307207137912468,
+      "learning_rate": 0.0029706496159986094,
+      "loss": 1.3745,
+      "step": 820
+    },
+    {
+      "epoch": 0.06314898853934313,
+      "learning_rate": 0.0029705782211331567,
+      "loss": 1.0971,
+      "step": 821
+    },
+    {
+      "epoch": 0.06322590569956157,
+      "learning_rate": 0.0029705067403991087,
+      "loss": 1.1213,
+      "step": 822
+    },
+    {
+      "epoch": 0.06330282285978002,
+      "learning_rate": 0.0029704351738006392,
+      "loss": 1.0577,
+      "step": 823
+    },
+    {
+      "epoch": 0.06337974001999847,
+      "learning_rate": 0.0029703635213419277,
+      "loss": 0.8942,
+      "step": 824
+    },
+    {
+      "epoch": 0.06345665718021691,
+      "learning_rate": 0.0029702917830271578,
+      "loss": 1.5873,
+      "step": 825
+    },
+    {
+      "epoch": 0.06353357434043536,
+      "learning_rate": 0.002970219958860519,
+      "loss": 1.5163,
+      "step": 826
+    },
+    {
+      "epoch": 0.0636104915006538,
+      "learning_rate": 0.0029701480488462037,
+      "loss": 0.8876,
+      "step": 827
+    },
+    {
+      "epoch": 0.06368740866087225,
+      "learning_rate": 0.002970076052988412,
+      "loss": 1.1676,
+      "step": 828
+    },
+    {
+      "epoch": 0.06376432582109069,
+      "learning_rate": 0.002970003971291348,
+      "loss": 1.2498,
+      "step": 829
+    },
+    {
+      "epoch": 0.06384124298130912,
+      "learning_rate": 0.002969931803759219,
+      "loss": 0.9933,
+      "step": 830
+    },
+    {
+      "epoch": 0.06391816014152757,
+      "learning_rate": 0.0029698595503962408,
+      "loss": 1.4362,
+      "step": 831
+    },
+    {
+      "epoch": 0.06399507730174601,
+      "learning_rate": 0.002969787211206632,
+      "loss": 0.9662,
+      "step": 832
+    },
+    {
+      "epoch": 0.06407199446196446,
+      "learning_rate": 0.0029697147861946155,
+      "loss": 0.9095,
+      "step": 833
+    },
+    {
+      "epoch": 0.0641489116221829,
+      "learning_rate": 0.0029696422753644207,
+      "loss": 1.1928,
+      "step": 834
+    },
+    {
+      "epoch": 0.06422582878240135,
+      "learning_rate": 0.0029695696787202823,
+      "loss": 1.0297,
+      "step": 835
+    },
+    {
+      "epoch": 0.0643027459426198,
+      "learning_rate": 0.002969496996266439,
+      "loss": 0.7538,
+      "step": 836
+    },
+    {
+      "epoch": 0.06437966310283824,
+      "learning_rate": 0.002969424228007134,
+      "loss": 1.1861,
+      "step": 837
+    },
+    {
+      "epoch": 0.06445658026305669,
+      "learning_rate": 0.0029693513739466174,
+      "loss": 1.2194,
+      "step": 838
+    },
+    {
+      "epoch": 0.06453349742327513,
+      "learning_rate": 0.0029692784340891423,
+      "loss": 1.1569,
+      "step": 839
+    },
+    {
+      "epoch": 0.06461041458349358,
+      "learning_rate": 0.0029692054084389688,
+      "loss": 1.3693,
+      "step": 840
+    },
+    {
+      "epoch": 0.06468733174371202,
+      "learning_rate": 0.0029691322970003594,
+      "loss": 1.275,
+      "step": 841
+    },
+    {
+      "epoch": 0.06476424890393047,
+      "learning_rate": 0.002969059099777585,
+      "loss": 0.8621,
+      "step": 842
+    },
+    {
+      "epoch": 0.06484116606414891,
+      "learning_rate": 0.0029689858167749185,
+      "loss": 1.258,
+      "step": 843
+    },
+    {
+      "epoch": 0.06491808322436736,
+      "learning_rate": 0.002968912447996639,
+      "loss": 1.0037,
+      "step": 844
+    },
+    {
+      "epoch": 0.0649950003845858,
+      "learning_rate": 0.0029688389934470314,
+      "loss": 1.148,
+      "step": 845
+    },
+    {
+      "epoch": 0.06507191754480425,
+      "learning_rate": 0.0029687654531303834,
+      "loss": 1.1333,
+      "step": 846
+    },
+    {
+      "epoch": 0.0651488347050227,
+      "learning_rate": 0.0029686918270509905,
+      "loss": 1.1125,
+      "step": 847
+    },
+    {
+      "epoch": 0.06522575186524114,
+      "learning_rate": 0.0029686181152131508,
+      "loss": 1.1169,
+      "step": 848
+    },
+    {
+      "epoch": 0.06530266902545959,
+      "learning_rate": 0.002968544317621169,
+      "loss": 1.3645,
+      "step": 849
+    },
+    {
+      "epoch": 0.06537958618567802,
+      "learning_rate": 0.0029684704342793538,
+      "loss": 1.1042,
+      "step": 850
+    },
+    {
+      "epoch": 0.06545650334589646,
+      "learning_rate": 0.0029683964651920198,
+      "loss": 1.1865,
+      "step": 851
+    },
+    {
+      "epoch": 0.06553342050611491,
+      "learning_rate": 0.0029683224103634856,
+      "loss": 1.224,
+      "step": 852
+    },
+    {
+      "epoch": 0.06561033766633335,
+      "learning_rate": 0.002968248269798076,
+      "loss": 1.6174,
+      "step": 853
+    },
+    {
+      "epoch": 0.0656872548265518,
+      "learning_rate": 0.0029681740435001194,
+      "loss": 1.5593,
+      "step": 854
+    },
+    {
+      "epoch": 0.06576417198677025,
+      "learning_rate": 0.00296809973147395,
+      "loss": 1.2401,
+      "step": 855
+    },
+    {
+      "epoch": 0.06584108914698869,
+      "learning_rate": 0.002968025333723908,
+      "loss": 1.0987,
+      "step": 856
+    },
+    {
+      "epoch": 0.06591800630720714,
+      "learning_rate": 0.0029679508502543366,
+      "loss": 1.2755,
+      "step": 857
+    },
+    {
+      "epoch": 0.06599492346742558,
+      "learning_rate": 0.002967876281069585,
+      "loss": 1.6386,
+      "step": 858
+    },
+    {
+      "epoch": 0.06607184062764403,
+      "learning_rate": 0.0029678016261740076,
+      "loss": 1.5125,
+      "step": 859
+    },
+    {
+      "epoch": 0.06614875778786247,
+      "learning_rate": 0.002967726885571964,
+      "loss": 1.4287,
+      "step": 860
+    },
+    {
+      "epoch": 0.06622567494808092,
+      "learning_rate": 0.002967652059267817,
+      "loss": 1.1767,
+      "step": 861
+    },
+    {
+      "epoch": 0.06630259210829936,
+      "learning_rate": 0.002967577147265937,
+      "loss": 1.3266,
+      "step": 862
+    },
+    {
+      "epoch": 0.06637950926851781,
+      "learning_rate": 0.0029675021495706986,
+      "loss": 1.2988,
+      "step": 863
+    },
+    {
+      "epoch": 0.06645642642873625,
+      "learning_rate": 0.00296742706618648,
+      "loss": 0.8827,
+      "step": 864
+    },
+    {
+      "epoch": 0.0665333435889547,
+      "learning_rate": 0.002967351897117665,
+      "loss": 1.3811,
+      "step": 865
+    },
+    {
+      "epoch": 0.06661026074917314,
+      "learning_rate": 0.0029672766423686444,
+      "loss": 1.1978,
+      "step": 866
+    },
+    {
+      "epoch": 0.06668717790939159,
+      "learning_rate": 0.002967201301943811,
+      "loss": 1.3499,
+      "step": 867
+    },
+    {
+      "epoch": 0.06676409506961004,
+      "learning_rate": 0.0029671258758475643,
+      "loss": 1.174,
+      "step": 868
+    },
+    {
+      "epoch": 0.06684101222982848,
+      "learning_rate": 0.0029670503640843093,
+      "loss": 1.0905,
+      "step": 869
+    },
+    {
+      "epoch": 0.06691792939004693,
+      "learning_rate": 0.002966974766658454,
+      "loss": 1.262,
+      "step": 870
+    },
+    {
+      "epoch": 0.06699484655026536,
+      "learning_rate": 0.0029668990835744137,
+      "loss": 0.9186,
+      "step": 871
+    },
+    {
+      "epoch": 0.0670717637104838,
+      "learning_rate": 0.0029668233148366072,
+      "loss": 0.9934,
+      "step": 872
+    },
+    {
+      "epoch": 0.06714868087070225,
+      "learning_rate": 0.002966747460449458,
+      "loss": 1.2823,
+      "step": 873
+    },
+    {
+      "epoch": 0.0672255980309207,
+      "learning_rate": 0.002966671520417397,
+      "loss": 1.0951,
+      "step": 874
+    },
+    {
+      "epoch": 0.06730251519113914,
+      "learning_rate": 0.0029665954947448565,
+      "loss": 1.2555,
+      "step": 875
+    },
+    {
+      "epoch": 0.06737943235135758,
+      "learning_rate": 0.002966519383436277,
+      "loss": 1.0027,
+      "step": 876
+    },
+    {
+      "epoch": 0.06745634951157603,
+      "learning_rate": 0.0029664431864961024,
+      "loss": 1.2581,
+      "step": 877
+    },
+    {
+      "epoch": 0.06753326667179448,
+      "learning_rate": 0.0029663669039287814,
+      "loss": 1.1946,
+      "step": 878
+    },
+    {
+      "epoch": 0.06761018383201292,
+      "learning_rate": 0.0029662905357387692,
+      "loss": 1.1934,
+      "step": 879
+    },
+    {
+      "epoch": 0.06768710099223137,
+      "learning_rate": 0.002966214081930524,
+      "loss": 1.2912,
+      "step": 880
+    },
+    {
+      "epoch": 0.06776401815244981,
+      "learning_rate": 0.002966137542508511,
+      "loss": 1.4963,
+      "step": 881
+    },
+    {
+      "epoch": 0.06784093531266826,
+      "learning_rate": 0.0029660609174771983,
+      "loss": 1.047,
+      "step": 882
+    },
+    {
+      "epoch": 0.0679178524728867,
+      "learning_rate": 0.0029659842068410617,
+      "loss": 1.3658,
+      "step": 883
+    },
+    {
+      "epoch": 0.06799476963310515,
+      "learning_rate": 0.0029659074106045784,
+      "loss": 1.5428,
+      "step": 884
+    },
+    {
+      "epoch": 0.0680716867933236,
+      "learning_rate": 0.0029658305287722344,
+      "loss": 1.0194,
+      "step": 885
+    },
+    {
+      "epoch": 0.06814860395354204,
+      "learning_rate": 0.002965753561348518,
+      "loss": 1.2364,
+      "step": 886
+    },
+    {
+      "epoch": 0.06822552111376048,
+      "learning_rate": 0.0029656765083379234,
+      "loss": 1.0498,
+      "step": 887
+    },
+    {
+      "epoch": 0.06830243827397893,
+      "learning_rate": 0.00296559936974495,
+      "loss": 1.0967,
+      "step": 888
+    },
+    {
+      "epoch": 0.06837935543419738,
+      "learning_rate": 0.002965522145574102,
+      "loss": 1.9147,
+      "step": 889
+    },
+    {
+      "epoch": 0.06845627259441582,
+      "learning_rate": 0.0029654448358298886,
+      "loss": 1.3101,
+      "step": 890
+    },
+    {
+      "epoch": 0.06853318975463425,
+      "learning_rate": 0.0029653674405168244,
+      "loss": 0.8805,
+      "step": 891
+    },
+    {
+      "epoch": 0.0686101069148527,
+      "learning_rate": 0.0029652899596394277,
+      "loss": 1.0904,
+      "step": 892
+    },
+    {
+      "epoch": 0.06868702407507114,
+      "learning_rate": 0.0029652123932022234,
+      "loss": 0.8722,
+      "step": 893
+    },
+    {
+      "epoch": 0.06876394123528959,
+      "learning_rate": 0.00296513474120974,
+      "loss": 1.304,
+      "step": 894
+    },
+    {
+      "epoch": 0.06884085839550803,
+      "learning_rate": 0.0029650570036665125,
+      "loss": 0.849,
+      "step": 895
+    },
+    {
+      "epoch": 0.06891777555572648,
+      "learning_rate": 0.0029649791805770797,
+      "loss": 0.9992,
+      "step": 896
+    },
+    {
+      "epoch": 0.06899469271594492,
+      "learning_rate": 0.002964901271945986,
+      "loss": 1.0795,
+      "step": 897
+    },
+    {
+      "epoch": 0.06907160987616337,
+      "learning_rate": 0.00296482327777778,
+      "loss": 1.304,
+      "step": 898
+    },
+    {
+      "epoch": 0.06914852703638182,
+      "learning_rate": 0.0029647451980770166,
+      "loss": 1.0488,
+      "step": 899
+    },
+    {
+      "epoch": 0.06922544419660026,
+      "learning_rate": 0.0029646670328482543,
+      "loss": 1.2691,
+      "step": 900
+    },
+    {
+      "epoch": 0.0693023613568187,
+      "learning_rate": 0.0029645887820960574,
+      "loss": 1.3094,
+      "step": 901
+    },
+    {
+      "epoch": 0.06937927851703715,
+      "learning_rate": 0.002964510445824995,
+      "loss": 1.2892,
+      "step": 902
+    },
+    {
+      "epoch": 0.0694561956772556,
+      "learning_rate": 0.002964432024039642,
+      "loss": 1.2423,
+      "step": 903
+    },
+    {
+      "epoch": 0.06953311283747404,
+      "learning_rate": 0.0029643535167445767,
+      "loss": 1.2492,
+      "step": 904
+    },
+    {
+      "epoch": 0.06961002999769249,
+      "learning_rate": 0.002964274923944383,
+      "loss": 1.3041,
+      "step": 905
+    },
+    {
+      "epoch": 0.06968694715791093,
+      "learning_rate": 0.002964196245643651,
+      "loss": 1.0715,
+      "step": 906
+    },
+    {
+      "epoch": 0.06976386431812938,
+      "learning_rate": 0.0029641174818469743,
+      "loss": 1.0728,
+      "step": 907
+    },
+    {
+      "epoch": 0.06984078147834782,
+      "learning_rate": 0.0029640386325589514,
+      "loss": 1.042,
+      "step": 908
+    },
+    {
+      "epoch": 0.06991769863856627,
+      "learning_rate": 0.0029639596977841876,
+      "loss": 1.4161,
+      "step": 909
+    },
+    {
+      "epoch": 0.06999461579878472,
+      "learning_rate": 0.0029638806775272907,
+      "loss": 1.4001,
+      "step": 910
+    },
+    {
+      "epoch": 0.07007153295900316,
+      "learning_rate": 0.0029638015717928755,
+      "loss": 1.355,
+      "step": 911
+    },
+    {
+      "epoch": 0.07014845011922159,
+      "learning_rate": 0.0029637223805855616,
+      "loss": 1.1305,
+      "step": 912
+    },
+    {
+      "epoch": 0.07022536727944004,
+      "learning_rate": 0.002963643103909972,
+      "loss": 1.1159,
+      "step": 913
+    },
+    {
+      "epoch": 0.07030228443965848,
+      "learning_rate": 0.002963563741770736,
+      "loss": 1.3449,
+      "step": 914
+    },
+    {
+      "epoch": 0.07037920159987693,
+      "learning_rate": 0.002963484294172488,
+      "loss": 1.338,
+      "step": 915
+    },
+    {
+      "epoch": 0.07045611876009537,
+      "learning_rate": 0.002963404761119867,
+      "loss": 1.2533,
+      "step": 916
+    },
+    {
+      "epoch": 0.07053303592031382,
+      "learning_rate": 0.002963325142617517,
+      "loss": 1.0319,
+      "step": 917
+    },
+    {
+      "epoch": 0.07060995308053226,
+      "learning_rate": 0.0029632454386700866,
+      "loss": 1.2467,
+      "step": 918
+    },
+    {
+      "epoch": 0.07068687024075071,
+      "learning_rate": 0.0029631656492822302,
+      "loss": 1.2088,
+      "step": 919
+    },
+    {
+      "epoch": 0.07076378740096916,
+      "learning_rate": 0.0029630857744586065,
+      "loss": 1.0837,
+      "step": 920
+    },
+    {
+      "epoch": 0.0708407045611876,
+      "learning_rate": 0.00296300581420388,
+      "loss": 1.2227,
+      "step": 921
+    },
+    {
+      "epoch": 0.07091762172140605,
+      "learning_rate": 0.002962925768522719,
+      "loss": 0.9956,
+      "step": 922
+    },
+    {
+      "epoch": 0.07099453888162449,
+      "learning_rate": 0.002962845637419798,
+      "loss": 1.1343,
+      "step": 923
+    },
+    {
+      "epoch": 0.07107145604184294,
+      "learning_rate": 0.0029627654208997955,
+      "loss": 1.0709,
+      "step": 924
+    },
+    {
+      "epoch": 0.07114837320206138,
+      "learning_rate": 0.0029626851189673955,
+      "loss": 1.1055,
+      "step": 925
+    },
+    {
+      "epoch": 0.07122529036227983,
+      "learning_rate": 0.002962604731627287,
+      "loss": 1.0965,
+      "step": 926
+    },
+    {
+      "epoch": 0.07130220752249827,
+      "learning_rate": 0.0029625242588841643,
+      "loss": 0.9764,
+      "step": 927
+    },
+    {
+      "epoch": 0.07137912468271672,
+      "learning_rate": 0.002962443700742726,
+      "loss": 1.503,
+      "step": 928
+    },
+    {
+      "epoch": 0.07145604184293516,
+      "learning_rate": 0.002962363057207675,
+      "loss": 1.1844,
+      "step": 929
+    },
+    {
+      "epoch": 0.07153295900315361,
+      "learning_rate": 0.002962282328283722,
+      "loss": 1.3772,
+      "step": 930
+    },
+    {
+      "epoch": 0.07160987616337205,
+      "learning_rate": 0.0029622015139755794,
+      "loss": 1.0582,
+      "step": 931
+    },
+    {
+      "epoch": 0.07168679332359049,
+      "learning_rate": 0.002962120614287967,
+      "loss": 1.126,
+      "step": 932
+    },
+    {
+      "epoch": 0.07176371048380893,
+      "learning_rate": 0.0029620396292256075,
+      "loss": 1.2031,
+      "step": 933
+    },
+    {
+      "epoch": 0.07184062764402738,
+      "learning_rate": 0.0029619585587932306,
+      "loss": 1.3518,
+      "step": 934
+    },
+    {
+      "epoch": 0.07191754480424582,
+      "learning_rate": 0.00296187740299557,
+      "loss": 0.9528,
+      "step": 935
+    },
+    {
+      "epoch": 0.07199446196446427,
+      "learning_rate": 0.0029617961618373647,
+      "loss": 1.2507,
+      "step": 936
+    },
+    {
+      "epoch": 0.07207137912468271,
+      "learning_rate": 0.002961714835323357,
+      "loss": 1.2534,
+      "step": 937
+    },
+    {
+      "epoch": 0.07214829628490116,
+      "learning_rate": 0.0029616334234582975,
+      "loss": 1.1605,
+      "step": 938
+    },
+    {
+      "epoch": 0.0722252134451196,
+      "learning_rate": 0.0029615519262469387,
+      "loss": 1.1648,
+      "step": 939
+    },
+    {
+      "epoch": 0.07230213060533805,
+      "learning_rate": 0.00296147034369404,
+      "loss": 1.0868,
+      "step": 940
+    },
+    {
+      "epoch": 0.0723790477655565,
+      "learning_rate": 0.0029613886758043653,
+      "loss": 1.1404,
+      "step": 941
+    },
+    {
+      "epoch": 0.07245596492577494,
+      "learning_rate": 0.002961306922582682,
+      "loss": 1.1598,
+      "step": 942
+    },
+    {
+      "epoch": 0.07253288208599339,
+      "learning_rate": 0.002961225084033765,
+      "loss": 0.9223,
+      "step": 943
+    },
+    {
+      "epoch": 0.07260979924621183,
+      "learning_rate": 0.002961143160162392,
+      "loss": 1.2376,
+      "step": 944
+    },
+    {
+      "epoch": 0.07268671640643028,
+      "learning_rate": 0.002961061150973347,
+      "loss": 1.1766,
+      "step": 945
+    },
+    {
+      "epoch": 0.07276363356664872,
+      "learning_rate": 0.0029609790564714194,
+      "loss": 1.0539,
+      "step": 946
+    },
+    {
+      "epoch": 0.07284055072686717,
+      "learning_rate": 0.002960896876661402,
+      "loss": 1.3398,
+      "step": 947
+    },
+    {
+      "epoch": 0.07291746788708561,
+      "learning_rate": 0.002960814611548093,
+      "loss": 1.0852,
+      "step": 948
+    },
+    {
+      "epoch": 0.07299438504730406,
+      "learning_rate": 0.0029607322611362973,
+      "loss": 1.0945,
+      "step": 949
+    },
+    {
+      "epoch": 0.0730713022075225,
+      "learning_rate": 0.0029606498254308214,
+      "loss": 1.0968,
+      "step": 950
+    },
+    {
+      "epoch": 0.07314821936774095,
+      "learning_rate": 0.0029605673044364807,
+      "loss": 0.8015,
+      "step": 951
+    },
+    {
+      "epoch": 0.0732251365279594,
+      "learning_rate": 0.0029604846981580927,
+      "loss": 1.1704,
+      "step": 952
+    },
+    {
+      "epoch": 0.07330205368817783,
+      "learning_rate": 0.002960402006600481,
+      "loss": 1.2461,
+      "step": 953
+    },
+    {
+      "epoch": 0.07337897084839627,
+      "learning_rate": 0.0029603192297684745,
+      "loss": 1.2593,
+      "step": 954
+    },
+    {
+      "epoch": 0.07345588800861472,
+      "learning_rate": 0.002960236367666906,
+      "loss": 0.9799,
+      "step": 955
+    },
+    {
+      "epoch": 0.07353280516883316,
+      "learning_rate": 0.0029601534203006143,
+      "loss": 1.098,
+      "step": 956
+    },
+    {
+      "epoch": 0.07360972232905161,
+      "learning_rate": 0.002960070387674442,
+      "loss": 1.2879,
+      "step": 957
+    },
+    {
+      "epoch": 0.07368663948927005,
+      "learning_rate": 0.0029599872697932393,
+      "loss": 1.3842,
+      "step": 958
+    },
+    {
+      "epoch": 0.0737635566494885,
+      "learning_rate": 0.002959904066661857,
+      "loss": 1.1728,
+      "step": 959
+    },
+    {
+      "epoch": 0.07384047380970694,
+      "learning_rate": 0.0029598207782851557,
+      "loss": 1.1599,
+      "step": 960
+    },
+    {
+      "epoch": 0.07391739096992539,
+      "learning_rate": 0.0029597374046679975,
+      "loss": 1.1854,
+      "step": 961
+    },
+    {
+      "epoch": 0.07399430813014383,
+      "learning_rate": 0.0029596539458152512,
+      "loss": 1.1199,
+      "step": 962
+    },
+    {
+      "epoch": 0.07407122529036228,
+      "learning_rate": 0.0029595704017317897,
+      "loss": 1.3585,
+      "step": 963
+    },
+    {
+      "epoch": 0.07414814245058073,
+      "learning_rate": 0.002959486772422491,
+      "loss": 1.2377,
+      "step": 964
+    },
+    {
+      "epoch": 0.07422505961079917,
+      "learning_rate": 0.0029594030578922385,
+      "loss": 1.1028,
+      "step": 965
+    },
+    {
+      "epoch": 0.07430197677101762,
+      "learning_rate": 0.002959319258145921,
+      "loss": 1.1915,
+      "step": 966
+    },
+    {
+      "epoch": 0.07437889393123606,
+      "learning_rate": 0.002959235373188431,
+      "loss": 0.9092,
+      "step": 967
+    },
+    {
+      "epoch": 0.07445581109145451,
+      "learning_rate": 0.0029591514030246667,
+      "loss": 0.9983,
+      "step": 968
+    },
+    {
+      "epoch": 0.07453272825167295,
+      "learning_rate": 0.002959067347659531,
+      "loss": 1.3972,
+      "step": 969
+    },
+    {
+      "epoch": 0.0746096454118914,
+      "learning_rate": 0.002958983207097933,
+      "loss": 1.1719,
+      "step": 970
+    },
+    {
+      "epoch": 0.07468656257210984,
+      "learning_rate": 0.0029588989813447843,
+      "loss": 1.009,
+      "step": 971
+    },
+    {
+      "epoch": 0.07476347973232829,
+      "learning_rate": 0.0029588146704050042,
+      "loss": 1.3163,
+      "step": 972
+    },
+    {
+      "epoch": 0.07484039689254672,
+      "learning_rate": 0.002958730274283515,
+      "loss": 1.0569,
+      "step": 973
+    },
+    {
+      "epoch": 0.07491731405276517,
+      "learning_rate": 0.0029586457929852445,
+      "loss": 1.647,
+      "step": 974
+    },
+    {
+      "epoch": 0.07499423121298361,
+      "learning_rate": 0.0029585612265151263,
+      "loss": 1.0079,
+      "step": 975
+    },
+    {
+      "epoch": 0.07507114837320206,
+      "learning_rate": 0.0029584765748780976,
+      "loss": 1.3997,
+      "step": 976
+    },
+    {
+      "epoch": 0.0751480655334205,
+      "learning_rate": 0.002958391838079102,
+      "loss": 0.9782,
+      "step": 977
+    },
+    {
+      "epoch": 0.07522498269363895,
+      "learning_rate": 0.0029583070161230875,
+      "loss": 0.8678,
+      "step": 978
+    },
+    {
+      "epoch": 0.07530189985385739,
+      "learning_rate": 0.002958222109015006,
+      "loss": 1.0651,
+      "step": 979
+    },
+    {
+      "epoch": 0.07537881701407584,
+      "learning_rate": 0.002958137116759816,
+      "loss": 1.3772,
+      "step": 980
+    },
+    {
+      "epoch": 0.07545573417429428,
+      "learning_rate": 0.00295805203936248,
+      "loss": 1.3194,
+      "step": 981
+    },
+    {
+      "epoch": 0.07553265133451273,
+      "learning_rate": 0.0029579668768279664,
+      "loss": 1.5207,
+      "step": 982
+    },
+    {
+      "epoch": 0.07560956849473117,
+      "learning_rate": 0.0029578816291612464,
+      "loss": 1.2981,
+      "step": 983
+    },
+    {
+      "epoch": 0.07568648565494962,
+      "learning_rate": 0.0029577962963672996,
+      "loss": 1.4814,
+      "step": 984
+    },
+    {
+      "epoch": 0.07576340281516807,
+      "learning_rate": 0.002957710878451107,
+      "loss": 1.1017,
+      "step": 985
+    },
+    {
+      "epoch": 0.07584031997538651,
+      "learning_rate": 0.0029576253754176576,
+      "loss": 1.1641,
+      "step": 986
+    },
+    {
+      "epoch": 0.07591723713560496,
+      "learning_rate": 0.0029575397872719433,
+      "loss": 1.505,
+      "step": 987
+    },
+    {
+      "epoch": 0.0759941542958234,
+      "learning_rate": 0.002957454114018962,
+      "loss": 1.3036,
+      "step": 988
+    },
+    {
+      "epoch": 0.07607107145604185,
+      "learning_rate": 0.0029573683556637156,
+      "loss": 0.8347,
+      "step": 989
+    },
+    {
+      "epoch": 0.07614798861626029,
+      "learning_rate": 0.0029572825122112126,
+      "loss": 1.1208,
+      "step": 990
+    },
+    {
+      "epoch": 0.07622490577647874,
+      "learning_rate": 0.0029571965836664646,
+      "loss": 0.9651,
+      "step": 991
+    },
+    {
+      "epoch": 0.07630182293669718,
+      "learning_rate": 0.00295711057003449,
+      "loss": 1.2866,
+      "step": 992
+    },
+    {
+      "epoch": 0.07637874009691561,
+      "learning_rate": 0.0029570244713203095,
+      "loss": 1.2155,
+      "step": 993
+    },
+    {
+      "epoch": 0.07645565725713406,
+      "learning_rate": 0.0029569382875289526,
+      "loss": 1.4266,
+      "step": 994
+    },
+    {
+      "epoch": 0.0765325744173525,
+      "learning_rate": 0.0029568520186654505,
+      "loss": 1.0505,
+      "step": 995
+    },
+    {
+      "epoch": 0.07660949157757095,
+      "learning_rate": 0.0029567656647348406,
+      "loss": 1.3778,
+      "step": 996
+    },
+    {
+      "epoch": 0.0766864087377894,
+      "learning_rate": 0.0029566792257421657,
+      "loss": 1.2593,
+      "step": 997
+    },
+    {
+      "epoch": 0.07676332589800784,
+      "learning_rate": 0.002956592701692472,
+      "loss": 1.1632,
+      "step": 998
+    },
+    {
+      "epoch": 0.07684024305822629,
+      "learning_rate": 0.002956506092590813,
+      "loss": 1.1734,
+      "step": 999
+    },
+    {
+      "epoch": 0.07691716021844473,
+      "learning_rate": 0.002956419398442245,
+      "loss": 1.1082,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07699407737866318,
+      "learning_rate": 0.0029563326192518307,
+      "loss": 1.7411,
+      "step": 1001
+    },
+    {
+      "epoch": 0.07707099453888162,
+      "learning_rate": 0.0029562457550246367,
+      "loss": 1.5031,
+      "step": 1002
+    },
+    {
+      "epoch": 0.07714791169910007,
+      "learning_rate": 0.0029561588057657358,
+      "loss": 0.7835,
+      "step": 1003
+    },
+    {
+      "epoch": 0.07722482885931851,
+      "learning_rate": 0.0029560717714802046,
+      "loss": 1.5095,
+      "step": 1004
+    },
+    {
+      "epoch": 0.07730174601953696,
+      "learning_rate": 0.002955984652173125,
+      "loss": 1.4018,
+      "step": 1005
+    },
+    {
+      "epoch": 0.0773786631797554,
+      "learning_rate": 0.002955897447849584,
+      "loss": 1.3244,
+      "step": 1006
+    },
+    {
+      "epoch": 0.07745558033997385,
+      "learning_rate": 0.0029558101585146746,
+      "loss": 1.0778,
+      "step": 1007
+    },
+    {
+      "epoch": 0.0775324975001923,
+      "learning_rate": 0.002955722784173492,
+      "loss": 1.5623,
+      "step": 1008
+    },
+    {
+      "epoch": 0.07760941466041074,
+      "learning_rate": 0.002955635324831139,
+      "loss": 0.9722,
+      "step": 1009
+    },
+    {
+      "epoch": 0.07768633182062919,
+      "learning_rate": 0.0029555477804927223,
+      "loss": 0.7629,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07776324898084763,
+      "learning_rate": 0.002955460151163354,
+      "loss": 1.4302,
+      "step": 1011
+    },
+    {
+      "epoch": 0.07784016614106608,
+      "learning_rate": 0.002955372436848151,
+      "loss": 1.8574,
+      "step": 1012
+    },
+    {
+      "epoch": 0.07791708330128452,
+      "learning_rate": 0.002955284637552234,
+      "loss": 1.0181,
+      "step": 1013
+    },
+    {
+      "epoch": 0.07799400046150295,
+      "learning_rate": 0.00295519675328073,
+      "loss": 1.1129,
+      "step": 1014
+    },
+    {
+      "epoch": 0.0780709176217214,
+      "learning_rate": 0.002955108784038772,
+      "loss": 1.1438,
+      "step": 1015
+    },
+    {
+      "epoch": 0.07814783478193985,
+      "learning_rate": 0.002955020729831495,
+      "loss": 1.0609,
+      "step": 1016
+    },
+    {
+      "epoch": 0.07822475194215829,
+      "learning_rate": 0.0029549325906640415,
+      "loss": 1.307,
+      "step": 1017
+    },
+    {
+      "epoch": 0.07830166910237674,
+      "learning_rate": 0.002954844366541558,
+      "loss": 0.9569,
+      "step": 1018
+    },
+    {
+      "epoch": 0.07837858626259518,
+      "learning_rate": 0.002954756057469195,
+      "loss": 0.955,
+      "step": 1019
+    },
+    {
+      "epoch": 0.07845550342281363,
+      "learning_rate": 0.00295466766345211,
+      "loss": 1.2528,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07853242058303207,
+      "learning_rate": 0.002954579184495464,
+      "loss": 1.4933,
+      "step": 1021
+    },
+    {
+      "epoch": 0.07860933774325052,
+      "learning_rate": 0.002954490620604424,
+      "loss": 1.0781,
+      "step": 1022
+    },
+    {
+      "epoch": 0.07868625490346896,
+      "learning_rate": 0.0029544019717841602,
+      "loss": 0.9809,
+      "step": 1023
+    },
+    {
+      "epoch": 0.07876317206368741,
+      "learning_rate": 0.00295431323803985,
+      "loss": 0.8879,
+      "step": 1024
+    },
+    {
+      "epoch": 0.07884008922390585,
+      "learning_rate": 0.0029542244193766737,
+      "loss": 0.7664,
+      "step": 1025
+    },
+    {
+      "epoch": 0.0789170063841243,
+      "learning_rate": 0.0029541355157998187,
+      "loss": 0.8898,
+      "step": 1026
+    },
+    {
+      "epoch": 0.07899392354434275,
+      "learning_rate": 0.002954046527314475,
+      "loss": 1.248,
+      "step": 1027
+    },
+    {
+      "epoch": 0.07907084070456119,
+      "learning_rate": 0.0029539574539258393,
+      "loss": 1.0316,
+      "step": 1028
+    },
+    {
+      "epoch": 0.07914775786477964,
+      "learning_rate": 0.0029538682956391127,
+      "loss": 0.968,
+      "step": 1029
+    },
+    {
+      "epoch": 0.07922467502499808,
+      "learning_rate": 0.0029537790524595004,
+      "loss": 1.1932,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07930159218521653,
+      "learning_rate": 0.0029536897243922153,
+      "loss": 1.1205,
+      "step": 1031
+    },
+    {
+      "epoch": 0.07937850934543497,
+      "learning_rate": 0.0029536003114424714,
+      "loss": 1.2685,
+      "step": 1032
+    },
+    {
+      "epoch": 0.07945542650565342,
+      "learning_rate": 0.0029535108136154904,
+      "loss": 1.2933,
+      "step": 1033
+    },
+    {
+      "epoch": 0.07953234366587185,
+      "learning_rate": 0.002953421230916499,
+      "loss": 1.051,
+      "step": 1034
+    },
+    {
+      "epoch": 0.0796092608260903,
+      "learning_rate": 0.0029533315633507263,
+      "loss": 1.0509,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07968617798630874,
+      "learning_rate": 0.0029532418109234093,
+      "loss": 0.8834,
+      "step": 1036
+    },
+    {
+      "epoch": 0.07976309514652719,
+      "learning_rate": 0.0029531519736397884,
+      "loss": 1.2259,
+      "step": 1037
+    },
+    {
+      "epoch": 0.07984001230674563,
+      "learning_rate": 0.0029530620515051094,
+      "loss": 1.0971,
+      "step": 1038
+    },
+    {
+      "epoch": 0.07991692946696408,
+      "learning_rate": 0.0029529720445246226,
+      "loss": 1.0709,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07999384662718252,
+      "learning_rate": 0.002952881952703584,
+      "loss": 1.0895,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08007076378740097,
+      "learning_rate": 0.0029527917760472544,
+      "loss": 1.3178,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08014768094761941,
+      "learning_rate": 0.0029527015145608985,
+      "loss": 1.3647,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08022459810783786,
+      "learning_rate": 0.0029526111682497873,
+      "loss": 0.541,
+      "step": 1043
+    },
+    {
+      "epoch": 0.0803015152680563,
+      "learning_rate": 0.002952520737119196,
+      "loss": 1.0206,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08037843242827475,
+      "learning_rate": 0.0029524302211744054,
+      "loss": 1.4368,
+      "step": 1045
+    },
+    {
+      "epoch": 0.0804553495884932,
+      "learning_rate": 0.0029523396204207,
+      "loss": 1.3588,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08053226674871164,
+      "learning_rate": 0.0029522489348633714,
+      "loss": 1.2697,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08060918390893008,
+      "learning_rate": 0.0029521581645077133,
+      "loss": 1.1707,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08068610106914853,
+      "learning_rate": 0.0029520673093590267,
+      "loss": 1.2145,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08076301822936698,
+      "learning_rate": 0.0029519763694226164,
+      "loss": 1.1222,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08083993538958542,
+      "learning_rate": 0.0029518853447037933,
+      "loss": 1.0963,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08091685254980387,
+      "learning_rate": 0.002951794235207871,
+      "loss": 1.1351,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08099376971002231,
+      "learning_rate": 0.002951703040940171,
+      "loss": 1.0011,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08107068687024076,
+      "learning_rate": 0.0029516117619060173,
+      "loss": 1.1586,
+      "step": 1054
+    },
+    {
+      "epoch": 0.08114760403045919,
+      "learning_rate": 0.0029515203981107397,
+      "loss": 0.9994,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08122452119067763,
+      "learning_rate": 0.002951428949559674,
+      "loss": 1.0701,
+      "step": 1056
+    },
+    {
+      "epoch": 0.08130143835089608,
+      "learning_rate": 0.0029513374162581586,
+      "loss": 0.9321,
+      "step": 1057
+    },
+    {
+      "epoch": 0.08137835551111453,
+      "learning_rate": 0.0029512457982115393,
+      "loss": 1.3266,
+      "step": 1058
+    },
+    {
+      "epoch": 0.08145527267133297,
+      "learning_rate": 0.0029511540954251653,
+      "loss": 1.2703,
+      "step": 1059
+    },
+    {
+      "epoch": 0.08153218983155142,
+      "learning_rate": 0.002951062307904391,
+      "loss": 1.1536,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08160910699176986,
+      "learning_rate": 0.002950970435654577,
+      "loss": 1.0513,
+      "step": 1061
+    },
+    {
+      "epoch": 0.0816860241519883,
+      "learning_rate": 0.0029508784786810866,
+      "loss": 1.0564,
+      "step": 1062
+    },
+    {
+      "epoch": 0.08176294131220675,
+      "learning_rate": 0.0029507864369892898,
+      "loss": 1.1673,
+      "step": 1063
+    },
+    {
+      "epoch": 0.0818398584724252,
+      "learning_rate": 0.002950694310584561,
+      "loss": 1.2823,
+      "step": 1064
+    },
+    {
+      "epoch": 0.08191677563264364,
+      "learning_rate": 0.002950602099472279,
+      "loss": 1.1043,
+      "step": 1065
+    },
+    {
+      "epoch": 0.08199369279286209,
+      "learning_rate": 0.002950509803657829,
+      "loss": 0.9982,
+      "step": 1066
+    },
+    {
+      "epoch": 0.08207060995308053,
+      "learning_rate": 0.0029504174231466005,
+      "loss": 0.8833,
+      "step": 1067
+    },
+    {
+      "epoch": 0.08214752711329898,
+      "learning_rate": 0.0029503249579439866,
+      "loss": 1.4851,
+      "step": 1068
+    },
+    {
+      "epoch": 0.08222444427351742,
+      "learning_rate": 0.002950232408055387,
+      "loss": 1.4989,
+      "step": 1069
+    },
+    {
+      "epoch": 0.08230136143373587,
+      "learning_rate": 0.0029501397734862053,
+      "loss": 1.3504,
+      "step": 1070
+    },
+    {
+      "epoch": 0.08237827859395432,
+      "learning_rate": 0.0029500470542418514,
+      "loss": 1.2241,
+      "step": 1071
+    },
+    {
+      "epoch": 0.08245519575417276,
+      "learning_rate": 0.0029499542503277384,
+      "loss": 1.1753,
+      "step": 1072
+    },
+    {
+      "epoch": 0.0825321129143912,
+      "learning_rate": 0.0029498613617492857,
+      "loss": 1.2218,
+      "step": 1073
+    },
+    {
+      "epoch": 0.08260903007460965,
+      "learning_rate": 0.0029497683885119173,
+      "loss": 0.9293,
+      "step": 1074
+    },
+    {
+      "epoch": 0.08268594723482808,
+      "learning_rate": 0.0029496753306210615,
+      "loss": 1.1212,
+      "step": 1075
+    },
+    {
+      "epoch": 0.08276286439504653,
+      "learning_rate": 0.002949582188082153,
+      "loss": 1.2674,
+      "step": 1076
+    },
+    {
+      "epoch": 0.08283978155526497,
+      "learning_rate": 0.002949488960900629,
+      "loss": 1.1751,
+      "step": 1077
+    },
+    {
+      "epoch": 0.08291669871548342,
+      "learning_rate": 0.0029493956490819345,
+      "loss": 1.1667,
+      "step": 1078
+    },
+    {
+      "epoch": 0.08299361587570186,
+      "learning_rate": 0.002949302252631517,
+      "loss": 1.0874,
+      "step": 1079
+    },
+    {
+      "epoch": 0.08307053303592031,
+      "learning_rate": 0.002949208771554831,
+      "loss": 1.1288,
+      "step": 1080
+    },
+    {
+      "epoch": 0.08314745019613876,
+      "learning_rate": 0.002949115205857334,
+      "loss": 0.9737,
+      "step": 1081
+    },
+    {
+      "epoch": 0.0832243673563572,
+      "learning_rate": 0.00294902155554449,
+      "loss": 1.3464,
+      "step": 1082
+    },
+    {
+      "epoch": 0.08330128451657565,
+      "learning_rate": 0.002948927820621768,
+      "loss": 1.1628,
+      "step": 1083
+    },
+    {
+      "epoch": 0.08337820167679409,
+      "learning_rate": 0.00294883400109464,
+      "loss": 1.159,
+      "step": 1084
+    },
+    {
+      "epoch": 0.08345511883701254,
+      "learning_rate": 0.0029487400969685847,
+      "loss": 1.2899,
+      "step": 1085
+    },
+    {
+      "epoch": 0.08353203599723098,
+      "learning_rate": 0.0029486461082490856,
+      "loss": 1.4605,
+      "step": 1086
+    },
+    {
+      "epoch": 0.08360895315744943,
+      "learning_rate": 0.00294855203494163,
+      "loss": 1.0857,
+      "step": 1087
+    },
+    {
+      "epoch": 0.08368587031766787,
+      "learning_rate": 0.0029484578770517115,
+      "loss": 1.1297,
+      "step": 1088
+    },
+    {
+      "epoch": 0.08376278747788632,
+      "learning_rate": 0.0029483636345848285,
+      "loss": 1.0929,
+      "step": 1089
+    },
+    {
+      "epoch": 0.08383970463810476,
+      "learning_rate": 0.002948269307546483,
+      "loss": 1.1001,
+      "step": 1090
+    },
+    {
+      "epoch": 0.08391662179832321,
+      "learning_rate": 0.0029481748959421833,
+      "loss": 1.2571,
+      "step": 1091
+    },
+    {
+      "epoch": 0.08399353895854166,
+      "learning_rate": 0.0029480803997774425,
+      "loss": 1.1372,
+      "step": 1092
+    },
+    {
+      "epoch": 0.0840704561187601,
+      "learning_rate": 0.0029479858190577775,
+      "loss": 1.5051,
+      "step": 1093
+    },
+    {
+      "epoch": 0.08414737327897855,
+      "learning_rate": 0.002947891153788712,
+      "loss": 1.0785,
+      "step": 1094
+    },
+    {
+      "epoch": 0.08422429043919699,
+      "learning_rate": 0.002947796403975773,
+      "loss": 1.3196,
+      "step": 1095
+    },
+    {
+      "epoch": 0.08430120759941542,
+      "learning_rate": 0.002947701569624493,
+      "loss": 0.9348,
+      "step": 1096
+    },
+    {
+      "epoch": 0.08437812475963387,
+      "learning_rate": 0.0029476066507404092,
+      "loss": 1.3281,
+      "step": 1097
+    },
+    {
+      "epoch": 0.08445504191985231,
+      "learning_rate": 0.002947511647329065,
+      "loss": 1.1641,
+      "step": 1098
+    },
+    {
+      "epoch": 0.08453195908007076,
+      "learning_rate": 0.0029474165593960065,
+      "loss": 1.2422,
+      "step": 1099
+    },
+    {
+      "epoch": 0.0846088762402892,
+      "learning_rate": 0.0029473213869467873,
+      "loss": 0.9473,
+      "step": 1100
+    },
+    {
+      "epoch": 0.08468579340050765,
+      "learning_rate": 0.0029472261299869633,
+      "loss": 1.2083,
+      "step": 1101
+    },
+    {
+      "epoch": 0.0847627105607261,
+      "learning_rate": 0.002947130788522098,
+      "loss": 1.1972,
+      "step": 1102
+    },
+    {
+      "epoch": 0.08483962772094454,
+      "learning_rate": 0.0029470353625577574,
+      "loss": 1.3026,
+      "step": 1103
+    },
+    {
+      "epoch": 0.08491654488116299,
+      "learning_rate": 0.002946939852099514,
+      "loss": 1.2338,
+      "step": 1104
+    },
+    {
+      "epoch": 0.08499346204138143,
+      "learning_rate": 0.002946844257152945,
+      "loss": 1.2036,
+      "step": 1105
+    },
+    {
+      "epoch": 0.08507037920159988,
+      "learning_rate": 0.0029467485777236314,
+      "loss": 0.9109,
+      "step": 1106
+    },
+    {
+      "epoch": 0.08514729636181832,
+      "learning_rate": 0.002946652813817161,
+      "loss": 1.0046,
+      "step": 1107
+    },
+    {
+      "epoch": 0.08522421352203677,
+      "learning_rate": 0.0029465569654391244,
+      "loss": 1.2388,
+      "step": 1108
+    },
+    {
+      "epoch": 0.08530113068225521,
+      "learning_rate": 0.00294646103259512,
+      "loss": 1.1558,
+      "step": 1109
+    },
+    {
+      "epoch": 0.08537804784247366,
+      "learning_rate": 0.0029463650152907477,
+      "loss": 0.8834,
+      "step": 1110
+    },
+    {
+      "epoch": 0.0854549650026921,
+      "learning_rate": 0.002946268913531615,
+      "loss": 1.1247,
+      "step": 1111
+    },
+    {
+      "epoch": 0.08553188216291055,
+      "learning_rate": 0.0029461727273233334,
+      "loss": 0.9333,
+      "step": 1112
+    },
+    {
+      "epoch": 0.085608799323129,
+      "learning_rate": 0.0029460764566715193,
+      "loss": 1.2425,
+      "step": 1113
+    },
+    {
+      "epoch": 0.08568571648334744,
+      "learning_rate": 0.0029459801015817933,
+      "loss": 0.9309,
+      "step": 1114
+    },
+    {
+      "epoch": 0.08576263364356589,
+      "learning_rate": 0.0029458836620597823,
+      "loss": 1.3583,
+      "step": 1115
+    },
+    {
+      "epoch": 0.08583955080378432,
+      "learning_rate": 0.0029457871381111172,
+      "loss": 0.9936,
+      "step": 1116
+    },
+    {
+      "epoch": 0.08591646796400276,
+      "learning_rate": 0.002945690529741435,
+      "loss": 1.1197,
+      "step": 1117
+    },
+    {
+      "epoch": 0.08599338512422121,
+      "learning_rate": 0.002945593836956376,
+      "loss": 0.935,
+      "step": 1118
+    },
+    {
+      "epoch": 0.08607030228443965,
+      "learning_rate": 0.002945497059761586,
+      "loss": 1.1282,
+      "step": 1119
+    },
+    {
+      "epoch": 0.0861472194446581,
+      "learning_rate": 0.002945400198162716,
+      "loss": 1.2256,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08622413660487654,
+      "learning_rate": 0.0029453032521654225,
+      "loss": 1.156,
+      "step": 1121
+    },
+    {
+      "epoch": 0.08630105376509499,
+      "learning_rate": 0.0029452062217753656,
+      "loss": 1.3209,
+      "step": 1122
+    },
+    {
+      "epoch": 0.08637797092531344,
+      "learning_rate": 0.0029451091069982115,
+      "loss": 1.1248,
+      "step": 1123
+    },
+    {
+      "epoch": 0.08645488808553188,
+      "learning_rate": 0.0029450119078396303,
+      "loss": 1.3843,
+      "step": 1124
+    },
+    {
+      "epoch": 0.08653180524575033,
+      "learning_rate": 0.002944914624305298,
+      "loss": 1.3654,
+      "step": 1125
+    },
+    {
+      "epoch": 0.08660872240596877,
+      "learning_rate": 0.0029448172564008946,
+      "loss": 1.0795,
+      "step": 1126
+    },
+    {
+      "epoch": 0.08668563956618722,
+      "learning_rate": 0.0029447198041321065,
+      "loss": 1.3927,
+      "step": 1127
+    },
+    {
+      "epoch": 0.08676255672640566,
+      "learning_rate": 0.002944622267504623,
+      "loss": 0.9433,
+      "step": 1128
+    },
+    {
+      "epoch": 0.08683947388662411,
+      "learning_rate": 0.00294452464652414,
+      "loss": 0.9105,
+      "step": 1129
+    },
+    {
+      "epoch": 0.08691639104684255,
+      "learning_rate": 0.0029444269411963565,
+      "loss": 1.0921,
+      "step": 1130
+    },
+    {
+      "epoch": 0.086993308207061,
+      "learning_rate": 0.0029443291515269796,
+      "loss": 1.4,
+      "step": 1131
+    },
+    {
+      "epoch": 0.08707022536727944,
+      "learning_rate": 0.002944231277521718,
+      "loss": 1.5666,
+      "step": 1132
+    },
+    {
+      "epoch": 0.08714714252749789,
+      "learning_rate": 0.002944133319186287,
+      "loss": 1.0999,
+      "step": 1133
+    },
+    {
+      "epoch": 0.08722405968771633,
+      "learning_rate": 0.002944035276526406,
+      "loss": 1.1352,
+      "step": 1134
+    },
+    {
+      "epoch": 0.08730097684793478,
+      "learning_rate": 0.002943937149547801,
+      "loss": 1.3064,
+      "step": 1135
+    },
+    {
+      "epoch": 0.08737789400815323,
+      "learning_rate": 0.0029438389382562004,
+      "loss": 1.0006,
+      "step": 1136
+    },
+    {
+      "epoch": 0.08745481116837166,
+      "learning_rate": 0.00294374064265734,
+      "loss": 1.2764,
+      "step": 1137
+    },
+    {
+      "epoch": 0.0875317283285901,
+      "learning_rate": 0.0029436422627569583,
+      "loss": 1.3404,
+      "step": 1138
+    },
+    {
+      "epoch": 0.08760864548880855,
+      "learning_rate": 0.002943543798560801,
+      "loss": 1.3129,
+      "step": 1139
+    },
+    {
+      "epoch": 0.087685562649027,
+      "learning_rate": 0.0029434452500746162,
+      "loss": 1.1383,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08776247980924544,
+      "learning_rate": 0.002943346617304159,
+      "loss": 1.2198,
+      "step": 1141
+    },
+    {
+      "epoch": 0.08783939696946388,
+      "learning_rate": 0.0029432479002551894,
+      "loss": 1.0944,
+      "step": 1142
+    },
+    {
+      "epoch": 0.08791631412968233,
+      "learning_rate": 0.00294314909893347,
+      "loss": 1.2941,
+      "step": 1143
+    },
+    {
+      "epoch": 0.08799323128990078,
+      "learning_rate": 0.0029430502133447707,
+      "loss": 1.426,
+      "step": 1144
+    },
+    {
+      "epoch": 0.08807014845011922,
+      "learning_rate": 0.002942951243494866,
+      "loss": 1.2258,
+      "step": 1145
+    },
+    {
+      "epoch": 0.08814706561033767,
+      "learning_rate": 0.002942852189389534,
+      "loss": 1.2612,
+      "step": 1146
+    },
+    {
+      "epoch": 0.08822398277055611,
+      "learning_rate": 0.002942753051034559,
+      "loss": 1.0856,
+      "step": 1147
+    },
+    {
+      "epoch": 0.08830089993077456,
+      "learning_rate": 0.0029426538284357297,
+      "loss": 1.2527,
+      "step": 1148
+    },
+    {
+      "epoch": 0.088377817090993,
+      "learning_rate": 0.00294255452159884,
+      "loss": 1.212,
+      "step": 1149
+    },
+    {
+      "epoch": 0.08845473425121145,
+      "learning_rate": 0.0029424551305296887,
+      "loss": 0.6959,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08853165141142989,
+      "learning_rate": 0.0029423556552340786,
+      "loss": 1.2818,
+      "step": 1151
+    },
+    {
+      "epoch": 0.08860856857164834,
+      "learning_rate": 0.0029422560957178185,
+      "loss": 0.7935,
+      "step": 1152
+    },
+    {
+      "epoch": 0.08868548573186678,
+      "learning_rate": 0.0029421564519867224,
+      "loss": 1.2913,
+      "step": 1153
+    },
+    {
+      "epoch": 0.08876240289208523,
+      "learning_rate": 0.0029420567240466076,
+      "loss": 0.7006,
+      "step": 1154
+    },
+    {
+      "epoch": 0.08883932005230367,
+      "learning_rate": 0.002941956911903298,
+      "loss": 1.3979,
+      "step": 1155
+    },
+    {
+      "epoch": 0.08891623721252212,
+      "learning_rate": 0.0029418570155626215,
+      "loss": 0.8993,
+      "step": 1156
+    },
+    {
+      "epoch": 0.08899315437274055,
+      "learning_rate": 0.002941757035030412,
+      "loss": 1.0958,
+      "step": 1157
+    },
+    {
+      "epoch": 0.089070071532959,
+      "learning_rate": 0.002941656970312505,
+      "loss": 1.0649,
+      "step": 1158
+    },
+    {
+      "epoch": 0.08914698869317744,
+      "learning_rate": 0.0029415568214147467,
+      "loss": 1.014,
+      "step": 1159
+    },
+    {
+      "epoch": 0.08922390585339589,
+      "learning_rate": 0.002941456588342982,
+      "loss": 1.2312,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08930082301361433,
+      "learning_rate": 0.0029413562711030658,
+      "loss": 1.2211,
+      "step": 1161
+    },
+    {
+      "epoch": 0.08937774017383278,
+      "learning_rate": 0.002941255869700854,
+      "loss": 1.1923,
+      "step": 1162
+    },
+    {
+      "epoch": 0.08945465733405122,
+      "learning_rate": 0.00294115538414221,
+      "loss": 1.2505,
+      "step": 1163
+    },
+    {
+      "epoch": 0.08953157449426967,
+      "learning_rate": 0.0029410548144330017,
+      "loss": 0.8847,
+      "step": 1164
+    },
+    {
+      "epoch": 0.08960849165448811,
+      "learning_rate": 0.0029409541605791008,
+      "loss": 1.242,
+      "step": 1165
+    },
+    {
+      "epoch": 0.08968540881470656,
+      "learning_rate": 0.0029408534225863845,
+      "loss": 1.1686,
+      "step": 1166
+    },
+    {
+      "epoch": 0.089762325974925,
+      "learning_rate": 0.002940752600460736,
+      "loss": 0.9645,
+      "step": 1167
+    },
+    {
+      "epoch": 0.08983924313514345,
+      "learning_rate": 0.0029406516942080403,
+      "loss": 1.215,
+      "step": 1168
+    },
+    {
+      "epoch": 0.0899161602953619,
+      "learning_rate": 0.0029405507038341916,
+      "loss": 1.117,
+      "step": 1169
+    },
+    {
+      "epoch": 0.08999307745558034,
+      "learning_rate": 0.002940449629345086,
+      "loss": 1.1072,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09006999461579879,
+      "learning_rate": 0.002940348470746625,
+      "loss": 1.0377,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09014691177601723,
+      "learning_rate": 0.002940247228044716,
+      "loss": 0.8495,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09022382893623568,
+      "learning_rate": 0.00294014590124527,
+      "loss": 1.0416,
+      "step": 1173
+    },
+    {
+      "epoch": 0.09030074609645412,
+      "learning_rate": 0.002940044490354205,
+      "loss": 1.1154,
+      "step": 1174
+    },
+    {
+      "epoch": 0.09037766325667257,
+      "learning_rate": 0.0029399429953774403,
+      "loss": 1.1448,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09045458041689101,
+      "learning_rate": 0.0029398414163209034,
+      "loss": 1.407,
+      "step": 1176
+    },
+    {
+      "epoch": 0.09053149757710946,
+      "learning_rate": 0.0029397397531905262,
+      "loss": 1.0101,
+      "step": 1177
+    },
+    {
+      "epoch": 0.09060841473732789,
+      "learning_rate": 0.002939638005992244,
+      "loss": 0.9946,
+      "step": 1178
+    },
+    {
+      "epoch": 0.09068533189754634,
+      "learning_rate": 0.002939536174731999,
+      "loss": 1.0516,
+      "step": 1179
+    },
+    {
+      "epoch": 0.09076224905776478,
+      "learning_rate": 0.0029394342594157357,
+      "loss": 1.7657,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09083916621798323,
+      "learning_rate": 0.002939332260049406,
+      "loss": 1.4462,
+      "step": 1181
+    },
+    {
+      "epoch": 0.09091608337820167,
+      "learning_rate": 0.0029392301766389657,
+      "loss": 1.2188,
+      "step": 1182
+    },
+    {
+      "epoch": 0.09099300053842012,
+      "learning_rate": 0.0029391280091903755,
+      "loss": 0.9906,
+      "step": 1183
+    },
+    {
+      "epoch": 0.09106991769863856,
+      "learning_rate": 0.002939025757709601,
+      "loss": 1.3368,
+      "step": 1184
+    },
+    {
+      "epoch": 0.09114683485885701,
+      "learning_rate": 0.0029389234222026127,
+      "loss": 1.2898,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09122375201907545,
+      "learning_rate": 0.0029388210026753863,
+      "loss": 1.1105,
+      "step": 1186
+    },
+    {
+      "epoch": 0.0913006691792939,
+      "learning_rate": 0.002938718499133902,
+      "loss": 1.0106,
+      "step": 1187
+    },
+    {
+      "epoch": 0.09137758633951235,
+      "learning_rate": 0.0029386159115841453,
+      "loss": 1.1443,
+      "step": 1188
+    },
+    {
+      "epoch": 0.09145450349973079,
+      "learning_rate": 0.0029385132400321055,
+      "loss": 1.2002,
+      "step": 1189
+    },
+    {
+      "epoch": 0.09153142065994924,
+      "learning_rate": 0.0029384104844837793,
+      "loss": 1.2615,
+      "step": 1190
+    },
+    {
+      "epoch": 0.09160833782016768,
+      "learning_rate": 0.0029383076449451657,
+      "loss": 1.1345,
+      "step": 1191
+    },
+    {
+      "epoch": 0.09168525498038613,
+      "learning_rate": 0.0029382047214222694,
+      "loss": 1.4055,
+      "step": 1192
+    },
+    {
+      "epoch": 0.09176217214060457,
+      "learning_rate": 0.0029381017139211004,
+      "loss": 1.4962,
+      "step": 1193
+    },
+    {
+      "epoch": 0.09183908930082302,
+      "learning_rate": 0.002937998622447674,
+      "loss": 1.1178,
+      "step": 1194
+    },
+    {
+      "epoch": 0.09191600646104146,
+      "learning_rate": 0.0029378954470080094,
+      "loss": 1.0983,
+      "step": 1195
+    },
+    {
+      "epoch": 0.09199292362125991,
+      "learning_rate": 0.0029377921876081307,
+      "loss": 0.9965,
+      "step": 1196
+    },
+    {
+      "epoch": 0.09206984078147835,
+      "learning_rate": 0.002937688844254068,
+      "loss": 1.1954,
+      "step": 1197
+    },
+    {
+      "epoch": 0.09214675794169679,
+      "learning_rate": 0.002937585416951856,
+      "loss": 1.6167,
+      "step": 1198
+    },
+    {
+      "epoch": 0.09222367510191523,
+      "learning_rate": 0.0029374819057075324,
+      "loss": 1.2936,
+      "step": 1199
+    },
+    {
+      "epoch": 0.09230059226213368,
+      "learning_rate": 0.0029373783105271427,
+      "loss": 0.9425,
+      "step": 1200
+    },
+    {
+      "epoch": 0.09237750942235212,
+      "learning_rate": 0.002937274631416735,
+      "loss": 1.5481,
+      "step": 1201
+    },
+    {
+      "epoch": 0.09245442658257057,
+      "learning_rate": 0.002937170868382364,
+      "loss": 1.1663,
+      "step": 1202
+    },
+    {
+      "epoch": 0.09253134374278901,
+      "learning_rate": 0.002937067021430088,
+      "loss": 1.0448,
+      "step": 1203
+    },
+    {
+      "epoch": 0.09260826090300746,
+      "learning_rate": 0.0029369630905659713,
+      "loss": 1.3691,
+      "step": 1204
+    },
+    {
+      "epoch": 0.0926851780632259,
+      "learning_rate": 0.002936859075796082,
+      "loss": 1.1063,
+      "step": 1205
+    },
+    {
+      "epoch": 0.09276209522344435,
+      "learning_rate": 0.0029367549771264936,
+      "loss": 1.2637,
+      "step": 1206
+    },
+    {
+      "epoch": 0.0928390123836628,
+      "learning_rate": 0.002936650794563285,
+      "loss": 1.298,
+      "step": 1207
+    },
+    {
+      "epoch": 0.09291592954388124,
+      "learning_rate": 0.0029365465281125393,
+      "loss": 1.0207,
+      "step": 1208
+    },
+    {
+      "epoch": 0.09299284670409969,
+      "learning_rate": 0.002936442177780344,
+      "loss": 0.8481,
+      "step": 1209
+    },
+    {
+      "epoch": 0.09306976386431813,
+      "learning_rate": 0.0029363377435727934,
+      "loss": 1.1319,
+      "step": 1210
+    },
+    {
+      "epoch": 0.09314668102453658,
+      "learning_rate": 0.0029362332254959855,
+      "loss": 1.1559,
+      "step": 1211
+    },
+    {
+      "epoch": 0.09322359818475502,
+      "learning_rate": 0.002936128623556022,
+      "loss": 1.5641,
+      "step": 1212
+    },
+    {
+      "epoch": 0.09330051534497347,
+      "learning_rate": 0.002936023937759012,
+      "loss": 1.5606,
+      "step": 1213
+    },
+    {
+      "epoch": 0.09337743250519191,
+      "learning_rate": 0.0029359191681110675,
+      "loss": 1.3077,
+      "step": 1214
+    },
+    {
+      "epoch": 0.09345434966541036,
+      "learning_rate": 0.0029358143146183064,
+      "loss": 1.3394,
+      "step": 1215
+    },
+    {
+      "epoch": 0.0935312668256288,
+      "learning_rate": 0.002935709377286851,
+      "loss": 1.0419,
+      "step": 1216
+    },
+    {
+      "epoch": 0.09360818398584725,
+      "learning_rate": 0.002935604356122829,
+      "loss": 1.3052,
+      "step": 1217
+    },
+    {
+      "epoch": 0.0936851011460657,
+      "learning_rate": 0.002935499251132372,
+      "loss": 1.0781,
+      "step": 1218
+    },
+    {
+      "epoch": 0.09376201830628413,
+      "learning_rate": 0.0029353940623216184,
+      "loss": 1.1302,
+      "step": 1219
+    },
+    {
+      "epoch": 0.09383893546650257,
+      "learning_rate": 0.002935288789696709,
+      "loss": 1.2968,
+      "step": 1220
+    },
+    {
+      "epoch": 0.09391585262672102,
+      "learning_rate": 0.0029351834332637912,
+      "loss": 1.3628,
+      "step": 1221
+    },
+    {
+      "epoch": 0.09399276978693946,
+      "learning_rate": 0.002935077993029018,
+      "loss": 1.2263,
+      "step": 1222
+    },
+    {
+      "epoch": 0.09406968694715791,
+      "learning_rate": 0.002934972468998544,
+      "loss": 1.4475,
+      "step": 1223
+    },
+    {
+      "epoch": 0.09414660410737635,
+      "learning_rate": 0.0029348668611785325,
+      "loss": 1.2735,
+      "step": 1224
+    },
+    {
+      "epoch": 0.0942235212675948,
+      "learning_rate": 0.0029347611695751496,
+      "loss": 1.124,
+      "step": 1225
+    },
+    {
+      "epoch": 0.09430043842781324,
+      "learning_rate": 0.0029346553941945667,
+      "loss": 1.5155,
+      "step": 1226
+    },
+    {
+      "epoch": 0.09437735558803169,
+      "learning_rate": 0.0029345495350429603,
+      "loss": 1.2345,
+      "step": 1227
+    },
+    {
+      "epoch": 0.09445427274825013,
+      "learning_rate": 0.002934443592126512,
+      "loss": 1.0955,
+      "step": 1228
+    },
+    {
+      "epoch": 0.09453118990846858,
+      "learning_rate": 0.002934337565451406,
+      "loss": 1.3493,
+      "step": 1229
+    },
+    {
+      "epoch": 0.09460810706868703,
+      "learning_rate": 0.002934231455023836,
+      "loss": 1.1968,
+      "step": 1230
+    },
+    {
+      "epoch": 0.09468502422890547,
+      "learning_rate": 0.0029341252608499962,
+      "loss": 1.3625,
+      "step": 1231
+    },
+    {
+      "epoch": 0.09476194138912392,
+      "learning_rate": 0.0029340189829360876,
+      "loss": 1.2965,
+      "step": 1232
+    },
+    {
+      "epoch": 0.09483885854934236,
+      "learning_rate": 0.0029339126212883167,
+      "loss": 1.1667,
+      "step": 1233
+    },
+    {
+      "epoch": 0.0949157757095608,
+      "learning_rate": 0.002933806175912893,
+      "loss": 1.4469,
+      "step": 1234
+    },
+    {
+      "epoch": 0.09499269286977925,
+      "learning_rate": 0.0029336996468160323,
+      "loss": 0.9501,
+      "step": 1235
+    },
+    {
+      "epoch": 0.0950696100299977,
+      "learning_rate": 0.0029335930340039554,
+      "loss": 1.132,
+      "step": 1236
+    },
+    {
+      "epoch": 0.09514652719021614,
+      "learning_rate": 0.002933486337482887,
+      "loss": 1.2609,
+      "step": 1237
+    },
+    {
+      "epoch": 0.09522344435043459,
+      "learning_rate": 0.0029333795572590573,
+      "loss": 1.0248,
+      "step": 1238
+    },
+    {
+      "epoch": 0.09530036151065302,
+      "learning_rate": 0.0029332726933387018,
+      "loss": 1.0056,
+      "step": 1239
+    },
+    {
+      "epoch": 0.09537727867087147,
+      "learning_rate": 0.0029331657457280593,
+      "loss": 1.3739,
+      "step": 1240
+    },
+    {
+      "epoch": 0.09545419583108991,
+      "learning_rate": 0.002933058714433376,
+      "loss": 1.1242,
+      "step": 1241
+    },
+    {
+      "epoch": 0.09553111299130836,
+      "learning_rate": 0.0029329515994609007,
+      "loss": 1.238,
+      "step": 1242
+    },
+    {
+      "epoch": 0.0956080301515268,
+      "learning_rate": 0.002932844400816888,
+      "loss": 1.7216,
+      "step": 1243
+    },
+    {
+      "epoch": 0.09568494731174525,
+      "learning_rate": 0.0029327371185075973,
+      "loss": 1.3232,
+      "step": 1244
+    },
+    {
+      "epoch": 0.09576186447196369,
+      "learning_rate": 0.0029326297525392934,
+      "loss": 0.8693,
+      "step": 1245
+    },
+    {
+      "epoch": 0.09583878163218214,
+      "learning_rate": 0.002932522302918245,
+      "loss": 1.1943,
+      "step": 1246
+    },
+    {
+      "epoch": 0.09591569879240058,
+      "learning_rate": 0.0029324147696507268,
+      "loss": 1.289,
+      "step": 1247
+    },
+    {
+      "epoch": 0.09599261595261903,
+      "learning_rate": 0.002932307152743017,
+      "loss": 0.98,
+      "step": 1248
+    },
+    {
+      "epoch": 0.09606953311283747,
+      "learning_rate": 0.0029321994522013995,
+      "loss": 1.4621,
+      "step": 1249
+    },
+    {
+      "epoch": 0.09614645027305592,
+      "learning_rate": 0.002932091668032164,
+      "loss": 1.1523,
+      "step": 1250
+    },
+    {
+      "epoch": 0.09622336743327436,
+      "learning_rate": 0.0029319838002416034,
+      "loss": 1.1686,
+      "step": 1251
+    },
+    {
+      "epoch": 0.09630028459349281,
+      "learning_rate": 0.0029318758488360163,
+      "loss": 1.2622,
+      "step": 1252
+    },
+    {
+      "epoch": 0.09637720175371126,
+      "learning_rate": 0.002931767813821706,
+      "loss": 1.2453,
+      "step": 1253
+    },
+    {
+      "epoch": 0.0964541189139297,
+      "learning_rate": 0.002931659695204981,
+      "loss": 1.4202,
+      "step": 1254
+    },
+    {
+      "epoch": 0.09653103607414815,
+      "learning_rate": 0.002931551492992154,
+      "loss": 1.1503,
+      "step": 1255
+    },
+    {
+      "epoch": 0.09660795323436659,
+      "learning_rate": 0.002931443207189544,
+      "loss": 1.1302,
+      "step": 1256
+    },
+    {
+      "epoch": 0.09668487039458504,
+      "learning_rate": 0.002931334837803473,
+      "loss": 1.2203,
+      "step": 1257
+    },
+    {
+      "epoch": 0.09676178755480348,
+      "learning_rate": 0.002931226384840269,
+      "loss": 1.2188,
+      "step": 1258
+    },
+    {
+      "epoch": 0.09683870471502193,
+      "learning_rate": 0.002931117848306265,
+      "loss": 1.0731,
+      "step": 1259
+    },
+    {
+      "epoch": 0.09691562187524036,
+      "learning_rate": 0.002931009228207798,
+      "loss": 1.2517,
+      "step": 1260
+    },
+    {
+      "epoch": 0.0969925390354588,
+      "learning_rate": 0.0029309005245512117,
+      "loss": 1.1722,
+      "step": 1261
+    },
+    {
+      "epoch": 0.09706945619567725,
+      "learning_rate": 0.0029307917373428516,
+      "loss": 1.2239,
+      "step": 1262
+    },
+    {
+      "epoch": 0.0971463733558957,
+      "learning_rate": 0.002930682866589071,
+      "loss": 0.8266,
+      "step": 1263
+    },
+    {
+      "epoch": 0.09722329051611414,
+      "learning_rate": 0.002930573912296227,
+      "loss": 1.2538,
+      "step": 1264
+    },
+    {
+      "epoch": 0.09730020767633259,
+      "learning_rate": 0.002930464874470681,
+      "loss": 1.2797,
+      "step": 1265
+    },
+    {
+      "epoch": 0.09737712483655103,
+      "learning_rate": 0.0029303557531188008,
+      "loss": 1.3732,
+      "step": 1266
+    },
+    {
+      "epoch": 0.09745404199676948,
+      "learning_rate": 0.0029302465482469574,
+      "loss": 1.1501,
+      "step": 1267
+    },
+    {
+      "epoch": 0.09753095915698792,
+      "learning_rate": 0.0029301372598615275,
+      "loss": 1.3214,
+      "step": 1268
+    },
+    {
+      "epoch": 0.09760787631720637,
+      "learning_rate": 0.002930027887968892,
+      "loss": 0.9647,
+      "step": 1269
+    },
+    {
+      "epoch": 0.09768479347742481,
+      "learning_rate": 0.002929918432575438,
+      "loss": 1.1907,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09776171063764326,
+      "learning_rate": 0.002929808893687556,
+      "loss": 0.8753,
+      "step": 1271
+    },
+    {
+      "epoch": 0.0978386277978617,
+      "learning_rate": 0.0029296992713116433,
+      "loss": 1.0652,
+      "step": 1272
+    },
+    {
+      "epoch": 0.09791554495808015,
+      "learning_rate": 0.0029295895654541004,
+      "loss": 0.9914,
+      "step": 1273
+    },
+    {
+      "epoch": 0.0979924621182986,
+      "learning_rate": 0.0029294797761213326,
+      "loss": 1.0565,
+      "step": 1274
+    },
+    {
+      "epoch": 0.09806937927851704,
+      "learning_rate": 0.0029293699033197504,
+      "loss": 1.1092,
+      "step": 1275
+    },
+    {
+      "epoch": 0.09814629643873549,
+      "learning_rate": 0.0029292599470557706,
+      "loss": 1.1121,
+      "step": 1276
+    },
+    {
+      "epoch": 0.09822321359895393,
+      "learning_rate": 0.0029291499073358126,
+      "loss": 1.0894,
+      "step": 1277
+    },
+    {
+      "epoch": 0.09830013075917238,
+      "learning_rate": 0.002929039784166302,
+      "loss": 0.7796,
+      "step": 1278
+    },
+    {
+      "epoch": 0.09837704791939082,
+      "learning_rate": 0.0029289295775536696,
+      "loss": 1.0671,
+      "step": 1279
+    },
+    {
+      "epoch": 0.09845396507960925,
+      "learning_rate": 0.0029288192875043494,
+      "loss": 0.8831,
+      "step": 1280
+    },
+    {
+      "epoch": 0.0985308822398277,
+      "learning_rate": 0.0029287089140247824,
+      "loss": 0.9901,
+      "step": 1281
+    },
+    {
+      "epoch": 0.09860779940004614,
+      "learning_rate": 0.0029285984571214125,
+      "loss": 1.1922,
+      "step": 1282
+    },
+    {
+      "epoch": 0.09868471656026459,
+      "learning_rate": 0.0029284879168006896,
+      "loss": 1.2796,
+      "step": 1283
+    },
+    {
+      "epoch": 0.09876163372048304,
+      "learning_rate": 0.0029283772930690692,
+      "loss": 0.9585,
+      "step": 1284
+    },
+    {
+      "epoch": 0.09883855088070148,
+      "learning_rate": 0.0029282665859330097,
+      "loss": 1.2213,
+      "step": 1285
+    },
+    {
+      "epoch": 0.09891546804091993,
+      "learning_rate": 0.002928155795398976,
+      "loss": 1.2803,
+      "step": 1286
+    },
+    {
+      "epoch": 0.09899238520113837,
+      "learning_rate": 0.0029280449214734365,
+      "loss": 1.4555,
+      "step": 1287
+    },
+    {
+      "epoch": 0.09906930236135682,
+      "learning_rate": 0.002927933964162866,
+      "loss": 1.3525,
+      "step": 1288
+    },
+    {
+      "epoch": 0.09914621952157526,
+      "learning_rate": 0.0029278229234737433,
+      "loss": 1.4479,
+      "step": 1289
+    },
+    {
+      "epoch": 0.09922313668179371,
+      "learning_rate": 0.002927711799412552,
+      "loss": 0.7878,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09930005384201215,
+      "learning_rate": 0.002927600591985781,
+      "loss": 0.7441,
+      "step": 1291
+    },
+    {
+      "epoch": 0.0993769710022306,
+      "learning_rate": 0.0029274893011999236,
+      "loss": 1.5054,
+      "step": 1292
+    },
+    {
+      "epoch": 0.09945388816244904,
+      "learning_rate": 0.0029273779270614776,
+      "loss": 1.2258,
+      "step": 1293
+    },
+    {
+      "epoch": 0.09953080532266749,
+      "learning_rate": 0.002927266469576947,
+      "loss": 1.2355,
+      "step": 1294
+    },
+    {
+      "epoch": 0.09960772248288594,
+      "learning_rate": 0.0029271549287528406,
+      "loss": 0.9092,
+      "step": 1295
+    },
+    {
+      "epoch": 0.09968463964310438,
+      "learning_rate": 0.00292704330459567,
+      "loss": 1.1802,
+      "step": 1296
+    },
+    {
+      "epoch": 0.09976155680332283,
+      "learning_rate": 0.002926931597111954,
+      "loss": 1.2831,
+      "step": 1297
+    },
+    {
+      "epoch": 0.09983847396354127,
+      "learning_rate": 0.0029268198063082144,
+      "loss": 1.1867,
+      "step": 1298
+    },
+    {
+      "epoch": 0.09991539112375972,
+      "learning_rate": 0.0029267079321909795,
+      "loss": 1.1651,
+      "step": 1299
+    },
+    {
+      "epoch": 0.09999230828397816,
+      "learning_rate": 0.0029265959747667813,
+      "loss": 1.1498,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1000692254441966,
+      "learning_rate": 0.002926483934042158,
+      "loss": 1.1721,
+      "step": 1301
+    },
+    {
+      "epoch": 0.10014614260441504,
+      "learning_rate": 0.0029263718100236507,
+      "loss": 1.0412,
+      "step": 1302
+    },
+    {
+      "epoch": 0.10022305976463348,
+      "learning_rate": 0.0029262596027178067,
+      "loss": 1.5262,
+      "step": 1303
+    },
+    {
+      "epoch": 0.10029997692485193,
+      "learning_rate": 0.0029261473121311786,
+      "loss": 1.2988,
+      "step": 1304
+    },
+    {
+      "epoch": 0.10037689408507038,
+      "learning_rate": 0.002926034938270322,
+      "loss": 0.9277,
+      "step": 1305
+    },
+    {
+      "epoch": 0.10045381124528882,
+      "learning_rate": 0.0029259224811418,
+      "loss": 1.4692,
+      "step": 1306
+    },
+    {
+      "epoch": 0.10053072840550727,
+      "learning_rate": 0.0029258099407521777,
+      "loss": 1.1353,
+      "step": 1307
+    },
+    {
+      "epoch": 0.10060764556572571,
+      "learning_rate": 0.0029256973171080274,
+      "loss": 1.0995,
+      "step": 1308
+    },
+    {
+      "epoch": 0.10068456272594416,
+      "learning_rate": 0.0029255846102159244,
+      "loss": 1.1535,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1007614798861626,
+      "learning_rate": 0.002925471820082451,
+      "loss": 1.4002,
+      "step": 1310
+    },
+    {
+      "epoch": 0.10083839704638105,
+      "learning_rate": 0.0029253589467141917,
+      "loss": 1.2245,
+      "step": 1311
+    },
+    {
+      "epoch": 0.1009153142065995,
+      "learning_rate": 0.0029252459901177386,
+      "loss": 1.2784,
+      "step": 1312
+    },
+    {
+      "epoch": 0.10099223136681794,
+      "learning_rate": 0.0029251329502996866,
+      "loss": 1.3721,
+      "step": 1313
+    },
+    {
+      "epoch": 0.10106914852703638,
+      "learning_rate": 0.0029250198272666363,
+      "loss": 1.288,
+      "step": 1314
+    },
+    {
+      "epoch": 0.10114606568725483,
+      "learning_rate": 0.002924906621025193,
+      "loss": 1.091,
+      "step": 1315
+    },
+    {
+      "epoch": 0.10122298284747328,
+      "learning_rate": 0.0029247933315819674,
+      "loss": 0.844,
+      "step": 1316
+    },
+    {
+      "epoch": 0.10129990000769172,
+      "learning_rate": 0.0029246799589435738,
+      "loss": 0.9985,
+      "step": 1317
+    },
+    {
+      "epoch": 0.10137681716791017,
+      "learning_rate": 0.0029245665031166334,
+      "loss": 1.2714,
+      "step": 1318
+    },
+    {
+      "epoch": 0.10145373432812861,
+      "learning_rate": 0.00292445296410777,
+      "loss": 1.3008,
+      "step": 1319
+    },
+    {
+      "epoch": 0.10153065148834706,
+      "learning_rate": 0.0029243393419236133,
+      "loss": 1.233,
+      "step": 1320
+    },
+    {
+      "epoch": 0.10160756864856549,
+      "learning_rate": 0.002924225636570798,
+      "loss": 0.8218,
+      "step": 1321
+    },
+    {
+      "epoch": 0.10168448580878393,
+      "learning_rate": 0.0029241118480559636,
+      "loss": 1.3125,
+      "step": 1322
+    },
+    {
+      "epoch": 0.10176140296900238,
+      "learning_rate": 0.002923997976385754,
+      "loss": 1.3914,
+      "step": 1323
+    },
+    {
+      "epoch": 0.10183832012922082,
+      "learning_rate": 0.0029238840215668185,
+      "loss": 1.1458,
+      "step": 1324
+    },
+    {
+      "epoch": 0.10191523728943927,
+      "learning_rate": 0.0029237699836058115,
+      "loss": 1.2458,
+      "step": 1325
+    },
+    {
+      "epoch": 0.10199215444965772,
+      "learning_rate": 0.0029236558625093906,
+      "loss": 1.1755,
+      "step": 1326
+    },
+    {
+      "epoch": 0.10206907160987616,
+      "learning_rate": 0.002923541658284221,
+      "loss": 1.1569,
+      "step": 1327
+    },
+    {
+      "epoch": 0.1021459887700946,
+      "learning_rate": 0.00292342737093697,
+      "loss": 1.127,
+      "step": 1328
+    },
+    {
+      "epoch": 0.10222290593031305,
+      "learning_rate": 0.0029233130004743114,
+      "loss": 1.0872,
+      "step": 1329
+    },
+    {
+      "epoch": 0.1022998230905315,
+      "learning_rate": 0.0029231985469029232,
+      "loss": 1.2195,
+      "step": 1330
+    },
+    {
+      "epoch": 0.10237674025074994,
+      "learning_rate": 0.002923084010229489,
+      "loss": 1.3099,
+      "step": 1331
+    },
+    {
+      "epoch": 0.10245365741096839,
+      "learning_rate": 0.002922969390460696,
+      "loss": 1.4398,
+      "step": 1332
+    },
+    {
+      "epoch": 0.10253057457118683,
+      "learning_rate": 0.0029228546876032373,
+      "loss": 0.9453,
+      "step": 1333
+    },
+    {
+      "epoch": 0.10260749173140528,
+      "learning_rate": 0.002922739901663811,
+      "loss": 1.2617,
+      "step": 1334
+    },
+    {
+      "epoch": 0.10268440889162372,
+      "learning_rate": 0.0029226250326491185,
+      "loss": 1.1086,
+      "step": 1335
+    },
+    {
+      "epoch": 0.10276132605184217,
+      "learning_rate": 0.002922510080565868,
+      "loss": 1.055,
+      "step": 1336
+    },
+    {
+      "epoch": 0.10283824321206061,
+      "learning_rate": 0.0029223950454207713,
+      "loss": 1.0562,
+      "step": 1337
+    },
+    {
+      "epoch": 0.10291516037227906,
+      "learning_rate": 0.002922279927220546,
+      "loss": 0.9574,
+      "step": 1338
+    },
+    {
+      "epoch": 0.1029920775324975,
+      "learning_rate": 0.002922164725971913,
+      "loss": 0.9332,
+      "step": 1339
+    },
+    {
+      "epoch": 0.10306899469271595,
+      "learning_rate": 0.0029220494416815996,
+      "loss": 1.2528,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1031459118529344,
+      "learning_rate": 0.002921934074356337,
+      "loss": 1.23,
+      "step": 1341
+    },
+    {
+      "epoch": 0.10322282901315283,
+      "learning_rate": 0.002921818624002862,
+      "loss": 1.1342,
+      "step": 1342
+    },
+    {
+      "epoch": 0.10329974617337127,
+      "learning_rate": 0.002921703090627916,
+      "loss": 1.157,
+      "step": 1343
+    },
+    {
+      "epoch": 0.10337666333358972,
+      "learning_rate": 0.002921587474238245,
+      "loss": 0.9476,
+      "step": 1344
+    },
+    {
+      "epoch": 0.10345358049380816,
+      "learning_rate": 0.0029214717748405994,
+      "loss": 1.3534,
+      "step": 1345
+    },
+    {
+      "epoch": 0.10353049765402661,
+      "learning_rate": 0.0029213559924417356,
+      "loss": 0.888,
+      "step": 1346
+    },
+    {
+      "epoch": 0.10360741481424506,
+      "learning_rate": 0.002921240127048414,
+      "loss": 0.6896,
+      "step": 1347
+    },
+    {
+      "epoch": 0.1036843319744635,
+      "learning_rate": 0.0029211241786674005,
+      "loss": 1.3125,
+      "step": 1348
+    },
+    {
+      "epoch": 0.10376124913468195,
+      "learning_rate": 0.0029210081473054652,
+      "loss": 1.1898,
+      "step": 1349
+    },
+    {
+      "epoch": 0.10383816629490039,
+      "learning_rate": 0.002920892032969383,
+      "loss": 1.3149,
+      "step": 1350
+    },
+    {
+      "epoch": 0.10391508345511884,
+      "learning_rate": 0.0029207758356659336,
+      "loss": 1.0003,
+      "step": 1351
+    },
+    {
+      "epoch": 0.10399200061533728,
+      "learning_rate": 0.0029206595554019037,
+      "loss": 1.3983,
+      "step": 1352
+    },
+    {
+      "epoch": 0.10406891777555573,
+      "learning_rate": 0.002920543192184081,
+      "loss": 1.2183,
+      "step": 1353
+    },
+    {
+      "epoch": 0.10414583493577417,
+      "learning_rate": 0.002920426746019261,
+      "loss": 0.9984,
+      "step": 1354
+    },
+    {
+      "epoch": 0.10422275209599262,
+      "learning_rate": 0.002920310216914243,
+      "loss": 1.2361,
+      "step": 1355
+    },
+    {
+      "epoch": 0.10429966925621106,
+      "learning_rate": 0.002920193604875831,
+      "loss": 1.258,
+      "step": 1356
+    },
+    {
+      "epoch": 0.10437658641642951,
+      "learning_rate": 0.0029200769099108344,
+      "loss": 0.6998,
+      "step": 1357
+    },
+    {
+      "epoch": 0.10445350357664795,
+      "learning_rate": 0.0029199601320260673,
+      "loss": 1.3332,
+      "step": 1358
+    },
+    {
+      "epoch": 0.1045304207368664,
+      "learning_rate": 0.002919843271228348,
+      "loss": 1.24,
+      "step": 1359
+    },
+    {
+      "epoch": 0.10460733789708485,
+      "learning_rate": 0.0029197263275245007,
+      "loss": 1.3677,
+      "step": 1360
+    },
+    {
+      "epoch": 0.10468425505730329,
+      "learning_rate": 0.0029196093009213535,
+      "loss": 1.3422,
+      "step": 1361
+    },
+    {
+      "epoch": 0.10476117221752172,
+      "learning_rate": 0.0029194921914257395,
+      "loss": 0.9152,
+      "step": 1362
+    },
+    {
+      "epoch": 0.10483808937774017,
+      "learning_rate": 0.002919374999044497,
+      "loss": 1.4853,
+      "step": 1363
+    },
+    {
+      "epoch": 0.10491500653795861,
+      "learning_rate": 0.002919257723784469,
+      "loss": 0.9956,
+      "step": 1364
+    },
+    {
+      "epoch": 0.10499192369817706,
+      "learning_rate": 0.0029191403656525037,
+      "loss": 1.1749,
+      "step": 1365
+    },
+    {
+      "epoch": 0.1050688408583955,
+      "learning_rate": 0.0029190229246554536,
+      "loss": 1.02,
+      "step": 1366
+    },
+    {
+      "epoch": 0.10514575801861395,
+      "learning_rate": 0.002918905400800176,
+      "loss": 1.0859,
+      "step": 1367
+    },
+    {
+      "epoch": 0.1052226751788324,
+      "learning_rate": 0.0029187877940935326,
+      "loss": 1.2528,
+      "step": 1368
+    },
+    {
+      "epoch": 0.10529959233905084,
+      "learning_rate": 0.002918670104542392,
+      "loss": 1.1351,
+      "step": 1369
+    },
+    {
+      "epoch": 0.10537650949926929,
+      "learning_rate": 0.0029185523321536254,
+      "loss": 0.9711,
+      "step": 1370
+    },
+    {
+      "epoch": 0.10545342665948773,
+      "learning_rate": 0.0029184344769341095,
+      "loss": 1.2338,
+      "step": 1371
+    },
+    {
+      "epoch": 0.10553034381970618,
+      "learning_rate": 0.0029183165388907265,
+      "loss": 0.9078,
+      "step": 1372
+    },
+    {
+      "epoch": 0.10560726097992462,
+      "learning_rate": 0.0029181985180303624,
+      "loss": 0.9939,
+      "step": 1373
+    },
+    {
+      "epoch": 0.10568417814014307,
+      "learning_rate": 0.002918080414359909,
+      "loss": 0.8116,
+      "step": 1374
+    },
+    {
+      "epoch": 0.10576109530036151,
+      "learning_rate": 0.0029179622278862625,
+      "loss": 1.3753,
+      "step": 1375
+    },
+    {
+      "epoch": 0.10583801246057996,
+      "learning_rate": 0.0029178439586163233,
+      "loss": 1.0247,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1059149296207984,
+      "learning_rate": 0.002917725606556998,
+      "loss": 1.6761,
+      "step": 1377
+    },
+    {
+      "epoch": 0.10599184678101685,
+      "learning_rate": 0.0029176071717151968,
+      "loss": 0.9708,
+      "step": 1378
+    },
+    {
+      "epoch": 0.1060687639412353,
+      "learning_rate": 0.0029174886540978356,
+      "loss": 1.2245,
+      "step": 1379
+    },
+    {
+      "epoch": 0.10614568110145374,
+      "learning_rate": 0.0029173700537118348,
+      "loss": 1.2728,
+      "step": 1380
+    },
+    {
+      "epoch": 0.10622259826167219,
+      "learning_rate": 0.002917251370564119,
+      "loss": 1.1181,
+      "step": 1381
+    },
+    {
+      "epoch": 0.10629951542189062,
+      "learning_rate": 0.002917132604661619,
+      "loss": 0.8526,
+      "step": 1382
+    },
+    {
+      "epoch": 0.10637643258210906,
+      "learning_rate": 0.002917013756011269,
+      "loss": 1.185,
+      "step": 1383
+    },
+    {
+      "epoch": 0.10645334974232751,
+      "learning_rate": 0.002916894824620009,
+      "loss": 1.3072,
+      "step": 1384
+    },
+    {
+      "epoch": 0.10653026690254595,
+      "learning_rate": 0.0029167758104947845,
+      "loss": 1.5782,
+      "step": 1385
+    },
+    {
+      "epoch": 0.1066071840627644,
+      "learning_rate": 0.0029166567136425427,
+      "loss": 1.4001,
+      "step": 1386
+    },
+    {
+      "epoch": 0.10668410122298284,
+      "learning_rate": 0.0029165375340702393,
+      "loss": 0.7133,
+      "step": 1387
+    },
+    {
+      "epoch": 0.10676101838320129,
+      "learning_rate": 0.002916418271784833,
+      "loss": 1.149,
+      "step": 1388
+    },
+    {
+      "epoch": 0.10683793554341973,
+      "learning_rate": 0.0029162989267932883,
+      "loss": 1.1076,
+      "step": 1389
+    },
+    {
+      "epoch": 0.10691485270363818,
+      "learning_rate": 0.0029161794991025723,
+      "loss": 1.0589,
+      "step": 1390
+    },
+    {
+      "epoch": 0.10699176986385663,
+      "learning_rate": 0.00291605998871966,
+      "loss": 0.9592,
+      "step": 1391
+    },
+    {
+      "epoch": 0.10706868702407507,
+      "learning_rate": 0.002915940395651529,
+      "loss": 1.1013,
+      "step": 1392
+    },
+    {
+      "epoch": 0.10714560418429352,
+      "learning_rate": 0.002915820719905163,
+      "loss": 1.2462,
+      "step": 1393
+    },
+    {
+      "epoch": 0.10722252134451196,
+      "learning_rate": 0.002915700961487549,
+      "loss": 0.9328,
+      "step": 1394
+    },
+    {
+      "epoch": 0.10729943850473041,
+      "learning_rate": 0.0029155811204056806,
+      "loss": 1.3122,
+      "step": 1395
+    },
+    {
+      "epoch": 0.10737635566494885,
+      "learning_rate": 0.0029154611966665557,
+      "loss": 1.1443,
+      "step": 1396
+    },
+    {
+      "epoch": 0.1074532728251673,
+      "learning_rate": 0.002915341190277176,
+      "loss": 0.9814,
+      "step": 1397
+    },
+    {
+      "epoch": 0.10753018998538574,
+      "learning_rate": 0.0029152211012445496,
+      "loss": 1.0536,
+      "step": 1398
+    },
+    {
+      "epoch": 0.10760710714560419,
+      "learning_rate": 0.002915100929575688,
+      "loss": 1.1837,
+      "step": 1399
+    },
+    {
+      "epoch": 0.10768402430582263,
+      "learning_rate": 0.002914980675277608,
+      "loss": 1.275,
+      "step": 1400
+    },
+    {
+      "epoch": 0.10776094146604108,
+      "learning_rate": 0.002914860338357332,
+      "loss": 1.1234,
+      "step": 1401
+    },
+    {
+      "epoch": 0.10783785862625953,
+      "learning_rate": 0.002914739918821886,
+      "loss": 1.1729,
+      "step": 1402
+    },
+    {
+      "epoch": 0.10791477578647796,
+      "learning_rate": 0.002914619416678302,
+      "loss": 0.9903,
+      "step": 1403
+    },
+    {
+      "epoch": 0.1079916929466964,
+      "learning_rate": 0.002914498831933616,
+      "loss": 1.1748,
+      "step": 1404
+    },
+    {
+      "epoch": 0.10806861010691485,
+      "learning_rate": 0.002914378164594869,
+      "loss": 1.4379,
+      "step": 1405
+    },
+    {
+      "epoch": 0.10814552726713329,
+      "learning_rate": 0.002914257414669107,
+      "loss": 0.8967,
+      "step": 1406
+    },
+    {
+      "epoch": 0.10822244442735174,
+      "learning_rate": 0.0029141365821633808,
+      "loss": 0.9649,
+      "step": 1407
+    },
+    {
+      "epoch": 0.10829936158757018,
+      "learning_rate": 0.0029140156670847454,
+      "loss": 1.4865,
+      "step": 1408
+    },
+    {
+      "epoch": 0.10837627874778863,
+      "learning_rate": 0.0029138946694402617,
+      "loss": 0.5231,
+      "step": 1409
+    },
+    {
+      "epoch": 0.10845319590800707,
+      "learning_rate": 0.002913773589236995,
+      "loss": 1.3454,
+      "step": 1410
+    },
+    {
+      "epoch": 0.10853011306822552,
+      "learning_rate": 0.0029136524264820147,
+      "loss": 1.2543,
+      "step": 1411
+    },
+    {
+      "epoch": 0.10860703022844397,
+      "learning_rate": 0.0029135311811823955,
+      "loss": 1.2049,
+      "step": 1412
+    },
+    {
+      "epoch": 0.10868394738866241,
+      "learning_rate": 0.0029134098533452183,
+      "loss": 1.1168,
+      "step": 1413
+    },
+    {
+      "epoch": 0.10876086454888086,
+      "learning_rate": 0.0029132884429775665,
+      "loss": 0.8141,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1088377817090993,
+      "learning_rate": 0.00291316695008653,
+      "loss": 1.2728,
+      "step": 1415
+    },
+    {
+      "epoch": 0.10891469886931775,
+      "learning_rate": 0.0029130453746792017,
+      "loss": 0.832,
+      "step": 1416
+    },
+    {
+      "epoch": 0.10899161602953619,
+      "learning_rate": 0.0029129237167626817,
+      "loss": 1.3316,
+      "step": 1417
+    },
+    {
+      "epoch": 0.10906853318975464,
+      "learning_rate": 0.0029128019763440733,
+      "loss": 1.3856,
+      "step": 1418
+    },
+    {
+      "epoch": 0.10914545034997308,
+      "learning_rate": 0.0029126801534304853,
+      "loss": 1.1343,
+      "step": 1419
+    },
+    {
+      "epoch": 0.10922236751019153,
+      "learning_rate": 0.002912558248029031,
+      "loss": 0.9837,
+      "step": 1420
+    },
+    {
+      "epoch": 0.10929928467040997,
+      "learning_rate": 0.002912436260146828,
+      "loss": 1.0679,
+      "step": 1421
+    },
+    {
+      "epoch": 0.10937620183062842,
+      "learning_rate": 0.002912314189791,
+      "loss": 1.2493,
+      "step": 1422
+    },
+    {
+      "epoch": 0.10945311899084685,
+      "learning_rate": 0.0029121920369686747,
+      "loss": 0.9199,
+      "step": 1423
+    },
+    {
+      "epoch": 0.1095300361510653,
+      "learning_rate": 0.0029120698016869844,
+      "loss": 1.4005,
+      "step": 1424
+    },
+    {
+      "epoch": 0.10960695331128374,
+      "learning_rate": 0.002911947483953067,
+      "loss": 1.2883,
+      "step": 1425
+    },
+    {
+      "epoch": 0.10968387047150219,
+      "learning_rate": 0.0029118250837740643,
+      "loss": 1.1278,
+      "step": 1426
+    },
+    {
+      "epoch": 0.10976078763172063,
+      "learning_rate": 0.0029117026011571236,
+      "loss": 1.1546,
+      "step": 1427
+    },
+    {
+      "epoch": 0.10983770479193908,
+      "learning_rate": 0.002911580036109397,
+      "loss": 1.0478,
+      "step": 1428
+    },
+    {
+      "epoch": 0.10991462195215752,
+      "learning_rate": 0.0029114573886380407,
+      "loss": 1.1066,
+      "step": 1429
+    },
+    {
+      "epoch": 0.10999153911237597,
+      "learning_rate": 0.0029113346587502165,
+      "loss": 1.1094,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11006845627259441,
+      "learning_rate": 0.002911211846453091,
+      "loss": 0.882,
+      "step": 1431
+    },
+    {
+      "epoch": 0.11014537343281286,
+      "learning_rate": 0.0029110889517538346,
+      "loss": 1.0255,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1102222905930313,
+      "learning_rate": 0.002910965974659624,
+      "loss": 1.2265,
+      "step": 1433
+    },
+    {
+      "epoch": 0.11029920775324975,
+      "learning_rate": 0.00291084291517764,
+      "loss": 1.204,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1103761249134682,
+      "learning_rate": 0.0029107197733150673,
+      "loss": 1.0782,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11045304207368664,
+      "learning_rate": 0.002910596549079097,
+      "loss": 0.9163,
+      "step": 1436
+    },
+    {
+      "epoch": 0.11052995923390509,
+      "learning_rate": 0.002910473242476924,
+      "loss": 1.0607,
+      "step": 1437
+    },
+    {
+      "epoch": 0.11060687639412353,
+      "learning_rate": 0.002910349853515749,
+      "loss": 1.1778,
+      "step": 1438
+    },
+    {
+      "epoch": 0.11068379355434198,
+      "learning_rate": 0.0029102263822027756,
+      "loss": 1.3244,
+      "step": 1439
+    },
+    {
+      "epoch": 0.11076071071456042,
+      "learning_rate": 0.002910102828545214,
+      "loss": 1.2257,
+      "step": 1440
+    },
+    {
+      "epoch": 0.11083762787477887,
+      "learning_rate": 0.0029099791925502787,
+      "loss": 1.2334,
+      "step": 1441
+    },
+    {
+      "epoch": 0.11091454503499731,
+      "learning_rate": 0.0029098554742251893,
+      "loss": 1.2848,
+      "step": 1442
+    },
+    {
+      "epoch": 0.11099146219521576,
+      "learning_rate": 0.002909731673577169,
+      "loss": 0.6049,
+      "step": 1443
+    },
+    {
+      "epoch": 0.11106837935543419,
+      "learning_rate": 0.002909607790613448,
+      "loss": 0.5912,
+      "step": 1444
+    },
+    {
+      "epoch": 0.11114529651565264,
+      "learning_rate": 0.002909483825341258,
+      "loss": 0.964,
+      "step": 1445
+    },
+    {
+      "epoch": 0.11122221367587108,
+      "learning_rate": 0.0029093597777678387,
+      "loss": 0.9893,
+      "step": 1446
+    },
+    {
+      "epoch": 0.11129913083608953,
+      "learning_rate": 0.0029092356479004334,
+      "loss": 1.3322,
+      "step": 1447
+    },
+    {
+      "epoch": 0.11137604799630797,
+      "learning_rate": 0.00290911143574629,
+      "loss": 1.5149,
+      "step": 1448
+    },
+    {
+      "epoch": 0.11145296515652642,
+      "learning_rate": 0.0029089871413126613,
+      "loss": 1.0509,
+      "step": 1449
+    },
+    {
+      "epoch": 0.11152988231674486,
+      "learning_rate": 0.002908862764606805,
+      "loss": 1.0904,
+      "step": 1450
+    },
+    {
+      "epoch": 0.11160679947696331,
+      "learning_rate": 0.0029087383056359832,
+      "loss": 1.0722,
+      "step": 1451
+    },
+    {
+      "epoch": 0.11168371663718175,
+      "learning_rate": 0.002908613764407464,
+      "loss": 1.4654,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1117606337974002,
+      "learning_rate": 0.0029084891409285186,
+      "loss": 0.9479,
+      "step": 1453
+    },
+    {
+      "epoch": 0.11183755095761864,
+      "learning_rate": 0.002908364435206425,
+      "loss": 1.4954,
+      "step": 1454
+    },
+    {
+      "epoch": 0.11191446811783709,
+      "learning_rate": 0.002908239647248464,
+      "loss": 1.2428,
+      "step": 1455
+    },
+    {
+      "epoch": 0.11199138527805554,
+      "learning_rate": 0.0029081147770619225,
+      "loss": 0.7893,
+      "step": 1456
+    },
+    {
+      "epoch": 0.11206830243827398,
+      "learning_rate": 0.0029079898246540917,
+      "loss": 1.233,
+      "step": 1457
+    },
+    {
+      "epoch": 0.11214521959849243,
+      "learning_rate": 0.0029078647900322675,
+      "loss": 1.2769,
+      "step": 1458
+    },
+    {
+      "epoch": 0.11222213675871087,
+      "learning_rate": 0.002907739673203751,
+      "loss": 0.9771,
+      "step": 1459
+    },
+    {
+      "epoch": 0.11229905391892932,
+      "learning_rate": 0.0029076144741758477,
+      "loss": 1.0689,
+      "step": 1460
+    },
+    {
+      "epoch": 0.11237597107914776,
+      "learning_rate": 0.0029074891929558684,
+      "loss": 1.2342,
+      "step": 1461
+    },
+    {
+      "epoch": 0.11245288823936621,
+      "learning_rate": 0.0029073638295511284,
+      "loss": 1.099,
+      "step": 1462
+    },
+    {
+      "epoch": 0.11252980539958465,
+      "learning_rate": 0.0029072383839689477,
+      "loss": 1.1543,
+      "step": 1463
+    },
+    {
+      "epoch": 0.11260672255980309,
+      "learning_rate": 0.002907112856216651,
+      "loss": 1.1191,
+      "step": 1464
+    },
+    {
+      "epoch": 0.11268363972002153,
+      "learning_rate": 0.0029069872463015674,
+      "loss": 1.0351,
+      "step": 1465
+    },
+    {
+      "epoch": 0.11276055688023998,
+      "learning_rate": 0.002906861554231033,
+      "loss": 1.2029,
+      "step": 1466
+    },
+    {
+      "epoch": 0.11283747404045842,
+      "learning_rate": 0.0029067357800123864,
+      "loss": 1.2132,
+      "step": 1467
+    },
+    {
+      "epoch": 0.11291439120067687,
+      "learning_rate": 0.002906609923652972,
+      "loss": 0.8952,
+      "step": 1468
+    },
+    {
+      "epoch": 0.11299130836089531,
+      "learning_rate": 0.002906483985160137,
+      "loss": 1.1924,
+      "step": 1469
+    },
+    {
+      "epoch": 0.11306822552111376,
+      "learning_rate": 0.0029063579645412373,
+      "loss": 0.6819,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1131451426813322,
+      "learning_rate": 0.0029062318618036298,
+      "loss": 1.1736,
+      "step": 1471
+    },
+    {
+      "epoch": 0.11322205984155065,
+      "learning_rate": 0.0029061056769546785,
+      "loss": 1.297,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1132989770017691,
+      "learning_rate": 0.0029059794100017512,
+      "loss": 1.1583,
+      "step": 1473
+    },
+    {
+      "epoch": 0.11337589416198754,
+      "learning_rate": 0.002905853060952221,
+      "loss": 1.4155,
+      "step": 1474
+    },
+    {
+      "epoch": 0.11345281132220598,
+      "learning_rate": 0.0029057266298134657,
+      "loss": 1.332,
+      "step": 1475
+    },
+    {
+      "epoch": 0.11352972848242443,
+      "learning_rate": 0.0029056001165928672,
+      "loss": 1.2283,
+      "step": 1476
+    },
+    {
+      "epoch": 0.11360664564264288,
+      "learning_rate": 0.002905473521297813,
+      "loss": 0.999,
+      "step": 1477
+    },
+    {
+      "epoch": 0.11368356280286132,
+      "learning_rate": 0.002905346843935695,
+      "loss": 1.3447,
+      "step": 1478
+    },
+    {
+      "epoch": 0.11376047996307977,
+      "learning_rate": 0.0029052200845139102,
+      "loss": 1.4668,
+      "step": 1479
+    },
+    {
+      "epoch": 0.11383739712329821,
+      "learning_rate": 0.0029050932430398605,
+      "loss": 1.0701,
+      "step": 1480
+    },
+    {
+      "epoch": 0.11391431428351666,
+      "learning_rate": 0.0029049663195209515,
+      "loss": 1.2138,
+      "step": 1481
+    },
+    {
+      "epoch": 0.1139912314437351,
+      "learning_rate": 0.0029048393139645954,
+      "loss": 1.2256,
+      "step": 1482
+    },
+    {
+      "epoch": 0.11406814860395355,
+      "learning_rate": 0.0029047122263782072,
+      "loss": 1.254,
+      "step": 1483
+    },
+    {
+      "epoch": 0.114145065764172,
+      "learning_rate": 0.0029045850567692086,
+      "loss": 0.9742,
+      "step": 1484
+    },
+    {
+      "epoch": 0.11422198292439042,
+      "learning_rate": 0.0029044578051450245,
+      "loss": 1.1314,
+      "step": 1485
+    },
+    {
+      "epoch": 0.11429890008460887,
+      "learning_rate": 0.002904330471513085,
+      "loss": 0.9521,
+      "step": 1486
+    },
+    {
+      "epoch": 0.11437581724482732,
+      "learning_rate": 0.002904203055880826,
+      "loss": 1.2706,
+      "step": 1487
+    },
+    {
+      "epoch": 0.11445273440504576,
+      "learning_rate": 0.0029040755582556877,
+      "loss": 0.8659,
+      "step": 1488
+    },
+    {
+      "epoch": 0.1145296515652642,
+      "learning_rate": 0.002903947978645114,
+      "loss": 1.2421,
+      "step": 1489
+    },
+    {
+      "epoch": 0.11460656872548265,
+      "learning_rate": 0.0029038203170565543,
+      "loss": 1.1359,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1146834858857011,
+      "learning_rate": 0.0029036925734974637,
+      "loss": 0.8697,
+      "step": 1491
+    },
+    {
+      "epoch": 0.11476040304591954,
+      "learning_rate": 0.0029035647479753006,
+      "loss": 1.4948,
+      "step": 1492
+    },
+    {
+      "epoch": 0.11483732020613799,
+      "learning_rate": 0.0029034368404975293,
+      "loss": 1.2606,
+      "step": 1493
+    },
+    {
+      "epoch": 0.11491423736635643,
+      "learning_rate": 0.002903308851071618,
+      "loss": 0.9511,
+      "step": 1494
+    },
+    {
+      "epoch": 0.11499115452657488,
+      "learning_rate": 0.0029031807797050405,
+      "loss": 1.1019,
+      "step": 1495
+    },
+    {
+      "epoch": 0.11506807168679332,
+      "learning_rate": 0.002903052626405275,
+      "loss": 1.3968,
+      "step": 1496
+    },
+    {
+      "epoch": 0.11514498884701177,
+      "learning_rate": 0.002902924391179804,
+      "loss": 0.9133,
+      "step": 1497
+    },
+    {
+      "epoch": 0.11522190600723022,
+      "learning_rate": 0.0029027960740361162,
+      "loss": 1.3496,
+      "step": 1498
+    },
+    {
+      "epoch": 0.11529882316744866,
+      "learning_rate": 0.002902667674981703,
+      "loss": 0.7661,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1153757403276671,
+      "learning_rate": 0.0029025391940240633,
+      "loss": 1.001,
+      "step": 1500
+    },
+    {
+      "epoch": 0.11545265748788555,
+      "learning_rate": 0.0029024106311706984,
+      "loss": 1.0003,
+      "step": 1501
+    },
+    {
+      "epoch": 0.115529574648104,
+      "learning_rate": 0.0029022819864291147,
+      "loss": 1.336,
+      "step": 1502
+    },
+    {
+      "epoch": 0.11560649180832244,
+      "learning_rate": 0.0029021532598068243,
+      "loss": 1.4581,
+      "step": 1503
+    },
+    {
+      "epoch": 0.11568340896854089,
+      "learning_rate": 0.0029020244513113438,
+      "loss": 1.194,
+      "step": 1504
+    },
+    {
+      "epoch": 0.11576032612875932,
+      "learning_rate": 0.002901895560950195,
+      "loss": 0.9644,
+      "step": 1505
+    },
+    {
+      "epoch": 0.11583724328897776,
+      "learning_rate": 0.0029017665887309026,
+      "loss": 1.1786,
+      "step": 1506
+    },
+    {
+      "epoch": 0.11591416044919621,
+      "learning_rate": 0.002901637534660999,
+      "loss": 1.0125,
+      "step": 1507
+    },
+    {
+      "epoch": 0.11599107760941466,
+      "learning_rate": 0.0029015083987480182,
+      "loss": 1.2137,
+      "step": 1508
+    },
+    {
+      "epoch": 0.1160679947696331,
+      "learning_rate": 0.002901379180999502,
+      "loss": 1.1543,
+      "step": 1509
+    },
+    {
+      "epoch": 0.11614491192985155,
+      "learning_rate": 0.0029012498814229943,
+      "loss": 1.0397,
+      "step": 1510
+    },
+    {
+      "epoch": 0.11622182909006999,
+      "learning_rate": 0.0029011205000260463,
+      "loss": 1.5104,
+      "step": 1511
+    },
+    {
+      "epoch": 0.11629874625028844,
+      "learning_rate": 0.002900991036816212,
+      "loss": 1.2308,
+      "step": 1512
+    },
+    {
+      "epoch": 0.11637566341050688,
+      "learning_rate": 0.0029008614918010506,
+      "loss": 1.1723,
+      "step": 1513
+    },
+    {
+      "epoch": 0.11645258057072533,
+      "learning_rate": 0.0029007318649881273,
+      "loss": 1.0011,
+      "step": 1514
+    },
+    {
+      "epoch": 0.11652949773094377,
+      "learning_rate": 0.0029006021563850104,
+      "loss": 1.0574,
+      "step": 1515
+    },
+    {
+      "epoch": 0.11660641489116222,
+      "learning_rate": 0.0029004723659992733,
+      "loss": 1.2372,
+      "step": 1516
+    },
+    {
+      "epoch": 0.11668333205138066,
+      "learning_rate": 0.002900342493838496,
+      "loss": 1.1541,
+      "step": 1517
+    },
+    {
+      "epoch": 0.11676024921159911,
+      "learning_rate": 0.0029002125399102607,
+      "loss": 1.0897,
+      "step": 1518
+    },
+    {
+      "epoch": 0.11683716637181756,
+      "learning_rate": 0.002900082504222156,
+      "loss": 1.3226,
+      "step": 1519
+    },
+    {
+      "epoch": 0.116914083532036,
+      "learning_rate": 0.0028999523867817745,
+      "loss": 1.8294,
+      "step": 1520
+    },
+    {
+      "epoch": 0.11699100069225445,
+      "learning_rate": 0.002899822187596715,
+      "loss": 1.4568,
+      "step": 1521
+    },
+    {
+      "epoch": 0.11706791785247289,
+      "learning_rate": 0.002899691906674578,
+      "loss": 1.2607,
+      "step": 1522
+    },
+    {
+      "epoch": 0.11714483501269134,
+      "learning_rate": 0.0028995615440229722,
+      "loss": 1.2988,
+      "step": 1523
+    },
+    {
+      "epoch": 0.11722175217290978,
+      "learning_rate": 0.0028994310996495096,
+      "loss": 1.024,
+      "step": 1524
+    },
+    {
+      "epoch": 0.11729866933312823,
+      "learning_rate": 0.0028993005735618066,
+      "loss": 1.2166,
+      "step": 1525
+    },
+    {
+      "epoch": 0.11737558649334666,
+      "learning_rate": 0.0028991699657674847,
+      "loss": 1.3122,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1174525036535651,
+      "learning_rate": 0.00289903927627417,
+      "loss": 1.1074,
+      "step": 1527
+    },
+    {
+      "epoch": 0.11752942081378355,
+      "learning_rate": 0.0028989085050894945,
+      "loss": 1.1349,
+      "step": 1528
+    },
+    {
+      "epoch": 0.117606337974002,
+      "learning_rate": 0.002898777652221093,
+      "loss": 0.9543,
+      "step": 1529
+    },
+    {
+      "epoch": 0.11768325513422044,
+      "learning_rate": 0.002898646717676607,
+      "loss": 1.1482,
+      "step": 1530
+    },
+    {
+      "epoch": 0.11776017229443889,
+      "learning_rate": 0.0028985157014636814,
+      "loss": 1.4885,
+      "step": 1531
+    },
+    {
+      "epoch": 0.11783708945465733,
+      "learning_rate": 0.002898384603589966,
+      "loss": 1.0584,
+      "step": 1532
+    },
+    {
+      "epoch": 0.11791400661487578,
+      "learning_rate": 0.0028982534240631166,
+      "loss": 1.3822,
+      "step": 1533
+    },
+    {
+      "epoch": 0.11799092377509422,
+      "learning_rate": 0.0028981221628907924,
+      "loss": 1.074,
+      "step": 1534
+    },
+    {
+      "epoch": 0.11806784093531267,
+      "learning_rate": 0.002897990820080659,
+      "loss": 1.5324,
+      "step": 1535
+    },
+    {
+      "epoch": 0.11814475809553111,
+      "learning_rate": 0.0028978593956403833,
+      "loss": 0.8464,
+      "step": 1536
+    },
+    {
+      "epoch": 0.11822167525574956,
+      "learning_rate": 0.0028977278895776413,
+      "loss": 1.0793,
+      "step": 1537
+    },
+    {
+      "epoch": 0.118298592415968,
+      "learning_rate": 0.002897596301900111,
+      "loss": 1.324,
+      "step": 1538
+    },
+    {
+      "epoch": 0.11837550957618645,
+      "learning_rate": 0.0028974646326154763,
+      "loss": 1.1548,
+      "step": 1539
+    },
+    {
+      "epoch": 0.1184524267364049,
+      "learning_rate": 0.002897332881731425,
+      "loss": 1.3605,
+      "step": 1540
+    },
+    {
+      "epoch": 0.11852934389662334,
+      "learning_rate": 0.0028972010492556506,
+      "loss": 1.192,
+      "step": 1541
+    },
+    {
+      "epoch": 0.11860626105684179,
+      "learning_rate": 0.002897069135195851,
+      "loss": 0.6856,
+      "step": 1542
+    },
+    {
+      "epoch": 0.11868317821706023,
+      "learning_rate": 0.0028969371395597284,
+      "loss": 1.0507,
+      "step": 1543
+    },
+    {
+      "epoch": 0.11876009537727868,
+      "learning_rate": 0.0028968050623549907,
+      "loss": 1.3073,
+      "step": 1544
+    },
+    {
+      "epoch": 0.11883701253749712,
+      "learning_rate": 0.002896672903589349,
+      "loss": 1.2736,
+      "step": 1545
+    },
+    {
+      "epoch": 0.11891392969771555,
+      "learning_rate": 0.0028965406632705214,
+      "loss": 1.3829,
+      "step": 1546
+    },
+    {
+      "epoch": 0.118990846857934,
+      "learning_rate": 0.002896408341406229,
+      "loss": 1.2508,
+      "step": 1547
+    },
+    {
+      "epoch": 0.11906776401815244,
+      "learning_rate": 0.0028962759380041984,
+      "loss": 1.176,
+      "step": 1548
+    },
+    {
+      "epoch": 0.11914468117837089,
+      "learning_rate": 0.00289614345307216,
+      "loss": 1.1064,
+      "step": 1549
+    },
+    {
+      "epoch": 0.11922159833858934,
+      "learning_rate": 0.002896010886617851,
+      "loss": 1.2934,
+      "step": 1550
+    },
+    {
+      "epoch": 0.11929851549880778,
+      "learning_rate": 0.002895878238649011,
+      "loss": 0.9442,
+      "step": 1551
+    },
+    {
+      "epoch": 0.11937543265902623,
+      "learning_rate": 0.002895745509173386,
+      "loss": 1.0226,
+      "step": 1552
+    },
+    {
+      "epoch": 0.11945234981924467,
+      "learning_rate": 0.002895612698198726,
+      "loss": 1.0044,
+      "step": 1553
+    },
+    {
+      "epoch": 0.11952926697946312,
+      "learning_rate": 0.0028954798057327863,
+      "loss": 1.4746,
+      "step": 1554
+    },
+    {
+      "epoch": 0.11960618413968156,
+      "learning_rate": 0.0028953468317833267,
+      "loss": 1.0465,
+      "step": 1555
+    },
+    {
+      "epoch": 0.11968310129990001,
+      "learning_rate": 0.002895213776358111,
+      "loss": 0.9802,
+      "step": 1556
+    },
+    {
+      "epoch": 0.11976001846011845,
+      "learning_rate": 0.002895080639464909,
+      "loss": 0.9997,
+      "step": 1557
+    },
+    {
+      "epoch": 0.1198369356203369,
+      "learning_rate": 0.0028949474211114943,
+      "loss": 0.9378,
+      "step": 1558
+    },
+    {
+      "epoch": 0.11991385278055534,
+      "learning_rate": 0.002894814121305646,
+      "loss": 1.0979,
+      "step": 1559
+    },
+    {
+      "epoch": 0.11999076994077379,
+      "learning_rate": 0.0028946807400551475,
+      "loss": 1.1207,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12006768710099223,
+      "learning_rate": 0.0028945472773677875,
+      "loss": 1.6021,
+      "step": 1561
+    },
+    {
+      "epoch": 0.12014460426121068,
+      "learning_rate": 0.0028944137332513583,
+      "loss": 1.1666,
+      "step": 1562
+    },
+    {
+      "epoch": 0.12022152142142913,
+      "learning_rate": 0.0028942801077136574,
+      "loss": 1.1044,
+      "step": 1563
+    },
+    {
+      "epoch": 0.12029843858164757,
+      "learning_rate": 0.0028941464007624885,
+      "loss": 1.0011,
+      "step": 1564
+    },
+    {
+      "epoch": 0.12037535574186602,
+      "learning_rate": 0.002894012612405659,
+      "loss": 1.0823,
+      "step": 1565
+    },
+    {
+      "epoch": 0.12045227290208446,
+      "learning_rate": 0.0028938787426509792,
+      "loss": 1.0855,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1205291900623029,
+      "learning_rate": 0.002893744791506267,
+      "loss": 1.0098,
+      "step": 1567
+    },
+    {
+      "epoch": 0.12060610722252134,
+      "learning_rate": 0.0028936107589793447,
+      "loss": 1.3746,
+      "step": 1568
+    },
+    {
+      "epoch": 0.12068302438273978,
+      "learning_rate": 0.0028934766450780373,
+      "loss": 1.0474,
+      "step": 1569
+    },
+    {
+      "epoch": 0.12075994154295823,
+      "learning_rate": 0.0028933424498101762,
+      "loss": 0.7657,
+      "step": 1570
+    },
+    {
+      "epoch": 0.12083685870317667,
+      "learning_rate": 0.0028932081731835976,
+      "loss": 1.2252,
+      "step": 1571
+    },
+    {
+      "epoch": 0.12091377586339512,
+      "learning_rate": 0.0028930738152061417,
+      "loss": 1.1658,
+      "step": 1572
+    },
+    {
+      "epoch": 0.12099069302361357,
+      "learning_rate": 0.002892939375885654,
+      "loss": 1.1699,
+      "step": 1573
+    },
+    {
+      "epoch": 0.12106761018383201,
+      "learning_rate": 0.002892804855229984,
+      "loss": 1.1706,
+      "step": 1574
+    },
+    {
+      "epoch": 0.12114452734405046,
+      "learning_rate": 0.0028926702532469876,
+      "loss": 1.1747,
+      "step": 1575
+    },
+    {
+      "epoch": 0.1212214445042689,
+      "learning_rate": 0.0028925355699445233,
+      "loss": 1.2883,
+      "step": 1576
+    },
+    {
+      "epoch": 0.12129836166448735,
+      "learning_rate": 0.002892400805330456,
+      "loss": 1.0269,
+      "step": 1577
+    },
+    {
+      "epoch": 0.12137527882470579,
+      "learning_rate": 0.0028922659594126546,
+      "loss": 1.1479,
+      "step": 1578
+    },
+    {
+      "epoch": 0.12145219598492424,
+      "learning_rate": 0.002892131032198992,
+      "loss": 1.1261,
+      "step": 1579
+    },
+    {
+      "epoch": 0.12152911314514268,
+      "learning_rate": 0.0028919960236973487,
+      "loss": 1.2145,
+      "step": 1580
+    },
+    {
+      "epoch": 0.12160603030536113,
+      "learning_rate": 0.0028918609339156067,
+      "loss": 1.2964,
+      "step": 1581
+    },
+    {
+      "epoch": 0.12168294746557957,
+      "learning_rate": 0.002891725762861654,
+      "loss": 0.8897,
+      "step": 1582
+    },
+    {
+      "epoch": 0.12175986462579802,
+      "learning_rate": 0.0028915905105433835,
+      "loss": 1.0423,
+      "step": 1583
+    },
+    {
+      "epoch": 0.12183678178601647,
+      "learning_rate": 0.0028914551769686925,
+      "loss": 1.2013,
+      "step": 1584
+    },
+    {
+      "epoch": 0.12191369894623491,
+      "learning_rate": 0.0028913197621454846,
+      "loss": 1.2252,
+      "step": 1585
+    },
+    {
+      "epoch": 0.12199061610645336,
+      "learning_rate": 0.0028911842660816647,
+      "loss": 1.0297,
+      "step": 1586
+    },
+    {
+      "epoch": 0.12206753326667179,
+      "learning_rate": 0.0028910486887851463,
+      "loss": 1.1314,
+      "step": 1587
+    },
+    {
+      "epoch": 0.12214445042689023,
+      "learning_rate": 0.002890913030263845,
+      "loss": 1.1527,
+      "step": 1588
+    },
+    {
+      "epoch": 0.12222136758710868,
+      "learning_rate": 0.0028907772905256823,
+      "loss": 1.4608,
+      "step": 1589
+    },
+    {
+      "epoch": 0.12229828474732712,
+      "learning_rate": 0.0028906414695785846,
+      "loss": 1.0999,
+      "step": 1590
+    },
+    {
+      "epoch": 0.12237520190754557,
+      "learning_rate": 0.0028905055674304813,
+      "loss": 1.2517,
+      "step": 1591
+    },
+    {
+      "epoch": 0.12245211906776401,
+      "learning_rate": 0.0028903695840893093,
+      "loss": 1.0303,
+      "step": 1592
+    },
+    {
+      "epoch": 0.12252903622798246,
+      "learning_rate": 0.0028902335195630084,
+      "loss": 1.153,
+      "step": 1593
+    },
+    {
+      "epoch": 0.1226059533882009,
+      "learning_rate": 0.002890097373859523,
+      "loss": 1.199,
+      "step": 1594
+    },
+    {
+      "epoch": 0.12268287054841935,
+      "learning_rate": 0.0028899611469868037,
+      "loss": 1.2185,
+      "step": 1595
+    },
+    {
+      "epoch": 0.1227597877086378,
+      "learning_rate": 0.002889824838952804,
+      "loss": 0.7877,
+      "step": 1596
+    },
+    {
+      "epoch": 0.12283670486885624,
+      "learning_rate": 0.002889688449765484,
+      "loss": 1.426,
+      "step": 1597
+    },
+    {
+      "epoch": 0.12291362202907469,
+      "learning_rate": 0.0028895519794328067,
+      "loss": 1.2392,
+      "step": 1598
+    },
+    {
+      "epoch": 0.12299053918929313,
+      "learning_rate": 0.0028894154279627414,
+      "loss": 1.1887,
+      "step": 1599
+    },
+    {
+      "epoch": 0.12306745634951158,
+      "learning_rate": 0.0028892787953632613,
+      "loss": 0.842,
+      "step": 1600
+    },
+    {
+      "epoch": 0.12314437350973002,
+      "learning_rate": 0.0028891420816423443,
+      "loss": 1.1089,
+      "step": 1601
+    },
+    {
+      "epoch": 0.12322129066994847,
+      "learning_rate": 0.0028890052868079737,
+      "loss": 1.207,
+      "step": 1602
+    },
+    {
+      "epoch": 0.12329820783016691,
+      "learning_rate": 0.0028888684108681366,
+      "loss": 1.1223,
+      "step": 1603
+    },
+    {
+      "epoch": 0.12337512499038536,
+      "learning_rate": 0.0028887314538308256,
+      "loss": 1.1937,
+      "step": 1604
+    },
+    {
+      "epoch": 0.1234520421506038,
+      "learning_rate": 0.0028885944157040376,
+      "loss": 1.0141,
+      "step": 1605
+    },
+    {
+      "epoch": 0.12352895931082225,
+      "learning_rate": 0.002888457296495774,
+      "loss": 1.2021,
+      "step": 1606
+    },
+    {
+      "epoch": 0.1236058764710407,
+      "learning_rate": 0.0028883200962140426,
+      "loss": 1.2874,
+      "step": 1607
+    },
+    {
+      "epoch": 0.12368279363125913,
+      "learning_rate": 0.0028881828148668537,
+      "loss": 1.3708,
+      "step": 1608
+    },
+    {
+      "epoch": 0.12375971079147757,
+      "learning_rate": 0.0028880454524622236,
+      "loss": 1.1228,
+      "step": 1609
+    },
+    {
+      "epoch": 0.12383662795169602,
+      "learning_rate": 0.002887908009008173,
+      "loss": 0.7563,
+      "step": 1610
+    },
+    {
+      "epoch": 0.12391354511191446,
+      "learning_rate": 0.0028877704845127273,
+      "loss": 1.0775,
+      "step": 1611
+    },
+    {
+      "epoch": 0.12399046227213291,
+      "learning_rate": 0.0028876328789839167,
+      "loss": 0.855,
+      "step": 1612
+    },
+    {
+      "epoch": 0.12406737943235135,
+      "learning_rate": 0.002887495192429776,
+      "loss": 1.0989,
+      "step": 1613
+    },
+    {
+      "epoch": 0.1241442965925698,
+      "learning_rate": 0.002887357424858345,
+      "loss": 1.3884,
+      "step": 1614
+    },
+    {
+      "epoch": 0.12422121375278825,
+      "learning_rate": 0.0028872195762776683,
+      "loss": 0.7945,
+      "step": 1615
+    },
+    {
+      "epoch": 0.12429813091300669,
+      "learning_rate": 0.0028870816466957942,
+      "loss": 1.2136,
+      "step": 1616
+    },
+    {
+      "epoch": 0.12437504807322514,
+      "learning_rate": 0.0028869436361207775,
+      "loss": 1.1295,
+      "step": 1617
+    },
+    {
+      "epoch": 0.12445196523344358,
+      "learning_rate": 0.0028868055445606766,
+      "loss": 1.2467,
+      "step": 1618
+    },
+    {
+      "epoch": 0.12452888239366203,
+      "learning_rate": 0.0028866673720235546,
+      "loss": 1.2908,
+      "step": 1619
+    },
+    {
+      "epoch": 0.12460579955388047,
+      "learning_rate": 0.0028865291185174798,
+      "loss": 1.2547,
+      "step": 1620
+    },
+    {
+      "epoch": 0.12468271671409892,
+      "learning_rate": 0.0028863907840505246,
+      "loss": 1.543,
+      "step": 1621
+    },
+    {
+      "epoch": 0.12475963387431736,
+      "learning_rate": 0.0028862523686307663,
+      "loss": 1.0831,
+      "step": 1622
+    },
+    {
+      "epoch": 0.12483655103453581,
+      "learning_rate": 0.002886113872266288,
+      "loss": 1.2436,
+      "step": 1623
+    },
+    {
+      "epoch": 0.12491346819475425,
+      "learning_rate": 0.0028859752949651762,
+      "loss": 1.5413,
+      "step": 1624
+    },
+    {
+      "epoch": 0.1249903853549727,
+      "learning_rate": 0.002885836636735522,
+      "loss": 1.1666,
+      "step": 1625
+    },
+    {
+      "epoch": 0.12506730251519113,
+      "learning_rate": 0.0028856978975854222,
+      "loss": 1.1694,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1251442196754096,
+      "learning_rate": 0.0028855590775229783,
+      "loss": 0.9153,
+      "step": 1627
+    },
+    {
+      "epoch": 0.12522113683562802,
+      "learning_rate": 0.002885420176556296,
+      "loss": 1.4449,
+      "step": 1628
+    },
+    {
+      "epoch": 0.12529805399584648,
+      "learning_rate": 0.0028852811946934857,
+      "loss": 1.1671,
+      "step": 1629
+    },
+    {
+      "epoch": 0.1253749711560649,
+      "learning_rate": 0.002885142131942663,
+      "loss": 1.253,
+      "step": 1630
+    },
+    {
+      "epoch": 0.12545188831628337,
+      "learning_rate": 0.0028850029883119473,
+      "loss": 1.1113,
+      "step": 1631
+    },
+    {
+      "epoch": 0.1255288054765018,
+      "learning_rate": 0.0028848637638094637,
+      "loss": 0.9715,
+      "step": 1632
+    },
+    {
+      "epoch": 0.12560572263672026,
+      "learning_rate": 0.0028847244584433414,
+      "loss": 0.9045,
+      "step": 1633
+    },
+    {
+      "epoch": 0.1256826397969387,
+      "learning_rate": 0.0028845850722217154,
+      "loss": 1.2711,
+      "step": 1634
+    },
+    {
+      "epoch": 0.12575955695715715,
+      "learning_rate": 0.0028844456051527237,
+      "loss": 0.9732,
+      "step": 1635
+    },
+    {
+      "epoch": 0.12583647411737559,
+      "learning_rate": 0.0028843060572445105,
+      "loss": 1.3341,
+      "step": 1636
+    },
+    {
+      "epoch": 0.12591339127759404,
+      "learning_rate": 0.0028841664285052236,
+      "loss": 0.8971,
+      "step": 1637
+    },
+    {
+      "epoch": 0.12599030843781248,
+      "learning_rate": 0.0028840267189430167,
+      "loss": 1.0697,
+      "step": 1638
+    },
+    {
+      "epoch": 0.1260672255980309,
+      "learning_rate": 0.002883886928566047,
+      "loss": 0.9462,
+      "step": 1639
+    },
+    {
+      "epoch": 0.12614414275824937,
+      "learning_rate": 0.0028837470573824775,
+      "loss": 0.4884,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1262210599184678,
+      "learning_rate": 0.0028836071054004755,
+      "loss": 1.8954,
+      "step": 1641
+    },
+    {
+      "epoch": 0.12629797707868626,
+      "learning_rate": 0.0028834670726282123,
+      "loss": 0.9863,
+      "step": 1642
+    },
+    {
+      "epoch": 0.1263748942389047,
+      "learning_rate": 0.002883326959073865,
+      "loss": 1.015,
+      "step": 1643
+    },
+    {
+      "epoch": 0.12645181139912315,
+      "learning_rate": 0.0028831867647456143,
+      "loss": 1.1744,
+      "step": 1644
+    },
+    {
+      "epoch": 0.12652872855934158,
+      "learning_rate": 0.002883046489651648,
+      "loss": 0.8641,
+      "step": 1645
+    },
+    {
+      "epoch": 0.12660564571956004,
+      "learning_rate": 0.0028829061338001547,
+      "loss": 1.0122,
+      "step": 1646
+    },
+    {
+      "epoch": 0.12668256287977847,
+      "learning_rate": 0.0028827656971993315,
+      "loss": 1.2379,
+      "step": 1647
+    },
+    {
+      "epoch": 0.12675948003999693,
+      "learning_rate": 0.0028826251798573783,
+      "loss": 0.9072,
+      "step": 1648
+    },
+    {
+      "epoch": 0.12683639720021536,
+      "learning_rate": 0.0028824845817825005,
+      "loss": 0.8084,
+      "step": 1649
+    },
+    {
+      "epoch": 0.12691331436043382,
+      "learning_rate": 0.002882343902982906,
+      "loss": 1.1727,
+      "step": 1650
+    },
+    {
+      "epoch": 0.12699023152065225,
+      "learning_rate": 0.0028822031434668108,
+      "loss": 1.1908,
+      "step": 1651
+    },
+    {
+      "epoch": 0.1270671486808707,
+      "learning_rate": 0.0028820623032424337,
+      "loss": 1.2055,
+      "step": 1652
+    },
+    {
+      "epoch": 0.12714406584108914,
+      "learning_rate": 0.0028819213823179984,
+      "loss": 1.3214,
+      "step": 1653
+    },
+    {
+      "epoch": 0.1272209830013076,
+      "learning_rate": 0.002881780380701733,
+      "loss": 1.1241,
+      "step": 1654
+    },
+    {
+      "epoch": 0.12729790016152603,
+      "learning_rate": 0.0028816392984018714,
+      "loss": 0.9759,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1273748173217445,
+      "learning_rate": 0.0028814981354266514,
+      "loss": 1.2301,
+      "step": 1656
+    },
+    {
+      "epoch": 0.12745173448196292,
+      "learning_rate": 0.0028813568917843155,
+      "loss": 1.2305,
+      "step": 1657
+    },
+    {
+      "epoch": 0.12752865164218138,
+      "learning_rate": 0.0028812155674831103,
+      "loss": 0.9721,
+      "step": 1658
+    },
+    {
+      "epoch": 0.12760556880239982,
+      "learning_rate": 0.0028810741625312896,
+      "loss": 1.0428,
+      "step": 1659
+    },
+    {
+      "epoch": 0.12768248596261825,
+      "learning_rate": 0.002880932676937109,
+      "loss": 1.1266,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1277594031228367,
+      "learning_rate": 0.00288079111070883,
+      "loss": 1.5436,
+      "step": 1661
+    },
+    {
+      "epoch": 0.12783632028305514,
+      "learning_rate": 0.002880649463854719,
+      "loss": 1.0405,
+      "step": 1662
+    },
+    {
+      "epoch": 0.1279132374432736,
+      "learning_rate": 0.002880507736383047,
+      "loss": 1.3491,
+      "step": 1663
+    },
+    {
+      "epoch": 0.12799015460349203,
+      "learning_rate": 0.0028803659283020895,
+      "loss": 1.2567,
+      "step": 1664
+    },
+    {
+      "epoch": 0.1280670717637105,
+      "learning_rate": 0.0028802240396201273,
+      "loss": 1.1244,
+      "step": 1665
+    },
+    {
+      "epoch": 0.12814398892392892,
+      "learning_rate": 0.0028800820703454447,
+      "loss": 1.061,
+      "step": 1666
+    },
+    {
+      "epoch": 0.12822090608414738,
+      "learning_rate": 0.0028799400204863315,
+      "loss": 1.0545,
+      "step": 1667
+    },
+    {
+      "epoch": 0.1282978232443658,
+      "learning_rate": 0.0028797978900510822,
+      "loss": 0.9163,
+      "step": 1668
+    },
+    {
+      "epoch": 0.12837474040458427,
+      "learning_rate": 0.0028796556790479967,
+      "loss": 1.3919,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1284516575648027,
+      "learning_rate": 0.0028795133874853778,
+      "loss": 1.1179,
+      "step": 1670
+    },
+    {
+      "epoch": 0.12852857472502116,
+      "learning_rate": 0.0028793710153715347,
+      "loss": 1.1639,
+      "step": 1671
+    },
+    {
+      "epoch": 0.1286054918852396,
+      "learning_rate": 0.0028792285627147808,
+      "loss": 1.3275,
+      "step": 1672
+    },
+    {
+      "epoch": 0.12868240904545805,
+      "learning_rate": 0.002879086029523433,
+      "loss": 1.1799,
+      "step": 1673
+    },
+    {
+      "epoch": 0.12875932620567648,
+      "learning_rate": 0.0028789434158058153,
+      "loss": 1.23,
+      "step": 1674
+    },
+    {
+      "epoch": 0.12883624336589494,
+      "learning_rate": 0.0028788007215702544,
+      "loss": 1.0948,
+      "step": 1675
+    },
+    {
+      "epoch": 0.12891316052611337,
+      "learning_rate": 0.002878657946825082,
+      "loss": 1.1994,
+      "step": 1676
+    },
+    {
+      "epoch": 0.12899007768633183,
+      "learning_rate": 0.0028785150915786356,
+      "loss": 1.4292,
+      "step": 1677
+    },
+    {
+      "epoch": 0.12906699484655026,
+      "learning_rate": 0.0028783721558392566,
+      "loss": 1.3145,
+      "step": 1678
+    },
+    {
+      "epoch": 0.1291439120067687,
+      "learning_rate": 0.00287822913961529,
+      "loss": 1.2676,
+      "step": 1679
+    },
+    {
+      "epoch": 0.12922082916698716,
+      "learning_rate": 0.0028780860429150883,
+      "loss": 1.4908,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1292977463272056,
+      "learning_rate": 0.0028779428657470063,
+      "loss": 1.2185,
+      "step": 1681
+    },
+    {
+      "epoch": 0.12937466348742405,
+      "learning_rate": 0.002877799608119404,
+      "loss": 1.4401,
+      "step": 1682
+    },
+    {
+      "epoch": 0.12945158064764248,
+      "learning_rate": 0.002877656270040647,
+      "loss": 1.0149,
+      "step": 1683
+    },
+    {
+      "epoch": 0.12952849780786094,
+      "learning_rate": 0.0028775128515191053,
+      "loss": 0.9654,
+      "step": 1684
+    },
+    {
+      "epoch": 0.12960541496807937,
+      "learning_rate": 0.0028773693525631513,
+      "loss": 1.1157,
+      "step": 1685
+    },
+    {
+      "epoch": 0.12968233212829783,
+      "learning_rate": 0.002877225773181166,
+      "loss": 0.9185,
+      "step": 1686
+    },
+    {
+      "epoch": 0.12975924928851626,
+      "learning_rate": 0.0028770821133815325,
+      "loss": 1.1889,
+      "step": 1687
+    },
+    {
+      "epoch": 0.12983616644873472,
+      "learning_rate": 0.002876938373172639,
+      "loss": 1.491,
+      "step": 1688
+    },
+    {
+      "epoch": 0.12991308360895315,
+      "learning_rate": 0.002876794552562879,
+      "loss": 1.3061,
+      "step": 1689
+    },
+    {
+      "epoch": 0.1299900007691716,
+      "learning_rate": 0.00287665065156065,
+      "loss": 1.2791,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13006691792939004,
+      "learning_rate": 0.0028765066701743553,
+      "loss": 1.3298,
+      "step": 1691
+    },
+    {
+      "epoch": 0.1301438350896085,
+      "learning_rate": 0.002876362608412401,
+      "loss": 1.3656,
+      "step": 1692
+    },
+    {
+      "epoch": 0.13022075224982693,
+      "learning_rate": 0.0028762184662832,
+      "loss": 1.0401,
+      "step": 1693
+    },
+    {
+      "epoch": 0.1302976694100454,
+      "learning_rate": 0.0028760742437951682,
+      "loss": 1.1322,
+      "step": 1694
+    },
+    {
+      "epoch": 0.13037458657026382,
+      "learning_rate": 0.002875929940956727,
+      "loss": 1.0038,
+      "step": 1695
+    },
+    {
+      "epoch": 0.13045150373048228,
+      "learning_rate": 0.002875785557776303,
+      "loss": 0.963,
+      "step": 1696
+    },
+    {
+      "epoch": 0.1305284208907007,
+      "learning_rate": 0.0028756410942623266,
+      "loss": 0.9686,
+      "step": 1697
+    },
+    {
+      "epoch": 0.13060533805091917,
+      "learning_rate": 0.002875496550423233,
+      "loss": 1.0261,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1306822552111376,
+      "learning_rate": 0.0028753519262674616,
+      "loss": 1.0338,
+      "step": 1699
+    },
+    {
+      "epoch": 0.13075917237135604,
+      "learning_rate": 0.0028752072218034587,
+      "loss": 1.2176,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1308360895315745,
+      "learning_rate": 0.002875062437039672,
+      "loss": 1.1064,
+      "step": 1701
+    },
+    {
+      "epoch": 0.13091300669179293,
+      "learning_rate": 0.0028749175719845576,
+      "loss": 1.1406,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1309899238520114,
+      "learning_rate": 0.002874772626646573,
+      "loss": 1.3195,
+      "step": 1703
+    },
+    {
+      "epoch": 0.13106684101222982,
+      "learning_rate": 0.0028746276010341815,
+      "loss": 0.963,
+      "step": 1704
+    },
+    {
+      "epoch": 0.13114375817244828,
+      "learning_rate": 0.0028744824951558523,
+      "loss": 0.9756,
+      "step": 1705
+    },
+    {
+      "epoch": 0.1312206753326667,
+      "learning_rate": 0.002874337309020057,
+      "loss": 0.9419,
+      "step": 1706
+    },
+    {
+      "epoch": 0.13129759249288517,
+      "learning_rate": 0.002874192042635275,
+      "loss": 0.9682,
+      "step": 1707
+    },
+    {
+      "epoch": 0.1313745096531036,
+      "learning_rate": 0.0028740466960099865,
+      "loss": 1.1678,
+      "step": 1708
+    },
+    {
+      "epoch": 0.13145142681332206,
+      "learning_rate": 0.00287390126915268,
+      "loss": 1.7509,
+      "step": 1709
+    },
+    {
+      "epoch": 0.1315283439735405,
+      "learning_rate": 0.0028737557620718466,
+      "loss": 0.9172,
+      "step": 1710
+    },
+    {
+      "epoch": 0.13160526113375895,
+      "learning_rate": 0.0028736101747759823,
+      "loss": 1.3299,
+      "step": 1711
+    },
+    {
+      "epoch": 0.13168217829397738,
+      "learning_rate": 0.0028734645072735886,
+      "loss": 1.1382,
+      "step": 1712
+    },
+    {
+      "epoch": 0.13175909545419584,
+      "learning_rate": 0.00287331875957317,
+      "loss": 1.2588,
+      "step": 1713
+    },
+    {
+      "epoch": 0.13183601261441427,
+      "learning_rate": 0.002873172931683239,
+      "loss": 0.9852,
+      "step": 1714
+    },
+    {
+      "epoch": 0.13191292977463273,
+      "learning_rate": 0.002873027023612309,
+      "loss": 1.3062,
+      "step": 1715
+    },
+    {
+      "epoch": 0.13198984693485116,
+      "learning_rate": 0.0028728810353689,
+      "loss": 0.9516,
+      "step": 1716
+    },
+    {
+      "epoch": 0.13206676409506962,
+      "learning_rate": 0.0028727349669615365,
+      "loss": 1.0101,
+      "step": 1717
+    },
+    {
+      "epoch": 0.13214368125528805,
+      "learning_rate": 0.0028725888183987478,
+      "loss": 0.9942,
+      "step": 1718
+    },
+    {
+      "epoch": 0.1322205984155065,
+      "learning_rate": 0.0028724425896890678,
+      "loss": 1.1523,
+      "step": 1719
+    },
+    {
+      "epoch": 0.13229751557572494,
+      "learning_rate": 0.0028722962808410343,
+      "loss": 1.1402,
+      "step": 1720
+    },
+    {
+      "epoch": 0.13237443273594338,
+      "learning_rate": 0.002872149891863191,
+      "loss": 1.0075,
+      "step": 1721
+    },
+    {
+      "epoch": 0.13245134989616184,
+      "learning_rate": 0.002872003422764085,
+      "loss": 1.3004,
+      "step": 1722
+    },
+    {
+      "epoch": 0.13252826705638027,
+      "learning_rate": 0.00287185687355227,
+      "loss": 1.0971,
+      "step": 1723
+    },
+    {
+      "epoch": 0.13260518421659873,
+      "learning_rate": 0.002871710244236302,
+      "loss": 1.4052,
+      "step": 1724
+    },
+    {
+      "epoch": 0.13268210137681716,
+      "learning_rate": 0.0028715635348247437,
+      "loss": 1.1598,
+      "step": 1725
+    },
+    {
+      "epoch": 0.13275901853703562,
+      "learning_rate": 0.0028714167453261605,
+      "loss": 1.0552,
+      "step": 1726
+    },
+    {
+      "epoch": 0.13283593569725405,
+      "learning_rate": 0.0028712698757491248,
+      "loss": 1.0522,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1329128528574725,
+      "learning_rate": 0.002871122926102212,
+      "loss": 1.2255,
+      "step": 1728
+    },
+    {
+      "epoch": 0.13298977001769094,
+      "learning_rate": 0.0028709758963940024,
+      "loss": 0.9807,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1330666871779094,
+      "learning_rate": 0.0028708287866330813,
+      "loss": 1.1592,
+      "step": 1730
+    },
+    {
+      "epoch": 0.13314360433812783,
+      "learning_rate": 0.0028706815968280387,
+      "loss": 1.0734,
+      "step": 1731
+    },
+    {
+      "epoch": 0.1332205214983463,
+      "learning_rate": 0.0028705343269874693,
+      "loss": 1.2188,
+      "step": 1732
+    },
+    {
+      "epoch": 0.13329743865856472,
+      "learning_rate": 0.002870386977119972,
+      "loss": 1.4174,
+      "step": 1733
+    },
+    {
+      "epoch": 0.13337435581878318,
+      "learning_rate": 0.0028702395472341513,
+      "loss": 1.3331,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1334512729790016,
+      "learning_rate": 0.0028700920373386156,
+      "loss": 1.1743,
+      "step": 1735
+    },
+    {
+      "epoch": 0.13352819013922007,
+      "learning_rate": 0.0028699444474419773,
+      "loss": 1.0421,
+      "step": 1736
+    },
+    {
+      "epoch": 0.1336051072994385,
+      "learning_rate": 0.0028697967775528554,
+      "loss": 1.0958,
+      "step": 1737
+    },
+    {
+      "epoch": 0.13368202445965696,
+      "learning_rate": 0.0028696490276798716,
+      "loss": 1.1462,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1337589416198754,
+      "learning_rate": 0.0028695011978316543,
+      "loss": 1.2942,
+      "step": 1739
+    },
+    {
+      "epoch": 0.13383585878009385,
+      "learning_rate": 0.0028693532880168346,
+      "loss": 1.1019,
+      "step": 1740
+    },
+    {
+      "epoch": 0.13391277594031228,
+      "learning_rate": 0.0028692052982440494,
+      "loss": 1.1589,
+      "step": 1741
+    },
+    {
+      "epoch": 0.13398969310053072,
+      "learning_rate": 0.0028690572285219396,
+      "loss": 1.3864,
+      "step": 1742
+    },
+    {
+      "epoch": 0.13406661026074917,
+      "learning_rate": 0.0028689090788591516,
+      "loss": 1.0047,
+      "step": 1743
+    },
+    {
+      "epoch": 0.1341435274209676,
+      "learning_rate": 0.0028687608492643357,
+      "loss": 1.1152,
+      "step": 1744
+    },
+    {
+      "epoch": 0.13422044458118607,
+      "learning_rate": 0.0028686125397461475,
+      "loss": 1.1046,
+      "step": 1745
+    },
+    {
+      "epoch": 0.1342973617414045,
+      "learning_rate": 0.002868464150313247,
+      "loss": 1.3783,
+      "step": 1746
+    },
+    {
+      "epoch": 0.13437427890162296,
+      "learning_rate": 0.0028683156809742975,
+      "loss": 1.0825,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1344511960618414,
+      "learning_rate": 0.0028681671317379704,
+      "loss": 1.1973,
+      "step": 1748
+    },
+    {
+      "epoch": 0.13452811322205985,
+      "learning_rate": 0.002868018502612938,
+      "loss": 0.8653,
+      "step": 1749
+    },
+    {
+      "epoch": 0.13460503038227828,
+      "learning_rate": 0.0028678697936078803,
+      "loss": 1.3479,
+      "step": 1750
+    },
+    {
+      "epoch": 0.13468194754249674,
+      "learning_rate": 0.0028677210047314793,
+      "loss": 0.9689,
+      "step": 1751
+    },
+    {
+      "epoch": 0.13475886470271517,
+      "learning_rate": 0.0028675721359924236,
+      "loss": 1.2804,
+      "step": 1752
+    },
+    {
+      "epoch": 0.13483578186293363,
+      "learning_rate": 0.002867423187399405,
+      "loss": 1.017,
+      "step": 1753
+    },
+    {
+      "epoch": 0.13491269902315206,
+      "learning_rate": 0.0028672741589611224,
+      "loss": 0.9328,
+      "step": 1754
+    },
+    {
+      "epoch": 0.13498961618337052,
+      "learning_rate": 0.0028671250506862762,
+      "loss": 1.1762,
+      "step": 1755
+    },
+    {
+      "epoch": 0.13506653334358895,
+      "learning_rate": 0.002866975862583574,
+      "loss": 1.1972,
+      "step": 1756
+    },
+    {
+      "epoch": 0.1351434505038074,
+      "learning_rate": 0.0028668265946617255,
+      "loss": 1.0895,
+      "step": 1757
+    },
+    {
+      "epoch": 0.13522036766402584,
+      "learning_rate": 0.0028666772469294484,
+      "loss": 1.2591,
+      "step": 1758
+    },
+    {
+      "epoch": 0.1352972848242443,
+      "learning_rate": 0.0028665278193954623,
+      "loss": 1.5245,
+      "step": 1759
+    },
+    {
+      "epoch": 0.13537420198446273,
+      "learning_rate": 0.0028663783120684927,
+      "loss": 1.2559,
+      "step": 1760
+    },
+    {
+      "epoch": 0.13545111914468116,
+      "learning_rate": 0.0028662287249572697,
+      "loss": 1.2404,
+      "step": 1761
+    },
+    {
+      "epoch": 0.13552803630489962,
+      "learning_rate": 0.0028660790580705277,
+      "loss": 0.9549,
+      "step": 1762
+    },
+    {
+      "epoch": 0.13560495346511806,
+      "learning_rate": 0.002865929311417005,
+      "loss": 0.9313,
+      "step": 1763
+    },
+    {
+      "epoch": 0.13568187062533651,
+      "learning_rate": 0.002865779485005447,
+      "loss": 1.2338,
+      "step": 1764
+    },
+    {
+      "epoch": 0.13575878778555495,
+      "learning_rate": 0.0028656295788446013,
+      "loss": 1.2663,
+      "step": 1765
+    },
+    {
+      "epoch": 0.1358357049457734,
+      "learning_rate": 0.002865479592943221,
+      "loss": 1.0019,
+      "step": 1766
+    },
+    {
+      "epoch": 0.13591262210599184,
+      "learning_rate": 0.0028653295273100648,
+      "loss": 1.5426,
+      "step": 1767
+    },
+    {
+      "epoch": 0.1359895392662103,
+      "learning_rate": 0.0028651793819538937,
+      "loss": 1.3661,
+      "step": 1768
+    },
+    {
+      "epoch": 0.13606645642642873,
+      "learning_rate": 0.0028650291568834765,
+      "loss": 1.0231,
+      "step": 1769
+    },
+    {
+      "epoch": 0.1361433735866472,
+      "learning_rate": 0.002864878852107584,
+      "loss": 1.4282,
+      "step": 1770
+    },
+    {
+      "epoch": 0.13622029074686562,
+      "learning_rate": 0.0028647284676349928,
+      "loss": 1.0069,
+      "step": 1771
+    },
+    {
+      "epoch": 0.13629720790708408,
+      "learning_rate": 0.002864578003474484,
+      "loss": 1.207,
+      "step": 1772
+    },
+    {
+      "epoch": 0.1363741250673025,
+      "learning_rate": 0.0028644274596348438,
+      "loss": 1.0187,
+      "step": 1773
+    },
+    {
+      "epoch": 0.13645104222752097,
+      "learning_rate": 0.0028642768361248623,
+      "loss": 1.1986,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1365279593877394,
+      "learning_rate": 0.0028641261329533344,
+      "loss": 1.0283,
+      "step": 1775
+    },
+    {
+      "epoch": 0.13660487654795786,
+      "learning_rate": 0.00286397535012906,
+      "loss": 1.2079,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1366817937081763,
+      "learning_rate": 0.0028638244876608428,
+      "loss": 0.8502,
+      "step": 1777
+    },
+    {
+      "epoch": 0.13675871086839475,
+      "learning_rate": 0.002863673545557493,
+      "loss": 0.8472,
+      "step": 1778
+    },
+    {
+      "epoch": 0.13683562802861318,
+      "learning_rate": 0.0028635225238278234,
+      "loss": 1.2772,
+      "step": 1779
+    },
+    {
+      "epoch": 0.13691254518883164,
+      "learning_rate": 0.002863371422480653,
+      "loss": 1.1688,
+      "step": 1780
+    },
+    {
+      "epoch": 0.13698946234905007,
+      "learning_rate": 0.002863220241524804,
+      "loss": 0.7301,
+      "step": 1781
+    },
+    {
+      "epoch": 0.1370663795092685,
+      "learning_rate": 0.0028630689809691044,
+      "loss": 1.0963,
+      "step": 1782
+    },
+    {
+      "epoch": 0.13714329666948696,
+      "learning_rate": 0.002862917640822387,
+      "loss": 0.9698,
+      "step": 1783
+    },
+    {
+      "epoch": 0.1372202138297054,
+      "learning_rate": 0.0028627662210934874,
+      "loss": 1.1856,
+      "step": 1784
+    },
+    {
+      "epoch": 0.13729713098992385,
+      "learning_rate": 0.0028626147217912477,
+      "loss": 0.9043,
+      "step": 1785
+    },
+    {
+      "epoch": 0.13737404815014229,
+      "learning_rate": 0.002862463142924515,
+      "loss": 1.1654,
+      "step": 1786
+    },
+    {
+      "epoch": 0.13745096531036075,
+      "learning_rate": 0.0028623114845021396,
+      "loss": 1.2161,
+      "step": 1787
+    },
+    {
+      "epoch": 0.13752788247057918,
+      "learning_rate": 0.0028621597465329765,
+      "loss": 1.1165,
+      "step": 1788
+    },
+    {
+      "epoch": 0.13760479963079764,
+      "learning_rate": 0.0028620079290258863,
+      "loss": 1.0131,
+      "step": 1789
+    },
+    {
+      "epoch": 0.13768171679101607,
+      "learning_rate": 0.002861856031989733,
+      "loss": 0.8552,
+      "step": 1790
+    },
+    {
+      "epoch": 0.13775863395123453,
+      "learning_rate": 0.0028617040554333876,
+      "loss": 1.0169,
+      "step": 1791
+    },
+    {
+      "epoch": 0.13783555111145296,
+      "learning_rate": 0.0028615519993657234,
+      "loss": 1.3486,
+      "step": 1792
+    },
+    {
+      "epoch": 0.13791246827167142,
+      "learning_rate": 0.002861399863795618,
+      "loss": 0.9617,
+      "step": 1793
+    },
+    {
+      "epoch": 0.13798938543188985,
+      "learning_rate": 0.002861247648731957,
+      "loss": 0.9184,
+      "step": 1794
+    },
+    {
+      "epoch": 0.1380663025921083,
+      "learning_rate": 0.0028610953541836257,
+      "loss": 1.1691,
+      "step": 1795
+    },
+    {
+      "epoch": 0.13814321975232674,
+      "learning_rate": 0.002860942980159519,
+      "loss": 1.2551,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1382201369125452,
+      "learning_rate": 0.0028607905266685336,
+      "loss": 1.2939,
+      "step": 1797
+    },
+    {
+      "epoch": 0.13829705407276363,
+      "learning_rate": 0.0028606379937195707,
+      "loss": 1.2948,
+      "step": 1798
+    },
+    {
+      "epoch": 0.1383739712329821,
+      "learning_rate": 0.0028604853813215376,
+      "loss": 0.8462,
+      "step": 1799
+    },
+    {
+      "epoch": 0.13845088839320052,
+      "learning_rate": 0.002860332689483345,
+      "loss": 1.3494,
+      "step": 1800
+    },
+    {
+      "epoch": 0.13852780555341898,
+      "learning_rate": 0.0028601799182139086,
+      "loss": 0.9626,
+      "step": 1801
+    },
+    {
+      "epoch": 0.1386047227136374,
+      "learning_rate": 0.00286002706752215,
+      "loss": 1.0662,
+      "step": 1802
+    },
+    {
+      "epoch": 0.13868163987385584,
+      "learning_rate": 0.002859874137416993,
+      "loss": 1.2386,
+      "step": 1803
+    },
+    {
+      "epoch": 0.1387585570340743,
+      "learning_rate": 0.002859721127907368,
+      "loss": 1.5507,
+      "step": 1804
+    },
+    {
+      "epoch": 0.13883547419429274,
+      "learning_rate": 0.0028595680390022093,
+      "loss": 1.2978,
+      "step": 1805
+    },
+    {
+      "epoch": 0.1389123913545112,
+      "learning_rate": 0.002859414870710456,
+      "loss": 1.1999,
+      "step": 1806
+    },
+    {
+      "epoch": 0.13898930851472963,
+      "learning_rate": 0.0028592616230410516,
+      "loss": 1.1199,
+      "step": 1807
+    },
+    {
+      "epoch": 0.13906622567494809,
+      "learning_rate": 0.0028591082960029443,
+      "loss": 0.9053,
+      "step": 1808
+    },
+    {
+      "epoch": 0.13914314283516652,
+      "learning_rate": 0.0028589548896050874,
+      "loss": 1.2272,
+      "step": 1809
+    },
+    {
+      "epoch": 0.13922005999538498,
+      "learning_rate": 0.002858801403856438,
+      "loss": 1.04,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1392969771556034,
+      "learning_rate": 0.0028586478387659588,
+      "loss": 1.0117,
+      "step": 1811
+    },
+    {
+      "epoch": 0.13937389431582187,
+      "learning_rate": 0.0028584941943426163,
+      "loss": 1.0784,
+      "step": 1812
+    },
+    {
+      "epoch": 0.1394508114760403,
+      "learning_rate": 0.002858340470595382,
+      "loss": 0.914,
+      "step": 1813
+    },
+    {
+      "epoch": 0.13952772863625876,
+      "learning_rate": 0.002858186667533232,
+      "loss": 1.0431,
+      "step": 1814
+    },
+    {
+      "epoch": 0.1396046457964772,
+      "learning_rate": 0.0028580327851651464,
+      "loss": 1.0706,
+      "step": 1815
+    },
+    {
+      "epoch": 0.13968156295669565,
+      "learning_rate": 0.0028578788235001117,
+      "loss": 1.2811,
+      "step": 1816
+    },
+    {
+      "epoch": 0.13975848011691408,
+      "learning_rate": 0.0028577247825471177,
+      "loss": 1.2036,
+      "step": 1817
+    },
+    {
+      "epoch": 0.13983539727713254,
+      "learning_rate": 0.002857570662315158,
+      "loss": 1.3338,
+      "step": 1818
+    },
+    {
+      "epoch": 0.13991231443735097,
+      "learning_rate": 0.002857416462813233,
+      "loss": 1.1993,
+      "step": 1819
+    },
+    {
+      "epoch": 0.13998923159756943,
+      "learning_rate": 0.0028572621840503454,
+      "loss": 1.1732,
+      "step": 1820
+    },
+    {
+      "epoch": 0.14006614875778786,
+      "learning_rate": 0.002857107826035505,
+      "loss": 1.1578,
+      "step": 1821
+    },
+    {
+      "epoch": 0.14014306591800632,
+      "learning_rate": 0.0028569533887777237,
+      "loss": 1.3077,
+      "step": 1822
+    },
+    {
+      "epoch": 0.14021998307822475,
+      "learning_rate": 0.0028567988722860206,
+      "loss": 0.9864,
+      "step": 1823
+    },
+    {
+      "epoch": 0.14029690023844318,
+      "learning_rate": 0.002856644276569417,
+      "loss": 1.3024,
+      "step": 1824
+    },
+    {
+      "epoch": 0.14037381739866164,
+      "learning_rate": 0.00285648960163694,
+      "loss": 1.3467,
+      "step": 1825
+    },
+    {
+      "epoch": 0.14045073455888007,
+      "learning_rate": 0.002856334847497622,
+      "loss": 1.1539,
+      "step": 1826
+    },
+    {
+      "epoch": 0.14052765171909853,
+      "learning_rate": 0.002856180014160498,
+      "loss": 1.0856,
+      "step": 1827
+    },
+    {
+      "epoch": 0.14060456887931697,
+      "learning_rate": 0.0028560251016346103,
+      "loss": 1.2987,
+      "step": 1828
+    },
+    {
+      "epoch": 0.14068148603953542,
+      "learning_rate": 0.0028558701099290033,
+      "loss": 1.3895,
+      "step": 1829
+    },
+    {
+      "epoch": 0.14075840319975386,
+      "learning_rate": 0.0028557150390527276,
+      "loss": 1.4298,
+      "step": 1830
+    },
+    {
+      "epoch": 0.14083532035997232,
+      "learning_rate": 0.002855559889014838,
+      "loss": 1.4291,
+      "step": 1831
+    },
+    {
+      "epoch": 0.14091223752019075,
+      "learning_rate": 0.0028554046598243937,
+      "loss": 1.0573,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1409891546804092,
+      "learning_rate": 0.0028552493514904585,
+      "loss": 1.2162,
+      "step": 1833
+    },
+    {
+      "epoch": 0.14106607184062764,
+      "learning_rate": 0.0028550939640221017,
+      "loss": 0.8612,
+      "step": 1834
+    },
+    {
+      "epoch": 0.1411429890008461,
+      "learning_rate": 0.002854938497428396,
+      "loss": 0.9214,
+      "step": 1835
+    },
+    {
+      "epoch": 0.14121990616106453,
+      "learning_rate": 0.002854782951718419,
+      "loss": 1.0045,
+      "step": 1836
+    },
+    {
+      "epoch": 0.141296823321283,
+      "learning_rate": 0.0028546273269012537,
+      "loss": 1.2892,
+      "step": 1837
+    },
+    {
+      "epoch": 0.14137374048150142,
+      "learning_rate": 0.002854471622985987,
+      "loss": 1.2209,
+      "step": 1838
+    },
+    {
+      "epoch": 0.14145065764171988,
+      "learning_rate": 0.0028543158399817106,
+      "loss": 1.263,
+      "step": 1839
+    },
+    {
+      "epoch": 0.1415275748019383,
+      "learning_rate": 0.0028541599778975205,
+      "loss": 1.2222,
+      "step": 1840
+    },
+    {
+      "epoch": 0.14160449196215677,
+      "learning_rate": 0.002854004036742519,
+      "loss": 1.1956,
+      "step": 1841
+    },
+    {
+      "epoch": 0.1416814091223752,
+      "learning_rate": 0.00285384801652581,
+      "loss": 1.1176,
+      "step": 1842
+    },
+    {
+      "epoch": 0.14175832628259363,
+      "learning_rate": 0.0028536919172565045,
+      "loss": 0.9716,
+      "step": 1843
+    },
+    {
+      "epoch": 0.1418352434428121,
+      "learning_rate": 0.0028535357389437173,
+      "loss": 1.0649,
+      "step": 1844
+    },
+    {
+      "epoch": 0.14191216060303052,
+      "learning_rate": 0.0028533794815965674,
+      "loss": 0.9432,
+      "step": 1845
+    },
+    {
+      "epoch": 0.14198907776324898,
+      "learning_rate": 0.0028532231452241797,
+      "loss": 0.5686,
+      "step": 1846
+    },
+    {
+      "epoch": 0.14206599492346741,
+      "learning_rate": 0.0028530667298356815,
+      "loss": 0.9443,
+      "step": 1847
+    },
+    {
+      "epoch": 0.14214291208368587,
+      "learning_rate": 0.002852910235440207,
+      "loss": 1.2505,
+      "step": 1848
+    },
+    {
+      "epoch": 0.1422198292439043,
+      "learning_rate": 0.0028527536620468943,
+      "loss": 1.3604,
+      "step": 1849
+    },
+    {
+      "epoch": 0.14229674640412276,
+      "learning_rate": 0.002852597009664885,
+      "loss": 1.1274,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1423736635643412,
+      "learning_rate": 0.002852440278303327,
+      "loss": 1.2805,
+      "step": 1851
+    },
+    {
+      "epoch": 0.14245058072455966,
+      "learning_rate": 0.0028522834679713715,
+      "loss": 1.4961,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1425274978847781,
+      "learning_rate": 0.0028521265786781752,
+      "loss": 1.5282,
+      "step": 1853
+    },
+    {
+      "epoch": 0.14260441504499655,
+      "learning_rate": 0.0028519696104328984,
+      "loss": 1.2071,
+      "step": 1854
+    },
+    {
+      "epoch": 0.14268133220521498,
+      "learning_rate": 0.0028518125632447074,
+      "loss": 0.8997,
+      "step": 1855
+    },
+    {
+      "epoch": 0.14275824936543344,
+      "learning_rate": 0.0028516554371227723,
+      "loss": 1.5829,
+      "step": 1856
+    },
+    {
+      "epoch": 0.14283516652565187,
+      "learning_rate": 0.0028514982320762667,
+      "loss": 1.0073,
+      "step": 1857
+    },
+    {
+      "epoch": 0.14291208368587033,
+      "learning_rate": 0.0028513409481143713,
+      "loss": 0.9761,
+      "step": 1858
+    },
+    {
+      "epoch": 0.14298900084608876,
+      "learning_rate": 0.0028511835852462696,
+      "loss": 1.5369,
+      "step": 1859
+    },
+    {
+      "epoch": 0.14306591800630722,
+      "learning_rate": 0.0028510261434811506,
+      "loss": 1.4548,
+      "step": 1860
+    },
+    {
+      "epoch": 0.14314283516652565,
+      "learning_rate": 0.0028508686228282064,
+      "loss": 0.9244,
+      "step": 1861
+    },
+    {
+      "epoch": 0.1432197523267441,
+      "learning_rate": 0.002850711023296635,
+      "loss": 1.0911,
+      "step": 1862
+    },
+    {
+      "epoch": 0.14329666948696254,
+      "learning_rate": 0.0028505533448956407,
+      "loss": 1.191,
+      "step": 1863
+    },
+    {
+      "epoch": 0.14337358664718097,
+      "learning_rate": 0.0028503955876344285,
+      "loss": 1.3685,
+      "step": 1864
+    },
+    {
+      "epoch": 0.14345050380739943,
+      "learning_rate": 0.00285023775152221,
+      "loss": 1.1106,
+      "step": 1865
+    },
+    {
+      "epoch": 0.14352742096761786,
+      "learning_rate": 0.002850079836568203,
+      "loss": 0.9043,
+      "step": 1866
+    },
+    {
+      "epoch": 0.14360433812783632,
+      "learning_rate": 0.0028499218427816267,
+      "loss": 1.0367,
+      "step": 1867
+    },
+    {
+      "epoch": 0.14368125528805475,
+      "learning_rate": 0.002849763770171707,
+      "loss": 0.9066,
+      "step": 1868
+    },
+    {
+      "epoch": 0.1437581724482732,
+      "learning_rate": 0.0028496056187476745,
+      "loss": 1.5269,
+      "step": 1869
+    },
+    {
+      "epoch": 0.14383508960849165,
+      "learning_rate": 0.002849447388518763,
+      "loss": 1.2532,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1439120067687101,
+      "learning_rate": 0.0028492890794942125,
+      "loss": 0.57,
+      "step": 1871
+    },
+    {
+      "epoch": 0.14398892392892854,
+      "learning_rate": 0.0028491306916832663,
+      "loss": 1.0754,
+      "step": 1872
+    },
+    {
+      "epoch": 0.144065841089147,
+      "learning_rate": 0.002848972225095173,
+      "loss": 1.199,
+      "step": 1873
+    },
+    {
+      "epoch": 0.14414275824936543,
+      "learning_rate": 0.0028488136797391855,
+      "loss": 1.037,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1442196754095839,
+      "learning_rate": 0.0028486550556245617,
+      "loss": 1.0015,
+      "step": 1875
+    },
+    {
+      "epoch": 0.14429659256980232,
+      "learning_rate": 0.0028484963527605637,
+      "loss": 1.2186,
+      "step": 1876
+    },
+    {
+      "epoch": 0.14437350973002078,
+      "learning_rate": 0.002848337571156458,
+      "loss": 1.2143,
+      "step": 1877
+    },
+    {
+      "epoch": 0.1444504268902392,
+      "learning_rate": 0.0028481787108215165,
+      "loss": 1.2161,
+      "step": 1878
+    },
+    {
+      "epoch": 0.14452734405045767,
+      "learning_rate": 0.0028480197717650154,
+      "loss": 0.712,
+      "step": 1879
+    },
+    {
+      "epoch": 0.1446042612106761,
+      "learning_rate": 0.002847860753996234,
+      "loss": 1.2457,
+      "step": 1880
+    },
+    {
+      "epoch": 0.14468117837089456,
+      "learning_rate": 0.002847701657524459,
+      "loss": 1.331,
+      "step": 1881
+    },
+    {
+      "epoch": 0.144758095531113,
+      "learning_rate": 0.0028475424823589802,
+      "loss": 1.4052,
+      "step": 1882
+    },
+    {
+      "epoch": 0.14483501269133145,
+      "learning_rate": 0.0028473832285090908,
+      "loss": 1.0963,
+      "step": 1883
+    },
+    {
+      "epoch": 0.14491192985154988,
+      "learning_rate": 0.002847223895984091,
+      "loss": 1.1275,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1449888470117683,
+      "learning_rate": 0.0028470644847932835,
+      "loss": 1.5927,
+      "step": 1885
+    },
+    {
+      "epoch": 0.14506576417198677,
+      "learning_rate": 0.002846904994945977,
+      "loss": 1.0431,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1451426813322052,
+      "learning_rate": 0.002846745426451484,
+      "loss": 1.37,
+      "step": 1887
+    },
+    {
+      "epoch": 0.14521959849242366,
+      "learning_rate": 0.0028465857793191223,
+      "loss": 1.3338,
+      "step": 1888
+    },
+    {
+      "epoch": 0.1452965156526421,
+      "learning_rate": 0.002846426053558213,
+      "loss": 0.9425,
+      "step": 1889
+    },
+    {
+      "epoch": 0.14537343281286055,
+      "learning_rate": 0.0028462662491780836,
+      "loss": 1.2258,
+      "step": 1890
+    },
+    {
+      "epoch": 0.14545034997307899,
+      "learning_rate": 0.0028461063661880653,
+      "loss": 0.8685,
+      "step": 1891
+    },
+    {
+      "epoch": 0.14552726713329744,
+      "learning_rate": 0.0028459464045974925,
+      "loss": 1.0388,
+      "step": 1892
+    },
+    {
+      "epoch": 0.14560418429351588,
+      "learning_rate": 0.0028457863644157067,
+      "loss": 1.2992,
+      "step": 1893
+    },
+    {
+      "epoch": 0.14568110145373434,
+      "learning_rate": 0.002845626245652053,
+      "loss": 1.0715,
+      "step": 1894
+    },
+    {
+      "epoch": 0.14575801861395277,
+      "learning_rate": 0.00284546604831588,
+      "loss": 1.5183,
+      "step": 1895
+    },
+    {
+      "epoch": 0.14583493577417123,
+      "learning_rate": 0.0028453057724165424,
+      "loss": 1.0368,
+      "step": 1896
+    },
+    {
+      "epoch": 0.14591185293438966,
+      "learning_rate": 0.0028451454179633986,
+      "loss": 1.0643,
+      "step": 1897
+    },
+    {
+      "epoch": 0.14598877009460812,
+      "learning_rate": 0.0028449849849658117,
+      "loss": 1.4871,
+      "step": 1898
+    },
+    {
+      "epoch": 0.14606568725482655,
+      "learning_rate": 0.0028448244734331505,
+      "loss": 0.7954,
+      "step": 1899
+    },
+    {
+      "epoch": 0.146142604415045,
+      "learning_rate": 0.0028446638833747864,
+      "loss": 0.9552,
+      "step": 1900
+    },
+    {
+      "epoch": 0.14621952157526344,
+      "learning_rate": 0.0028445032148000965,
+      "loss": 1.2515,
+      "step": 1901
+    },
+    {
+      "epoch": 0.1462964387354819,
+      "learning_rate": 0.002844342467718463,
+      "loss": 1.0473,
+      "step": 1902
+    },
+    {
+      "epoch": 0.14637335589570033,
+      "learning_rate": 0.0028441816421392718,
+      "loss": 0.8347,
+      "step": 1903
+    },
+    {
+      "epoch": 0.1464502730559188,
+      "learning_rate": 0.0028440207380719135,
+      "loss": 1.0339,
+      "step": 1904
+    },
+    {
+      "epoch": 0.14652719021613722,
+      "learning_rate": 0.0028438597555257837,
+      "loss": 1.388,
+      "step": 1905
+    },
+    {
+      "epoch": 0.14660410737635565,
+      "learning_rate": 0.002843698694510282,
+      "loss": 0.9474,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1466810245365741,
+      "learning_rate": 0.0028435375550348134,
+      "loss": 1.3463,
+      "step": 1907
+    },
+    {
+      "epoch": 0.14675794169679254,
+      "learning_rate": 0.002843376337108787,
+      "loss": 1.4996,
+      "step": 1908
+    },
+    {
+      "epoch": 0.146834858857011,
+      "learning_rate": 0.002843215040741616,
+      "loss": 1.5202,
+      "step": 1909
+    },
+    {
+      "epoch": 0.14691177601722943,
+      "learning_rate": 0.002843053665942719,
+      "loss": 0.8599,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1469886931774479,
+      "learning_rate": 0.0028428922127215184,
+      "loss": 0.9895,
+      "step": 1911
+    },
+    {
+      "epoch": 0.14706561033766632,
+      "learning_rate": 0.002842730681087442,
+      "loss": 1.2132,
+      "step": 1912
+    },
+    {
+      "epoch": 0.14714252749788478,
+      "learning_rate": 0.002842569071049922,
+      "loss": 1.0408,
+      "step": 1913
+    },
+    {
+      "epoch": 0.14721944465810322,
+      "learning_rate": 0.002842407382618395,
+      "loss": 0.9303,
+      "step": 1914
+    },
+    {
+      "epoch": 0.14729636181832167,
+      "learning_rate": 0.0028422456158023017,
+      "loss": 1.5299,
+      "step": 1915
+    },
+    {
+      "epoch": 0.1473732789785401,
+      "learning_rate": 0.0028420837706110876,
+      "loss": 1.1313,
+      "step": 1916
+    },
+    {
+      "epoch": 0.14745019613875857,
+      "learning_rate": 0.002841921847054204,
+      "loss": 1.3461,
+      "step": 1917
+    },
+    {
+      "epoch": 0.147527113298977,
+      "learning_rate": 0.002841759845141105,
+      "loss": 1.2145,
+      "step": 1918
+    },
+    {
+      "epoch": 0.14760403045919546,
+      "learning_rate": 0.0028415977648812503,
+      "loss": 1.4611,
+      "step": 1919
+    },
+    {
+      "epoch": 0.1476809476194139,
+      "learning_rate": 0.002841435606284104,
+      "loss": 0.866,
+      "step": 1920
+    },
+    {
+      "epoch": 0.14775786477963235,
+      "learning_rate": 0.0028412733693591345,
+      "loss": 1.0392,
+      "step": 1921
+    },
+    {
+      "epoch": 0.14783478193985078,
+      "learning_rate": 0.002841111054115815,
+      "loss": 1.1667,
+      "step": 1922
+    },
+    {
+      "epoch": 0.14791169910006924,
+      "learning_rate": 0.0028409486605636236,
+      "loss": 1.4217,
+      "step": 1923
+    },
+    {
+      "epoch": 0.14798861626028767,
+      "learning_rate": 0.002840786188712042,
+      "loss": 1.1517,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1480655334205061,
+      "learning_rate": 0.0028406236385705584,
+      "loss": 0.9707,
+      "step": 1925
+    },
+    {
+      "epoch": 0.14814245058072456,
+      "learning_rate": 0.0028404610101486626,
+      "loss": 1.2975,
+      "step": 1926
+    },
+    {
+      "epoch": 0.148219367740943,
+      "learning_rate": 0.002840298303455852,
+      "loss": 1.2556,
+      "step": 1927
+    },
+    {
+      "epoch": 0.14829628490116145,
+      "learning_rate": 0.0028401355185016263,
+      "loss": 1.1761,
+      "step": 1928
+    },
+    {
+      "epoch": 0.14837320206137988,
+      "learning_rate": 0.002839972655295491,
+      "loss": 0.9831,
+      "step": 1929
+    },
+    {
+      "epoch": 0.14845011922159834,
+      "learning_rate": 0.002839809713846956,
+      "loss": 1.1987,
+      "step": 1930
+    },
+    {
+      "epoch": 0.14852703638181677,
+      "learning_rate": 0.0028396466941655357,
+      "loss": 1.0997,
+      "step": 1931
+    },
+    {
+      "epoch": 0.14860395354203523,
+      "learning_rate": 0.002839483596260748,
+      "loss": 1.0292,
+      "step": 1932
+    },
+    {
+      "epoch": 0.14868087070225366,
+      "learning_rate": 0.002839320420142118,
+      "loss": 1.0209,
+      "step": 1933
+    },
+    {
+      "epoch": 0.14875778786247212,
+      "learning_rate": 0.0028391571658191723,
+      "loss": 1.2296,
+      "step": 1934
+    },
+    {
+      "epoch": 0.14883470502269056,
+      "learning_rate": 0.0028389938333014445,
+      "loss": 1.1078,
+      "step": 1935
+    },
+    {
+      "epoch": 0.14891162218290901,
+      "learning_rate": 0.002838830422598471,
+      "loss": 0.9301,
+      "step": 1936
+    },
+    {
+      "epoch": 0.14898853934312745,
+      "learning_rate": 0.002838666933719794,
+      "loss": 1.3642,
+      "step": 1937
+    },
+    {
+      "epoch": 0.1490654565033459,
+      "learning_rate": 0.0028385033666749595,
+      "loss": 1.2228,
+      "step": 1938
+    },
+    {
+      "epoch": 0.14914237366356434,
+      "learning_rate": 0.002838339721473518,
+      "loss": 1.2052,
+      "step": 1939
+    },
+    {
+      "epoch": 0.1492192908237828,
+      "learning_rate": 0.0028381759981250263,
+      "loss": 1.1428,
+      "step": 1940
+    },
+    {
+      "epoch": 0.14929620798400123,
+      "learning_rate": 0.0028380121966390427,
+      "loss": 1.3212,
+      "step": 1941
+    },
+    {
+      "epoch": 0.1493731251442197,
+      "learning_rate": 0.0028378483170251327,
+      "loss": 0.8909,
+      "step": 1942
+    },
+    {
+      "epoch": 0.14945004230443812,
+      "learning_rate": 0.0028376843592928655,
+      "loss": 1.0285,
+      "step": 1943
+    },
+    {
+      "epoch": 0.14952695946465658,
+      "learning_rate": 0.002837520323451814,
+      "loss": 0.9375,
+      "step": 1944
+    },
+    {
+      "epoch": 0.149603876624875,
+      "learning_rate": 0.002837356209511557,
+      "loss": 1.2237,
+      "step": 1945
+    },
+    {
+      "epoch": 0.14968079378509344,
+      "learning_rate": 0.0028371920174816777,
+      "loss": 1.1826,
+      "step": 1946
+    },
+    {
+      "epoch": 0.1497577109453119,
+      "learning_rate": 0.0028370277473717628,
+      "loss": 1.1488,
+      "step": 1947
+    },
+    {
+      "epoch": 0.14983462810553033,
+      "learning_rate": 0.0028368633991914037,
+      "loss": 1.0076,
+      "step": 1948
+    },
+    {
+      "epoch": 0.1499115452657488,
+      "learning_rate": 0.0028366989729501977,
+      "loss": 0.9484,
+      "step": 1949
+    },
+    {
+      "epoch": 0.14998846242596722,
+      "learning_rate": 0.0028365344686577462,
+      "loss": 1.055,
+      "step": 1950
+    },
+    {
+      "epoch": 0.15006537958618568,
+      "learning_rate": 0.0028363698863236534,
+      "loss": 1.0813,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1501422967464041,
+      "learning_rate": 0.0028362052259575305,
+      "loss": 1.2597,
+      "step": 1952
+    },
+    {
+      "epoch": 0.15021921390662257,
+      "learning_rate": 0.002836040487568992,
+      "loss": 1.1157,
+      "step": 1953
+    },
+    {
+      "epoch": 0.150296131066841,
+      "learning_rate": 0.002835875671167657,
+      "loss": 1.2661,
+      "step": 1954
+    },
+    {
+      "epoch": 0.15037304822705946,
+      "learning_rate": 0.002835710776763149,
+      "loss": 1.3811,
+      "step": 1955
+    },
+    {
+      "epoch": 0.1504499653872779,
+      "learning_rate": 0.002835545804365097,
+      "loss": 1.2374,
+      "step": 1956
+    },
+    {
+      "epoch": 0.15052688254749635,
+      "learning_rate": 0.0028353807539831335,
+      "loss": 1.2996,
+      "step": 1957
+    },
+    {
+      "epoch": 0.15060379970771479,
+      "learning_rate": 0.002835215625626896,
+      "loss": 1.39,
+      "step": 1958
+    },
+    {
+      "epoch": 0.15068071686793325,
+      "learning_rate": 0.002835050419306027,
+      "loss": 0.9618,
+      "step": 1959
+    },
+    {
+      "epoch": 0.15075763402815168,
+      "learning_rate": 0.0028348851350301725,
+      "loss": 0.8978,
+      "step": 1960
+    },
+    {
+      "epoch": 0.15083455118837014,
+      "learning_rate": 0.0028347197728089838,
+      "loss": 1.002,
+      "step": 1961
+    },
+    {
+      "epoch": 0.15091146834858857,
+      "learning_rate": 0.0028345543326521163,
+      "loss": 1.0203,
+      "step": 1962
+    },
+    {
+      "epoch": 0.15098838550880703,
+      "learning_rate": 0.002834388814569231,
+      "loss": 1.1797,
+      "step": 1963
+    },
+    {
+      "epoch": 0.15106530266902546,
+      "learning_rate": 0.0028342232185699914,
+      "loss": 1.3316,
+      "step": 1964
+    },
+    {
+      "epoch": 0.15114221982924392,
+      "learning_rate": 0.002834057544664068,
+      "loss": 0.8223,
+      "step": 1965
+    },
+    {
+      "epoch": 0.15121913698946235,
+      "learning_rate": 0.002833891792861134,
+      "loss": 1.2784,
+      "step": 1966
+    },
+    {
+      "epoch": 0.15129605414968078,
+      "learning_rate": 0.002833725963170868,
+      "loss": 0.9315,
+      "step": 1967
+    },
+    {
+      "epoch": 0.15137297130989924,
+      "learning_rate": 0.002833560055602954,
+      "loss": 0.9381,
+      "step": 1968
+    },
+    {
+      "epoch": 0.15144988847011767,
+      "learning_rate": 0.0028333940701670774,
+      "loss": 0.8995,
+      "step": 1969
+    },
+    {
+      "epoch": 0.15152680563033613,
+      "learning_rate": 0.002833228006872932,
+      "loss": 1.3886,
+      "step": 1970
+    },
+    {
+      "epoch": 0.15160372279055456,
+      "learning_rate": 0.0028330618657302136,
+      "loss": 1.0036,
+      "step": 1971
+    },
+    {
+      "epoch": 0.15168063995077302,
+      "learning_rate": 0.0028328956467486236,
+      "loss": 1.2779,
+      "step": 1972
+    },
+    {
+      "epoch": 0.15175755711099145,
+      "learning_rate": 0.0028327293499378677,
+      "loss": 1.2057,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1518344742712099,
+      "learning_rate": 0.002832562975307656,
+      "loss": 1.0431,
+      "step": 1974
+    },
+    {
+      "epoch": 0.15191139143142834,
+      "learning_rate": 0.0028323965228677036,
+      "loss": 1.0746,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1519883085916468,
+      "learning_rate": 0.00283222999262773,
+      "loss": 1.0914,
+      "step": 1976
+    },
+    {
+      "epoch": 0.15206522575186524,
+      "learning_rate": 0.0028320633845974577,
+      "loss": 1.291,
+      "step": 1977
+    },
+    {
+      "epoch": 0.1521421429120837,
+      "learning_rate": 0.002831896698786617,
+      "loss": 1.1502,
+      "step": 1978
+    },
+    {
+      "epoch": 0.15221906007230213,
+      "learning_rate": 0.00283172993520494,
+      "loss": 1.2315,
+      "step": 1979
+    },
+    {
+      "epoch": 0.15229597723252059,
+      "learning_rate": 0.0028315630938621633,
+      "loss": 1.1341,
+      "step": 1980
+    },
+    {
+      "epoch": 0.15237289439273902,
+      "learning_rate": 0.0028313961747680307,
+      "loss": 1.3974,
+      "step": 1981
+    },
+    {
+      "epoch": 0.15244981155295748,
+      "learning_rate": 0.0028312291779322876,
+      "loss": 1.0439,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1525267287131759,
+      "learning_rate": 0.0028310621033646854,
+      "loss": 1.0077,
+      "step": 1983
+    },
+    {
+      "epoch": 0.15260364587339437,
+      "learning_rate": 0.00283089495107498,
+      "loss": 1.6118,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1526805630336128,
+      "learning_rate": 0.0028307277210729313,
+      "loss": 0.9113,
+      "step": 1985
+    },
+    {
+      "epoch": 0.15275748019383123,
+      "learning_rate": 0.002830560413368304,
+      "loss": 1.3026,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1528343973540497,
+      "learning_rate": 0.002830393027970868,
+      "loss": 1.131,
+      "step": 1987
+    },
+    {
+      "epoch": 0.15291131451426812,
+      "learning_rate": 0.002830225564890396,
+      "loss": 1.2247,
+      "step": 1988
+    },
+    {
+      "epoch": 0.15298823167448658,
+      "learning_rate": 0.002830058024136667,
+      "loss": 1.2465,
+      "step": 1989
+    },
+    {
+      "epoch": 0.153065148834705,
+      "learning_rate": 0.002829890405719464,
+      "loss": 1.0329,
+      "step": 1990
+    },
+    {
+      "epoch": 0.15314206599492347,
+      "learning_rate": 0.0028297227096485736,
+      "loss": 1.2137,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1532189831551419,
+      "learning_rate": 0.0028295549359337894,
+      "loss": 1.1542,
+      "step": 1992
+    },
+    {
+      "epoch": 0.15329590031536036,
+      "learning_rate": 0.002829387084584906,
+      "loss": 1.127,
+      "step": 1993
+    },
+    {
+      "epoch": 0.1533728174755788,
+      "learning_rate": 0.0028292191556117258,
+      "loss": 0.8747,
+      "step": 1994
+    },
+    {
+      "epoch": 0.15344973463579725,
+      "learning_rate": 0.002829051149024054,
+      "loss": 1.1015,
+      "step": 1995
+    },
+    {
+      "epoch": 0.15352665179601568,
+      "learning_rate": 0.0028288830648317,
+      "loss": 0.9848,
+      "step": 1996
+    },
+    {
+      "epoch": 0.15360356895623414,
+      "learning_rate": 0.0028287149030444786,
+      "loss": 1.3574,
+      "step": 1997
+    },
+    {
+      "epoch": 0.15368048611645257,
+      "learning_rate": 0.00282854666367221,
+      "loss": 1.0259,
+      "step": 1998
+    },
+    {
+      "epoch": 0.15375740327667103,
+      "learning_rate": 0.002828378346724717,
+      "loss": 1.081,
+      "step": 1999
+    },
+    {
+      "epoch": 0.15383432043688947,
+      "learning_rate": 0.0028282099522118275,
+      "loss": 0.9127,
+      "step": 2000
+    },
+    {
+      "epoch": 0.15391123759710792,
+      "learning_rate": 0.002828041480143375,
+      "loss": 1.1192,
+      "step": 2001
+    },
+    {
+      "epoch": 0.15398815475732636,
+      "learning_rate": 0.0028278729305291966,
+      "loss": 1.2785,
+      "step": 2002
+    },
+    {
+      "epoch": 0.15406507191754482,
+      "learning_rate": 0.0028277043033791338,
+      "loss": 1.0425,
+      "step": 2003
+    },
+    {
+      "epoch": 0.15414198907776325,
+      "learning_rate": 0.0028275355987030333,
+      "loss": 1.3098,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1542189062379817,
+      "learning_rate": 0.002827366816510745,
+      "loss": 1.0905,
+      "step": 2005
+    },
+    {
+      "epoch": 0.15429582339820014,
+      "learning_rate": 0.0028271979568121252,
+      "loss": 1.1252,
+      "step": 2006
+    },
+    {
+      "epoch": 0.15437274055841857,
+      "learning_rate": 0.0028270290196170337,
+      "loss": 1.1481,
+      "step": 2007
+    },
+    {
+      "epoch": 0.15444965771863703,
+      "learning_rate": 0.0028268600049353343,
+      "loss": 1.1522,
+      "step": 2008
+    },
+    {
+      "epoch": 0.15452657487885546,
+      "learning_rate": 0.002826690912776897,
+      "loss": 1.1175,
+      "step": 2009
+    },
+    {
+      "epoch": 0.15460349203907392,
+      "learning_rate": 0.0028265217431515943,
+      "loss": 0.8834,
+      "step": 2010
+    },
+    {
+      "epoch": 0.15468040919929235,
+      "learning_rate": 0.0028263524960693044,
+      "loss": 1.1786,
+      "step": 2011
+    },
+    {
+      "epoch": 0.1547573263595108,
+      "learning_rate": 0.0028261831715399107,
+      "loss": 1.085,
+      "step": 2012
+    },
+    {
+      "epoch": 0.15483424351972924,
+      "learning_rate": 0.0028260137695732984,
+      "loss": 1.3008,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1549111606799477,
+      "learning_rate": 0.0028258442901793607,
+      "loss": 1.0226,
+      "step": 2014
+    },
+    {
+      "epoch": 0.15498807784016613,
+      "learning_rate": 0.0028256747333679933,
+      "loss": 1.5447,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1550649950003846,
+      "learning_rate": 0.0028255050991490963,
+      "loss": 1.1576,
+      "step": 2016
+    },
+    {
+      "epoch": 0.15514191216060302,
+      "learning_rate": 0.0028253353875325755,
+      "loss": 0.6988,
+      "step": 2017
+    },
+    {
+      "epoch": 0.15521882932082148,
+      "learning_rate": 0.00282516559852834,
+      "loss": 1.0037,
+      "step": 2018
+    },
+    {
+      "epoch": 0.15529574648103991,
+      "learning_rate": 0.0028249957321463044,
+      "loss": 1.1871,
+      "step": 2019
+    },
+    {
+      "epoch": 0.15537266364125837,
+      "learning_rate": 0.002824825788396387,
+      "loss": 0.9937,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1554495808014768,
+      "learning_rate": 0.0028246557672885106,
+      "loss": 1.0193,
+      "step": 2021
+    },
+    {
+      "epoch": 0.15552649796169526,
+      "learning_rate": 0.0028244856688326037,
+      "loss": 1.2292,
+      "step": 2022
+    },
+    {
+      "epoch": 0.1556034151219137,
+      "learning_rate": 0.0028243154930385986,
+      "loss": 1.3331,
+      "step": 2023
+    },
+    {
+      "epoch": 0.15568033228213216,
+      "learning_rate": 0.002824145239916432,
+      "loss": 1.2292,
+      "step": 2024
+    },
+    {
+      "epoch": 0.1557572494423506,
+      "learning_rate": 0.002823974909476044,
+      "loss": 1.4199,
+      "step": 2025
+    },
+    {
+      "epoch": 0.15583416660256905,
+      "learning_rate": 0.0028238045017273815,
+      "loss": 1.2479,
+      "step": 2026
+    },
+    {
+      "epoch": 0.15591108376278748,
+      "learning_rate": 0.0028236340166803947,
+      "loss": 1.1911,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1559880009230059,
+      "learning_rate": 0.0028234634543450377,
+      "loss": 0.9725,
+      "step": 2028
+    },
+    {
+      "epoch": 0.15606491808322437,
+      "learning_rate": 0.002823292814731271,
+      "loss": 1.4719,
+      "step": 2029
+    },
+    {
+      "epoch": 0.1561418352434428,
+      "learning_rate": 0.0028231220978490574,
+      "loss": 0.932,
+      "step": 2030
+    },
+    {
+      "epoch": 0.15621875240366126,
+      "learning_rate": 0.002822951303708366,
+      "loss": 1.1865,
+      "step": 2031
+    },
+    {
+      "epoch": 0.1562956695638797,
+      "learning_rate": 0.0028227804323191687,
+      "loss": 1.0766,
+      "step": 2032
+    },
+    {
+      "epoch": 0.15637258672409815,
+      "learning_rate": 0.0028226094836914435,
+      "loss": 1.0572,
+      "step": 2033
+    },
+    {
+      "epoch": 0.15644950388431658,
+      "learning_rate": 0.002822438457835172,
+      "loss": 1.0668,
+      "step": 2034
+    },
+    {
+      "epoch": 0.15652642104453504,
+      "learning_rate": 0.002822267354760341,
+      "loss": 1.2249,
+      "step": 2035
+    },
+    {
+      "epoch": 0.15660333820475347,
+      "learning_rate": 0.002822096174476941,
+      "loss": 1.1557,
+      "step": 2036
+    },
+    {
+      "epoch": 0.15668025536497193,
+      "learning_rate": 0.0028219249169949678,
+      "loss": 1.1815,
+      "step": 2037
+    },
+    {
+      "epoch": 0.15675717252519036,
+      "learning_rate": 0.0028217535823244206,
+      "loss": 0.876,
+      "step": 2038
+    },
+    {
+      "epoch": 0.15683408968540882,
+      "learning_rate": 0.0028215821704753044,
+      "loss": 1.365,
+      "step": 2039
+    },
+    {
+      "epoch": 0.15691100684562725,
+      "learning_rate": 0.002821410681457628,
+      "loss": 1.1182,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1569879240058457,
+      "learning_rate": 0.0028212391152814048,
+      "loss": 0.9862,
+      "step": 2041
+    },
+    {
+      "epoch": 0.15706484116606415,
+      "learning_rate": 0.0028210674719566526,
+      "loss": 1.2278,
+      "step": 2042
+    },
+    {
+      "epoch": 0.1571417583262826,
+      "learning_rate": 0.002820895751493394,
+      "loss": 1.8615,
+      "step": 2043
+    },
+    {
+      "epoch": 0.15721867548650104,
+      "learning_rate": 0.002820723953901656,
+      "loss": 1.2583,
+      "step": 2044
+    },
+    {
+      "epoch": 0.1572955926467195,
+      "learning_rate": 0.0028205520791914694,
+      "loss": 0.9442,
+      "step": 2045
+    },
+    {
+      "epoch": 0.15737250980693793,
+      "learning_rate": 0.0028203801273728713,
+      "loss": 1.3713,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1574494269671564,
+      "learning_rate": 0.0028202080984559012,
+      "loss": 1.2726,
+      "step": 2047
+    },
+    {
+      "epoch": 0.15752634412737482,
+      "learning_rate": 0.002820035992450604,
+      "loss": 1.0752,
+      "step": 2048
+    },
+    {
+      "epoch": 0.15760326128759325,
+      "learning_rate": 0.00281986380936703,
+      "loss": 1.2511,
+      "step": 2049
+    },
+    {
+      "epoch": 0.1576801784478117,
+      "learning_rate": 0.002819691549215233,
+      "loss": 1.3435,
+      "step": 2050
+    },
+    {
+      "epoch": 0.15775709560803014,
+      "learning_rate": 0.00281951921200527,
+      "loss": 1.2046,
+      "step": 2051
+    },
+    {
+      "epoch": 0.1578340127682486,
+      "learning_rate": 0.0028193467977472055,
+      "loss": 1.1837,
+      "step": 2052
+    },
+    {
+      "epoch": 0.15791092992846703,
+      "learning_rate": 0.002819174306451107,
+      "loss": 1.4107,
+      "step": 2053
+    },
+    {
+      "epoch": 0.1579878470886855,
+      "learning_rate": 0.002819001738127045,
+      "loss": 1.3604,
+      "step": 2054
+    },
+    {
+      "epoch": 0.15806476424890392,
+      "learning_rate": 0.002818829092785098,
+      "loss": 1.2017,
+      "step": 2055
+    },
+    {
+      "epoch": 0.15814168140912238,
+      "learning_rate": 0.002818656370435345,
+      "loss": 1.2019,
+      "step": 2056
+    },
+    {
+      "epoch": 0.1582185985693408,
+      "learning_rate": 0.0028184835710878724,
+      "loss": 0.9851,
+      "step": 2057
+    },
+    {
+      "epoch": 0.15829551572955927,
+      "learning_rate": 0.00281831069475277,
+      "loss": 0.898,
+      "step": 2058
+    },
+    {
+      "epoch": 0.1583724328897777,
+      "learning_rate": 0.0028181377414401325,
+      "loss": 0.919,
+      "step": 2059
+    },
+    {
+      "epoch": 0.15844935004999616,
+      "learning_rate": 0.002817964711160058,
+      "loss": 1.5117,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1585262672102146,
+      "learning_rate": 0.0028177916039226506,
+      "loss": 1.1425,
+      "step": 2061
+    },
+    {
+      "epoch": 0.15860318437043305,
+      "learning_rate": 0.0028176184197380182,
+      "loss": 1.2193,
+      "step": 2062
+    },
+    {
+      "epoch": 0.15868010153065149,
+      "learning_rate": 0.0028174451586162738,
+      "loss": 0.95,
+      "step": 2063
+    },
+    {
+      "epoch": 0.15875701869086994,
+      "learning_rate": 0.002817271820567533,
+      "loss": 1.0872,
+      "step": 2064
+    },
+    {
+      "epoch": 0.15883393585108838,
+      "learning_rate": 0.002817098405601917,
+      "loss": 1.1749,
+      "step": 2065
+    },
+    {
+      "epoch": 0.15891085301130684,
+      "learning_rate": 0.0028169249137295536,
+      "loss": 1.0426,
+      "step": 2066
+    },
+    {
+      "epoch": 0.15898777017152527,
+      "learning_rate": 0.0028167513449605715,
+      "loss": 1.4273,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1590646873317437,
+      "learning_rate": 0.0028165776993051062,
+      "loss": 0.7961,
+      "step": 2068
+    },
+    {
+      "epoch": 0.15914160449196216,
+      "learning_rate": 0.002816403976773297,
+      "loss": 1.0494,
+      "step": 2069
+    },
+    {
+      "epoch": 0.1592185216521806,
+      "learning_rate": 0.0028162301773752875,
+      "loss": 1.3031,
+      "step": 2070
+    },
+    {
+      "epoch": 0.15929543881239905,
+      "learning_rate": 0.002816056301121226,
+      "loss": 0.9875,
+      "step": 2071
+    },
+    {
+      "epoch": 0.15937235597261748,
+      "learning_rate": 0.0028158823480212663,
+      "loss": 1.3561,
+      "step": 2072
+    },
+    {
+      "epoch": 0.15944927313283594,
+      "learning_rate": 0.002815708318085564,
+      "loss": 1.1828,
+      "step": 2073
+    },
+    {
+      "epoch": 0.15952619029305437,
+      "learning_rate": 0.0028155342113242825,
+      "loss": 1.1529,
+      "step": 2074
+    },
+    {
+      "epoch": 0.15960310745327283,
+      "learning_rate": 0.002815360027747587,
+      "loss": 1.2199,
+      "step": 2075
+    },
+    {
+      "epoch": 0.15968002461349126,
+      "learning_rate": 0.002815185767365649,
+      "loss": 1.6978,
+      "step": 2076
+    },
+    {
+      "epoch": 0.15975694177370972,
+      "learning_rate": 0.0028150114301886426,
+      "loss": 1.0705,
+      "step": 2077
+    },
+    {
+      "epoch": 0.15983385893392815,
+      "learning_rate": 0.0028148370162267493,
+      "loss": 1.4204,
+      "step": 2078
+    },
+    {
+      "epoch": 0.1599107760941466,
+      "learning_rate": 0.0028146625254901517,
+      "loss": 1.2823,
+      "step": 2079
+    },
+    {
+      "epoch": 0.15998769325436504,
+      "learning_rate": 0.0028144879579890393,
+      "loss": 1.0765,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1600646104145835,
+      "learning_rate": 0.0028143133137336056,
+      "loss": 1.034,
+      "step": 2081
+    },
+    {
+      "epoch": 0.16014152757480193,
+      "learning_rate": 0.002814138592734047,
+      "loss": 1.1956,
+      "step": 2082
+    },
+    {
+      "epoch": 0.1602184447350204,
+      "learning_rate": 0.0028139637950005667,
+      "loss": 1.0665,
+      "step": 2083
+    },
+    {
+      "epoch": 0.16029536189523882,
+      "learning_rate": 0.002813788920543371,
+      "loss": 1.1069,
+      "step": 2084
+    },
+    {
+      "epoch": 0.16037227905545728,
+      "learning_rate": 0.002813613969372671,
+      "loss": 1.3421,
+      "step": 2085
+    },
+    {
+      "epoch": 0.16044919621567572,
+      "learning_rate": 0.002813438941498682,
+      "loss": 1.2892,
+      "step": 2086
+    },
+    {
+      "epoch": 0.16052611337589417,
+      "learning_rate": 0.002813263836931625,
+      "loss": 1.2271,
+      "step": 2087
+    },
+    {
+      "epoch": 0.1606030305361126,
+      "learning_rate": 0.002813088655681724,
+      "loss": 0.9485,
+      "step": 2088
+    },
+    {
+      "epoch": 0.16067994769633104,
+      "learning_rate": 0.0028129133977592073,
+      "loss": 1.8525,
+      "step": 2089
+    },
+    {
+      "epoch": 0.1607568648565495,
+      "learning_rate": 0.002812738063174309,
+      "loss": 1.0006,
+      "step": 2090
+    },
+    {
+      "epoch": 0.16083378201676793,
+      "learning_rate": 0.0028125626519372674,
+      "loss": 1.1479,
+      "step": 2091
+    },
+    {
+      "epoch": 0.1609106991769864,
+      "learning_rate": 0.002812387164058325,
+      "loss": 0.7483,
+      "step": 2092
+    },
+    {
+      "epoch": 0.16098761633720482,
+      "learning_rate": 0.002812211599547728,
+      "loss": 1.3081,
+      "step": 2093
+    },
+    {
+      "epoch": 0.16106453349742328,
+      "learning_rate": 0.002812035958415728,
+      "loss": 1.1483,
+      "step": 2094
+    },
+    {
+      "epoch": 0.1611414506576417,
+      "learning_rate": 0.0028118602406725815,
+      "loss": 1.334,
+      "step": 2095
+    },
+    {
+      "epoch": 0.16121836781786017,
+      "learning_rate": 0.002811684446328548,
+      "loss": 1.0362,
+      "step": 2096
+    },
+    {
+      "epoch": 0.1612952849780786,
+      "learning_rate": 0.0028115085753938926,
+      "loss": 0.8121,
+      "step": 2097
+    },
+    {
+      "epoch": 0.16137220213829706,
+      "learning_rate": 0.0028113326278788845,
+      "loss": 1.4638,
+      "step": 2098
+    },
+    {
+      "epoch": 0.1614491192985155,
+      "learning_rate": 0.002811156603793798,
+      "loss": 1.3577,
+      "step": 2099
+    },
+    {
+      "epoch": 0.16152603645873395,
+      "learning_rate": 0.002810980503148911,
+      "loss": 1.1596,
+      "step": 2100
+    },
+    {
+      "epoch": 0.16160295361895238,
+      "learning_rate": 0.0028108043259545057,
+      "loss": 1.3218,
+      "step": 2101
+    },
+    {
+      "epoch": 0.16167987077917084,
+      "learning_rate": 0.0028106280722208703,
+      "loss": 1.068,
+      "step": 2102
+    },
+    {
+      "epoch": 0.16175678793938927,
+      "learning_rate": 0.0028104517419582954,
+      "loss": 1.2523,
+      "step": 2103
+    },
+    {
+      "epoch": 0.16183370509960773,
+      "learning_rate": 0.0028102753351770776,
+      "loss": 1.2487,
+      "step": 2104
+    },
+    {
+      "epoch": 0.16191062225982616,
+      "learning_rate": 0.0028100988518875176,
+      "loss": 1.0539,
+      "step": 2105
+    },
+    {
+      "epoch": 0.16198753942004462,
+      "learning_rate": 0.0028099222920999205,
+      "loss": 0.9122,
+      "step": 2106
+    },
+    {
+      "epoch": 0.16206445658026306,
+      "learning_rate": 0.002809745655824595,
+      "loss": 1.2692,
+      "step": 2107
+    },
+    {
+      "epoch": 0.16214137374048151,
+      "learning_rate": 0.002809568943071856,
+      "loss": 1.0734,
+      "step": 2108
+    },
+    {
+      "epoch": 0.16221829090069995,
+      "learning_rate": 0.002809392153852022,
+      "loss": 1.3023,
+      "step": 2109
+    },
+    {
+      "epoch": 0.16229520806091838,
+      "learning_rate": 0.002809215288175415,
+      "loss": 1.0414,
+      "step": 2110
+    },
+    {
+      "epoch": 0.16237212522113684,
+      "learning_rate": 0.0028090383460523624,
+      "loss": 1.6666,
+      "step": 2111
+    },
+    {
+      "epoch": 0.16244904238135527,
+      "learning_rate": 0.0028088613274931977,
+      "loss": 1.0372,
+      "step": 2112
+    },
+    {
+      "epoch": 0.16252595954157373,
+      "learning_rate": 0.0028086842325082554,
+      "loss": 1.187,
+      "step": 2113
+    },
+    {
+      "epoch": 0.16260287670179216,
+      "learning_rate": 0.002808507061107877,
+      "loss": 1.1738,
+      "step": 2114
+    },
+    {
+      "epoch": 0.16267979386201062,
+      "learning_rate": 0.0028083298133024072,
+      "loss": 0.9106,
+      "step": 2115
+    },
+    {
+      "epoch": 0.16275671102222905,
+      "learning_rate": 0.0028081524891021966,
+      "loss": 1.1799,
+      "step": 2116
+    },
+    {
+      "epoch": 0.1628336281824475,
+      "learning_rate": 0.002807975088517599,
+      "loss": 1.3328,
+      "step": 2117
+    },
+    {
+      "epoch": 0.16291054534266594,
+      "learning_rate": 0.002807797611558972,
+      "loss": 1.0491,
+      "step": 2118
+    },
+    {
+      "epoch": 0.1629874625028844,
+      "learning_rate": 0.00280762005823668,
+      "loss": 1.1215,
+      "step": 2119
+    },
+    {
+      "epoch": 0.16306437966310283,
+      "learning_rate": 0.0028074424285610905,
+      "loss": 0.8615,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1631412968233213,
+      "learning_rate": 0.0028072647225425747,
+      "loss": 1.1726,
+      "step": 2121
+    },
+    {
+      "epoch": 0.16321821398353972,
+      "learning_rate": 0.002807086940191509,
+      "loss": 1.2703,
+      "step": 2122
+    },
+    {
+      "epoch": 0.16329513114375818,
+      "learning_rate": 0.0028069090815182746,
+      "loss": 1.3984,
+      "step": 2123
+    },
+    {
+      "epoch": 0.1633720483039766,
+      "learning_rate": 0.0028067311465332576,
+      "loss": 0.8942,
+      "step": 2124
+    },
+    {
+      "epoch": 0.16344896546419507,
+      "learning_rate": 0.0028065531352468465,
+      "loss": 1.2872,
+      "step": 2125
+    },
+    {
+      "epoch": 0.1635258826244135,
+      "learning_rate": 0.002806375047669437,
+      "loss": 1.2614,
+      "step": 2126
+    },
+    {
+      "epoch": 0.16360279978463196,
+      "learning_rate": 0.002806196883811426,
+      "loss": 1.0975,
+      "step": 2127
+    },
+    {
+      "epoch": 0.1636797169448504,
+      "learning_rate": 0.002806018643683218,
+      "loss": 1.0155,
+      "step": 2128
+    },
+    {
+      "epoch": 0.16375663410506885,
+      "learning_rate": 0.0028058403272952204,
+      "loss": 1.3019,
+      "step": 2129
+    },
+    {
+      "epoch": 0.16383355126528729,
+      "learning_rate": 0.002805661934657845,
+      "loss": 1.2465,
+      "step": 2130
+    },
+    {
+      "epoch": 0.16391046842550572,
+      "learning_rate": 0.002805483465781509,
+      "loss": 1.0104,
+      "step": 2131
+    },
+    {
+      "epoch": 0.16398738558572418,
+      "learning_rate": 0.0028053049206766324,
+      "loss": 0.8387,
+      "step": 2132
+    },
+    {
+      "epoch": 0.1640643027459426,
+      "learning_rate": 0.002805126299353641,
+      "loss": 1.1378,
+      "step": 2133
+    },
+    {
+      "epoch": 0.16414121990616107,
+      "learning_rate": 0.002804947601822965,
+      "loss": 1.3031,
+      "step": 2134
+    },
+    {
+      "epoch": 0.1642181370663795,
+      "learning_rate": 0.002804768828095039,
+      "loss": 1.0428,
+      "step": 2135
+    },
+    {
+      "epoch": 0.16429505422659796,
+      "learning_rate": 0.0028045899781803008,
+      "loss": 1.4127,
+      "step": 2136
+    },
+    {
+      "epoch": 0.1643719713868164,
+      "learning_rate": 0.002804411052089194,
+      "loss": 1.2058,
+      "step": 2137
+    },
+    {
+      "epoch": 0.16444888854703485,
+      "learning_rate": 0.0028042320498321673,
+      "loss": 1.3751,
+      "step": 2138
+    },
+    {
+      "epoch": 0.16452580570725328,
+      "learning_rate": 0.002804052971419671,
+      "loss": 0.6894,
+      "step": 2139
+    },
+    {
+      "epoch": 0.16460272286747174,
+      "learning_rate": 0.002803873816862163,
+      "loss": 1.0824,
+      "step": 2140
+    },
+    {
+      "epoch": 0.16467964002769017,
+      "learning_rate": 0.002803694586170104,
+      "loss": 0.9984,
+      "step": 2141
+    },
+    {
+      "epoch": 0.16475655718790863,
+      "learning_rate": 0.0028035152793539596,
+      "loss": 1.0274,
+      "step": 2142
+    },
+    {
+      "epoch": 0.16483347434812706,
+      "learning_rate": 0.0028033358964241993,
+      "loss": 1.2265,
+      "step": 2143
+    },
+    {
+      "epoch": 0.16491039150834552,
+      "learning_rate": 0.0028031564373912983,
+      "loss": 1.3405,
+      "step": 2144
+    },
+    {
+      "epoch": 0.16498730866856395,
+      "learning_rate": 0.002802976902265734,
+      "loss": 1.0912,
+      "step": 2145
+    },
+    {
+      "epoch": 0.1650642258287824,
+      "learning_rate": 0.0028027972910579907,
+      "loss": 1.3862,
+      "step": 2146
+    },
+    {
+      "epoch": 0.16514114298900084,
+      "learning_rate": 0.002802617603778556,
+      "loss": 1.3538,
+      "step": 2147
+    },
+    {
+      "epoch": 0.1652180601492193,
+      "learning_rate": 0.0028024378404379216,
+      "loss": 1.4892,
+      "step": 2148
+    },
+    {
+      "epoch": 0.16529497730943774,
+      "learning_rate": 0.002802258001046585,
+      "loss": 1.4131,
+      "step": 2149
+    },
+    {
+      "epoch": 0.16537189446965617,
+      "learning_rate": 0.002802078085615046,
+      "loss": 1.2851,
+      "step": 2150
+    },
+    {
+      "epoch": 0.16544881162987463,
+      "learning_rate": 0.0028018980941538103,
+      "loss": 1.3294,
+      "step": 2151
+    },
+    {
+      "epoch": 0.16552572879009306,
+      "learning_rate": 0.0028017180266733882,
+      "loss": 1.2027,
+      "step": 2152
+    },
+    {
+      "epoch": 0.16560264595031152,
+      "learning_rate": 0.002801537883184294,
+      "loss": 1.0872,
+      "step": 2153
+    },
+    {
+      "epoch": 0.16567956311052995,
+      "learning_rate": 0.0028013576636970465,
+      "loss": 1.2291,
+      "step": 2154
+    },
+    {
+      "epoch": 0.1657564802707484,
+      "learning_rate": 0.0028011773682221686,
+      "loss": 1.332,
+      "step": 2155
+    },
+    {
+      "epoch": 0.16583339743096684,
+      "learning_rate": 0.0028009969967701886,
+      "loss": 0.7623,
+      "step": 2156
+    },
+    {
+      "epoch": 0.1659103145911853,
+      "learning_rate": 0.0028008165493516378,
+      "loss": 1.2742,
+      "step": 2157
+    },
+    {
+      "epoch": 0.16598723175140373,
+      "learning_rate": 0.002800636025977053,
+      "loss": 0.6873,
+      "step": 2158
+    },
+    {
+      "epoch": 0.1660641489116222,
+      "learning_rate": 0.002800455426656975,
+      "loss": 1.072,
+      "step": 2159
+    },
+    {
+      "epoch": 0.16614106607184062,
+      "learning_rate": 0.0028002747514019495,
+      "loss": 1.1598,
+      "step": 2160
+    },
+    {
+      "epoch": 0.16621798323205908,
+      "learning_rate": 0.002800094000222526,
+      "loss": 1.0351,
+      "step": 2161
+    },
+    {
+      "epoch": 0.1662949003922775,
+      "learning_rate": 0.002799913173129259,
+      "loss": 1.3774,
+      "step": 2162
+    },
+    {
+      "epoch": 0.16637181755249597,
+      "learning_rate": 0.0027997322701327067,
+      "loss": 1.1639,
+      "step": 2163
+    },
+    {
+      "epoch": 0.1664487347127144,
+      "learning_rate": 0.0027995512912434334,
+      "loss": 1.3926,
+      "step": 2164
+    },
+    {
+      "epoch": 0.16652565187293286,
+      "learning_rate": 0.0027993702364720054,
+      "loss": 1.5027,
+      "step": 2165
+    },
+    {
+      "epoch": 0.1666025690331513,
+      "learning_rate": 0.002799189105828995,
+      "loss": 1.4819,
+      "step": 2166
+    },
+    {
+      "epoch": 0.16667948619336975,
+      "learning_rate": 0.002799007899324979,
+      "loss": 1.1601,
+      "step": 2167
+    },
+    {
+      "epoch": 0.16675640335358818,
+      "learning_rate": 0.002798826616970538,
+      "loss": 1.0782,
+      "step": 2168
+    },
+    {
+      "epoch": 0.16683332051380664,
+      "learning_rate": 0.002798645258776257,
+      "loss": 1.1136,
+      "step": 2169
+    },
+    {
+      "epoch": 0.16691023767402507,
+      "learning_rate": 0.002798463824752726,
+      "loss": 1.3599,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1669871548342435,
+      "learning_rate": 0.0027982823149105386,
+      "loss": 1.3108,
+      "step": 2171
+    },
+    {
+      "epoch": 0.16706407199446197,
+      "learning_rate": 0.0027981007292602945,
+      "loss": 0.9383,
+      "step": 2172
+    },
+    {
+      "epoch": 0.1671409891546804,
+      "learning_rate": 0.0027979190678125957,
+      "loss": 1.1164,
+      "step": 2173
+    },
+    {
+      "epoch": 0.16721790631489886,
+      "learning_rate": 0.00279773733057805,
+      "loss": 1.2451,
+      "step": 2174
+    },
+    {
+      "epoch": 0.1672948234751173,
+      "learning_rate": 0.0027975555175672687,
+      "loss": 0.9756,
+      "step": 2175
+    },
+    {
+      "epoch": 0.16737174063533575,
+      "learning_rate": 0.0027973736287908685,
+      "loss": 1.1213,
+      "step": 2176
+    },
+    {
+      "epoch": 0.16744865779555418,
+      "learning_rate": 0.0027971916642594704,
+      "loss": 1.0953,
+      "step": 2177
+    },
+    {
+      "epoch": 0.16752557495577264,
+      "learning_rate": 0.002797009623983699,
+      "loss": 1.499,
+      "step": 2178
+    },
+    {
+      "epoch": 0.16760249211599107,
+      "learning_rate": 0.0027968275079741844,
+      "loss": 1.0063,
+      "step": 2179
+    },
+    {
+      "epoch": 0.16767940927620953,
+      "learning_rate": 0.002796645316241559,
+      "loss": 1.2216,
+      "step": 2180
+    },
+    {
+      "epoch": 0.16775632643642796,
+      "learning_rate": 0.0027964630487964633,
+      "loss": 1.3844,
+      "step": 2181
+    },
+    {
+      "epoch": 0.16783324359664642,
+      "learning_rate": 0.0027962807056495383,
+      "loss": 1.2136,
+      "step": 2182
+    },
+    {
+      "epoch": 0.16791016075686485,
+      "learning_rate": 0.002796098286811432,
+      "loss": 1.3936,
+      "step": 2183
+    },
+    {
+      "epoch": 0.1679870779170833,
+      "learning_rate": 0.002795915792292797,
+      "loss": 1.2363,
+      "step": 2184
+    },
+    {
+      "epoch": 0.16806399507730174,
+      "learning_rate": 0.0027957332221042867,
+      "loss": 1.2134,
+      "step": 2185
+    },
+    {
+      "epoch": 0.1681409122375202,
+      "learning_rate": 0.002795550576256564,
+      "loss": 1.7939,
+      "step": 2186
+    },
+    {
+      "epoch": 0.16821782939773863,
+      "learning_rate": 0.0027953678547602934,
+      "loss": 1.2313,
+      "step": 2187
+    },
+    {
+      "epoch": 0.1682947465579571,
+      "learning_rate": 0.002795185057626143,
+      "loss": 1.0828,
+      "step": 2188
+    },
+    {
+      "epoch": 0.16837166371817552,
+      "learning_rate": 0.0027950021848647875,
+      "loss": 1.3092,
+      "step": 2189
+    },
+    {
+      "epoch": 0.16844858087839398,
+      "learning_rate": 0.002794819236486905,
+      "loss": 1.3984,
+      "step": 2190
+    },
+    {
+      "epoch": 0.16852549803861241,
+      "learning_rate": 0.0027946362125031783,
+      "loss": 1.1199,
+      "step": 2191
+    },
+    {
+      "epoch": 0.16860241519883085,
+      "learning_rate": 0.0027944531129242933,
+      "loss": 0.9542,
+      "step": 2192
+    },
+    {
+      "epoch": 0.1686793323590493,
+      "learning_rate": 0.002794269937760942,
+      "loss": 1.0747,
+      "step": 2193
+    },
+    {
+      "epoch": 0.16875624951926774,
+      "learning_rate": 0.00279408668702382,
+      "loss": 1.178,
+      "step": 2194
+    },
+    {
+      "epoch": 0.1688331666794862,
+      "learning_rate": 0.002793903360723628,
+      "loss": 1.1669,
+      "step": 2195
+    },
+    {
+      "epoch": 0.16891008383970463,
+      "learning_rate": 0.0027937199588710703,
+      "loss": 0.9588,
+      "step": 2196
+    },
+    {
+      "epoch": 0.1689870009999231,
+      "learning_rate": 0.0027935364814768555,
+      "loss": 1.1285,
+      "step": 2197
+    },
+    {
+      "epoch": 0.16906391816014152,
+      "learning_rate": 0.0027933529285516985,
+      "loss": 0.8729,
+      "step": 2198
+    },
+    {
+      "epoch": 0.16914083532035998,
+      "learning_rate": 0.002793169300106315,
+      "loss": 1.1388,
+      "step": 2199
+    },
+    {
+      "epoch": 0.1692177524805784,
+      "learning_rate": 0.002792985596151429,
+      "loss": 1.0768,
+      "step": 2200
+    },
+    {
+      "epoch": 0.16929466964079687,
+      "learning_rate": 0.002792801816697766,
+      "loss": 0.9755,
+      "step": 2201
+    },
+    {
+      "epoch": 0.1693715868010153,
+      "learning_rate": 0.002792617961756058,
+      "loss": 0.8531,
+      "step": 2202
+    },
+    {
+      "epoch": 0.16944850396123376,
+      "learning_rate": 0.00279243403133704,
+      "loss": 1.5626,
+      "step": 2203
+    },
+    {
+      "epoch": 0.1695254211214522,
+      "learning_rate": 0.0027922500254514517,
+      "loss": 1.0973,
+      "step": 2204
+    },
+    {
+      "epoch": 0.16960233828167065,
+      "learning_rate": 0.002792065944110038,
+      "loss": 0.9663,
+      "step": 2205
+    },
+    {
+      "epoch": 0.16967925544188908,
+      "learning_rate": 0.0027918817873235475,
+      "loss": 1.0623,
+      "step": 2206
+    },
+    {
+      "epoch": 0.16975617260210754,
+      "learning_rate": 0.0027916975551027325,
+      "loss": 1.3159,
+      "step": 2207
+    },
+    {
+      "epoch": 0.16983308976232597,
+      "learning_rate": 0.0027915132474583514,
+      "loss": 1.1395,
+      "step": 2208
+    },
+    {
+      "epoch": 0.16991000692254443,
+      "learning_rate": 0.0027913288644011657,
+      "loss": 1.2279,
+      "step": 2209
+    },
+    {
+      "epoch": 0.16998692408276286,
+      "learning_rate": 0.0027911444059419413,
+      "loss": 1.2001,
+      "step": 2210
+    },
+    {
+      "epoch": 0.17006384124298132,
+      "learning_rate": 0.0027909598720914502,
+      "loss": 0.9281,
+      "step": 2211
+    },
+    {
+      "epoch": 0.17014075840319975,
+      "learning_rate": 0.002790775262860466,
+      "loss": 0.7931,
+      "step": 2212
+    },
+    {
+      "epoch": 0.17021767556341819,
+      "learning_rate": 0.0027905905782597703,
+      "loss": 0.9781,
+      "step": 2213
+    },
+    {
+      "epoch": 0.17029459272363665,
+      "learning_rate": 0.002790405818300145,
+      "loss": 0.8806,
+      "step": 2214
+    },
+    {
+      "epoch": 0.17037150988385508,
+      "learning_rate": 0.002790220982992379,
+      "loss": 1.2641,
+      "step": 2215
+    },
+    {
+      "epoch": 0.17044842704407354,
+      "learning_rate": 0.002790036072347265,
+      "loss": 1.3273,
+      "step": 2216
+    },
+    {
+      "epoch": 0.17052534420429197,
+      "learning_rate": 0.0027898510863756006,
+      "loss": 0.8347,
+      "step": 2217
+    },
+    {
+      "epoch": 0.17060226136451043,
+      "learning_rate": 0.002789666025088187,
+      "loss": 1.4892,
+      "step": 2218
+    },
+    {
+      "epoch": 0.17067917852472886,
+      "learning_rate": 0.00278948088849583,
+      "loss": 0.8616,
+      "step": 2219
+    },
+    {
+      "epoch": 0.17075609568494732,
+      "learning_rate": 0.00278929567660934,
+      "loss": 1.0284,
+      "step": 2220
+    },
+    {
+      "epoch": 0.17083301284516575,
+      "learning_rate": 0.002789110389439532,
+      "loss": 0.8165,
+      "step": 2221
+    },
+    {
+      "epoch": 0.1709099300053842,
+      "learning_rate": 0.0027889250269972246,
+      "loss": 0.7955,
+      "step": 2222
+    },
+    {
+      "epoch": 0.17098684716560264,
+      "learning_rate": 0.0027887395892932415,
+      "loss": 1.1538,
+      "step": 2223
+    },
+    {
+      "epoch": 0.1710637643258211,
+      "learning_rate": 0.0027885540763384112,
+      "loss": 1.2271,
+      "step": 2224
+    },
+    {
+      "epoch": 0.17114068148603953,
+      "learning_rate": 0.0027883684881435653,
+      "loss": 1.0206,
+      "step": 2225
+    },
+    {
+      "epoch": 0.171217598646258,
+      "learning_rate": 0.002788182824719541,
+      "loss": 1.3821,
+      "step": 2226
+    },
+    {
+      "epoch": 0.17129451580647642,
+      "learning_rate": 0.0027879970860771783,
+      "loss": 1.13,
+      "step": 2227
+    },
+    {
+      "epoch": 0.17137143296669488,
+      "learning_rate": 0.002787811272227324,
+      "loss": 1.5313,
+      "step": 2228
+    },
+    {
+      "epoch": 0.1714483501269133,
+      "learning_rate": 0.002787625383180827,
+      "loss": 1.2021,
+      "step": 2229
+    },
+    {
+      "epoch": 0.17152526728713177,
+      "learning_rate": 0.002787439418948542,
+      "loss": 1.143,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1716021844473502,
+      "learning_rate": 0.002787253379541328,
+      "loss": 0.9585,
+      "step": 2231
+    },
+    {
+      "epoch": 0.17167910160756863,
+      "learning_rate": 0.0027870672649700475,
+      "loss": 0.9756,
+      "step": 2232
+    },
+    {
+      "epoch": 0.1717560187677871,
+      "learning_rate": 0.0027868810752455678,
+      "loss": 1.1284,
+      "step": 2233
+    },
+    {
+      "epoch": 0.17183293592800553,
+      "learning_rate": 0.0027866948103787608,
+      "loss": 1.5215,
+      "step": 2234
+    },
+    {
+      "epoch": 0.17190985308822399,
+      "learning_rate": 0.0027865084703805033,
+      "loss": 0.7911,
+      "step": 2235
+    },
+    {
+      "epoch": 0.17198677024844242,
+      "learning_rate": 0.0027863220552616754,
+      "loss": 1.0903,
+      "step": 2236
+    },
+    {
+      "epoch": 0.17206368740866088,
+      "learning_rate": 0.0027861355650331617,
+      "loss": 1.073,
+      "step": 2237
+    },
+    {
+      "epoch": 0.1721406045688793,
+      "learning_rate": 0.0027859489997058523,
+      "loss": 1.1637,
+      "step": 2238
+    },
+    {
+      "epoch": 0.17221752172909777,
+      "learning_rate": 0.0027857623592906404,
+      "loss": 1.2024,
+      "step": 2239
+    },
+    {
+      "epoch": 0.1722944388893162,
+      "learning_rate": 0.002785575643798425,
+      "loss": 1.2597,
+      "step": 2240
+    },
+    {
+      "epoch": 0.17237135604953466,
+      "learning_rate": 0.0027853888532401072,
+      "loss": 1.2187,
+      "step": 2241
+    },
+    {
+      "epoch": 0.1724482732097531,
+      "learning_rate": 0.0027852019876265953,
+      "loss": 1.2744,
+      "step": 2242
+    },
+    {
+      "epoch": 0.17252519036997155,
+      "learning_rate": 0.0027850150469687996,
+      "loss": 1.0699,
+      "step": 2243
+    },
+    {
+      "epoch": 0.17260210753018998,
+      "learning_rate": 0.002784828031277636,
+      "loss": 1.2331,
+      "step": 2244
+    },
+    {
+      "epoch": 0.17267902469040844,
+      "learning_rate": 0.002784640940564025,
+      "loss": 1.2178,
+      "step": 2245
+    },
+    {
+      "epoch": 0.17275594185062687,
+      "learning_rate": 0.0027844537748388898,
+      "loss": 1.3137,
+      "step": 2246
+    },
+    {
+      "epoch": 0.17283285901084533,
+      "learning_rate": 0.002784266534113161,
+      "loss": 1.3004,
+      "step": 2247
+    },
+    {
+      "epoch": 0.17290977617106376,
+      "learning_rate": 0.00278407921839777,
+      "loss": 1.21,
+      "step": 2248
+    },
+    {
+      "epoch": 0.17298669333128222,
+      "learning_rate": 0.0027838918277036558,
+      "loss": 1.0841,
+      "step": 2249
+    },
+    {
+      "epoch": 0.17306361049150065,
+      "learning_rate": 0.0027837043620417595,
+      "loss": 1.35,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1731405276517191,
+      "learning_rate": 0.002783516821423028,
+      "loss": 0.6665,
+      "step": 2251
+    },
+    {
+      "epoch": 0.17321744481193754,
+      "learning_rate": 0.0027833292058584117,
+      "loss": 1.0572,
+      "step": 2252
+    },
+    {
+      "epoch": 0.17329436197215597,
+      "learning_rate": 0.0027831415153588652,
+      "loss": 1.0276,
+      "step": 2253
+    },
+    {
+      "epoch": 0.17337127913237443,
+      "learning_rate": 0.0027829537499353485,
+      "loss": 1.426,
+      "step": 2254
+    },
+    {
+      "epoch": 0.17344819629259287,
+      "learning_rate": 0.0027827659095988255,
+      "loss": 1.1894,
+      "step": 2255
+    },
+    {
+      "epoch": 0.17352511345281132,
+      "learning_rate": 0.0027825779943602635,
+      "loss": 1.0175,
+      "step": 2256
+    },
+    {
+      "epoch": 0.17360203061302976,
+      "learning_rate": 0.0027823900042306363,
+      "loss": 1.2415,
+      "step": 2257
+    },
+    {
+      "epoch": 0.17367894777324822,
+      "learning_rate": 0.0027822019392209204,
+      "loss": 1.3457,
+      "step": 2258
+    },
+    {
+      "epoch": 0.17375586493346665,
+      "learning_rate": 0.0027820137993420967,
+      "loss": 1.537,
+      "step": 2259
+    },
+    {
+      "epoch": 0.1738327820936851,
+      "learning_rate": 0.002781825584605151,
+      "loss": 1.5206,
+      "step": 2260
+    },
+    {
+      "epoch": 0.17390969925390354,
+      "learning_rate": 0.0027816372950210744,
+      "loss": 0.7763,
+      "step": 2261
+    },
+    {
+      "epoch": 0.173986616414122,
+      "learning_rate": 0.002781448930600859,
+      "loss": 1.408,
+      "step": 2262
+    },
+    {
+      "epoch": 0.17406353357434043,
+      "learning_rate": 0.002781260491355506,
+      "loss": 1.8919,
+      "step": 2263
+    },
+    {
+      "epoch": 0.1741404507345589,
+      "learning_rate": 0.002781071977296018,
+      "loss": 1.3851,
+      "step": 2264
+    },
+    {
+      "epoch": 0.17421736789477732,
+      "learning_rate": 0.002780883388433402,
+      "loss": 0.706,
+      "step": 2265
+    },
+    {
+      "epoch": 0.17429428505499578,
+      "learning_rate": 0.002780694724778669,
+      "loss": 1.2864,
+      "step": 2266
+    },
+    {
+      "epoch": 0.1743712022152142,
+      "learning_rate": 0.0027805059863428372,
+      "loss": 1.0764,
+      "step": 2267
+    },
+    {
+      "epoch": 0.17444811937543267,
+      "learning_rate": 0.0027803171731369263,
+      "loss": 1.3292,
+      "step": 2268
+    },
+    {
+      "epoch": 0.1745250365356511,
+      "learning_rate": 0.002780128285171961,
+      "loss": 1.0075,
+      "step": 2269
+    },
+    {
+      "epoch": 0.17460195369586956,
+      "learning_rate": 0.0027799393224589714,
+      "loss": 1.3501,
+      "step": 2270
+    },
+    {
+      "epoch": 0.174678870856088,
+      "learning_rate": 0.0027797502850089905,
+      "loss": 0.8623,
+      "step": 2271
+    },
+    {
+      "epoch": 0.17475578801630645,
+      "learning_rate": 0.0027795611728330567,
+      "loss": 0.828,
+      "step": 2272
+    },
+    {
+      "epoch": 0.17483270517652488,
+      "learning_rate": 0.002779371985942213,
+      "loss": 1.1315,
+      "step": 2273
+    },
+    {
+      "epoch": 0.17490962233674331,
+      "learning_rate": 0.002779182724347505,
+      "loss": 1.4622,
+      "step": 2274
+    },
+    {
+      "epoch": 0.17498653949696177,
+      "learning_rate": 0.0027789933880599848,
+      "loss": 0.8753,
+      "step": 2275
+    },
+    {
+      "epoch": 0.1750634566571802,
+      "learning_rate": 0.0027788039770907085,
+      "loss": 1.1484,
+      "step": 2276
+    },
+    {
+      "epoch": 0.17514037381739866,
+      "learning_rate": 0.0027786144914507347,
+      "loss": 1.1718,
+      "step": 2277
+    },
+    {
+      "epoch": 0.1752172909776171,
+      "learning_rate": 0.0027784249311511282,
+      "loss": 1.1838,
+      "step": 2278
+    },
+    {
+      "epoch": 0.17529420813783556,
+      "learning_rate": 0.002778235296202957,
+      "loss": 1.3006,
+      "step": 2279
+    },
+    {
+      "epoch": 0.175371125298054,
+      "learning_rate": 0.0027780455866172954,
+      "loss": 1.0545,
+      "step": 2280
+    },
+    {
+      "epoch": 0.17544804245827245,
+      "learning_rate": 0.00277785580240522,
+      "loss": 1.2178,
+      "step": 2281
+    },
+    {
+      "epoch": 0.17552495961849088,
+      "learning_rate": 0.002777665943577813,
+      "loss": 0.7611,
+      "step": 2282
+    },
+    {
+      "epoch": 0.17560187677870934,
+      "learning_rate": 0.002777476010146159,
+      "loss": 1.3487,
+      "step": 2283
+    },
+    {
+      "epoch": 0.17567879393892777,
+      "learning_rate": 0.00277728600212135,
+      "loss": 1.2143,
+      "step": 2284
+    },
+    {
+      "epoch": 0.17575571109914623,
+      "learning_rate": 0.00277709591951448,
+      "loss": 1.2719,
+      "step": 2285
+    },
+    {
+      "epoch": 0.17583262825936466,
+      "learning_rate": 0.0027769057623366483,
+      "loss": 1.5202,
+      "step": 2286
+    },
+    {
+      "epoch": 0.17590954541958312,
+      "learning_rate": 0.0027767155305989583,
+      "loss": 1.0031,
+      "step": 2287
+    },
+    {
+      "epoch": 0.17598646257980155,
+      "learning_rate": 0.0027765252243125184,
+      "loss": 1.3378,
+      "step": 2288
+    },
+    {
+      "epoch": 0.17606337974002,
+      "learning_rate": 0.0027763348434884398,
+      "loss": 1.3234,
+      "step": 2289
+    },
+    {
+      "epoch": 0.17614029690023844,
+      "learning_rate": 0.00277614438813784,
+      "loss": 1.1323,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1762172140604569,
+      "learning_rate": 0.0027759538582718393,
+      "loss": 1.294,
+      "step": 2291
+    },
+    {
+      "epoch": 0.17629413122067533,
+      "learning_rate": 0.0027757632539015626,
+      "loss": 1.2474,
+      "step": 2292
+    },
+    {
+      "epoch": 0.1763710483808938,
+      "learning_rate": 0.00277557257503814,
+      "loss": 0.7762,
+      "step": 2293
+    },
+    {
+      "epoch": 0.17644796554111222,
+      "learning_rate": 0.0027753818216927056,
+      "loss": 1.5522,
+      "step": 2294
+    },
+    {
+      "epoch": 0.17652488270133065,
+      "learning_rate": 0.0027751909938763973,
+      "loss": 1.964,
+      "step": 2295
+    },
+    {
+      "epoch": 0.1766017998615491,
+      "learning_rate": 0.002775000091600358,
+      "loss": 1.2278,
+      "step": 2296
+    },
+    {
+      "epoch": 0.17667871702176755,
+      "learning_rate": 0.002774809114875735,
+      "loss": 1.1071,
+      "step": 2297
+    },
+    {
+      "epoch": 0.176755634181986,
+      "learning_rate": 0.0027746180637136784,
+      "loss": 1.3481,
+      "step": 2298
+    },
+    {
+      "epoch": 0.17683255134220444,
+      "learning_rate": 0.002774426938125345,
+      "loss": 0.9673,
+      "step": 2299
+    },
+    {
+      "epoch": 0.1769094685024229,
+      "learning_rate": 0.0027742357381218947,
+      "loss": 0.9776,
+      "step": 2300
+    },
+    {
+      "epoch": 0.17698638566264133,
+      "learning_rate": 0.0027740444637144915,
+      "loss": 0.8895,
+      "step": 2301
+    },
+    {
+      "epoch": 0.17706330282285979,
+      "learning_rate": 0.0027738531149143044,
+      "loss": 1.5286,
+      "step": 2302
+    },
+    {
+      "epoch": 0.17714021998307822,
+      "learning_rate": 0.0027736616917325065,
+      "loss": 1.1316,
+      "step": 2303
+    },
+    {
+      "epoch": 0.17721713714329668,
+      "learning_rate": 0.002773470194180275,
+      "loss": 1.0178,
+      "step": 2304
+    },
+    {
+      "epoch": 0.1772940543035151,
+      "learning_rate": 0.0027732786222687913,
+      "loss": 1.1816,
+      "step": 2305
+    },
+    {
+      "epoch": 0.17737097146373357,
+      "learning_rate": 0.002773086976009242,
+      "loss": 1.172,
+      "step": 2306
+    },
+    {
+      "epoch": 0.177447888623952,
+      "learning_rate": 0.0027728952554128172,
+      "loss": 1.0449,
+      "step": 2307
+    },
+    {
+      "epoch": 0.17752480578417046,
+      "learning_rate": 0.0027727034604907125,
+      "loss": 0.8095,
+      "step": 2308
+    },
+    {
+      "epoch": 0.1776017229443889,
+      "learning_rate": 0.0027725115912541256,
+      "loss": 0.899,
+      "step": 2309
+    },
+    {
+      "epoch": 0.17767864010460735,
+      "learning_rate": 0.0027723196477142612,
+      "loss": 1.2357,
+      "step": 2310
+    },
+    {
+      "epoch": 0.17775555726482578,
+      "learning_rate": 0.0027721276298823265,
+      "loss": 1.2257,
+      "step": 2311
+    },
+    {
+      "epoch": 0.17783247442504424,
+      "learning_rate": 0.002771935537769534,
+      "loss": 1.3586,
+      "step": 2312
+    },
+    {
+      "epoch": 0.17790939158526267,
+      "learning_rate": 0.0027717433713870996,
+      "loss": 1.4614,
+      "step": 2313
+    },
+    {
+      "epoch": 0.1779863087454811,
+      "learning_rate": 0.0027715511307462443,
+      "loss": 1.0453,
+      "step": 2314
+    },
+    {
+      "epoch": 0.17806322590569956,
+      "learning_rate": 0.002771358815858193,
+      "loss": 1.1424,
+      "step": 2315
+    },
+    {
+      "epoch": 0.178140143065918,
+      "learning_rate": 0.002771166426734176,
+      "loss": 0.7497,
+      "step": 2316
+    },
+    {
+      "epoch": 0.17821706022613645,
+      "learning_rate": 0.0027709739633854265,
+      "loss": 1.0717,
+      "step": 2317
+    },
+    {
+      "epoch": 0.17829397738635488,
+      "learning_rate": 0.002770781425823182,
+      "loss": 1.2349,
+      "step": 2318
+    },
+    {
+      "epoch": 0.17837089454657334,
+      "learning_rate": 0.002770588814058687,
+      "loss": 1.1313,
+      "step": 2319
+    },
+    {
+      "epoch": 0.17844781170679178,
+      "learning_rate": 0.0027703961281031857,
+      "loss": 1.3433,
+      "step": 2320
+    },
+    {
+      "epoch": 0.17852472886701024,
+      "learning_rate": 0.0027702033679679315,
+      "loss": 1.1871,
+      "step": 2321
+    },
+    {
+      "epoch": 0.17860164602722867,
+      "learning_rate": 0.0027700105336641782,
+      "loss": 1.0887,
+      "step": 2322
+    },
+    {
+      "epoch": 0.17867856318744713,
+      "learning_rate": 0.0027698176252031865,
+      "loss": 1.3306,
+      "step": 2323
+    },
+    {
+      "epoch": 0.17875548034766556,
+      "learning_rate": 0.00276962464259622,
+      "loss": 1.1755,
+      "step": 2324
+    },
+    {
+      "epoch": 0.17883239750788402,
+      "learning_rate": 0.0027694315858545477,
+      "loss": 1.2721,
+      "step": 2325
+    },
+    {
+      "epoch": 0.17890931466810245,
+      "learning_rate": 0.002769238454989442,
+      "loss": 1.1679,
+      "step": 2326
+    },
+    {
+      "epoch": 0.1789862318283209,
+      "learning_rate": 0.0027690452500121806,
+      "loss": 1.0343,
+      "step": 2327
+    },
+    {
+      "epoch": 0.17906314898853934,
+      "learning_rate": 0.002768851970934044,
+      "loss": 0.8905,
+      "step": 2328
+    },
+    {
+      "epoch": 0.1791400661487578,
+      "learning_rate": 0.0027686586177663184,
+      "loss": 1.1715,
+      "step": 2329
+    },
+    {
+      "epoch": 0.17921698330897623,
+      "learning_rate": 0.0027684651905202943,
+      "loss": 1.1536,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1792939004691947,
+      "learning_rate": 0.0027682716892072656,
+      "loss": 1.1649,
+      "step": 2331
+    },
+    {
+      "epoch": 0.17937081762941312,
+      "learning_rate": 0.0027680781138385313,
+      "loss": 1.1529,
+      "step": 2332
+    },
+    {
+      "epoch": 0.17944773478963158,
+      "learning_rate": 0.0027678844644253944,
+      "loss": 0.8628,
+      "step": 2333
+    },
+    {
+      "epoch": 0.17952465194985,
+      "learning_rate": 0.002767690740979162,
+      "loss": 0.9465,
+      "step": 2334
+    },
+    {
+      "epoch": 0.17960156911006844,
+      "learning_rate": 0.0027674969435111463,
+      "loss": 1.1379,
+      "step": 2335
+    },
+    {
+      "epoch": 0.1796784862702869,
+      "learning_rate": 0.002767303072032663,
+      "loss": 1.0461,
+      "step": 2336
+    },
+    {
+      "epoch": 0.17975540343050533,
+      "learning_rate": 0.0027671091265550323,
+      "loss": 1.126,
+      "step": 2337
+    },
+    {
+      "epoch": 0.1798323205907238,
+      "learning_rate": 0.00276691510708958,
+      "loss": 1.2887,
+      "step": 2338
+    },
+    {
+      "epoch": 0.17990923775094222,
+      "learning_rate": 0.0027667210136476336,
+      "loss": 0.9879,
+      "step": 2339
+    },
+    {
+      "epoch": 0.17998615491116068,
+      "learning_rate": 0.002766526846240527,
+      "loss": 1.0731,
+      "step": 2340
+    },
+    {
+      "epoch": 0.18006307207137912,
+      "learning_rate": 0.002766332604879598,
+      "loss": 1.0854,
+      "step": 2341
+    },
+    {
+      "epoch": 0.18013998923159757,
+      "learning_rate": 0.0027661382895761886,
+      "loss": 1.3078,
+      "step": 2342
+    },
+    {
+      "epoch": 0.180216906391816,
+      "learning_rate": 0.0027659439003416453,
+      "loss": 1.3538,
+      "step": 2343
+    },
+    {
+      "epoch": 0.18029382355203447,
+      "learning_rate": 0.0027657494371873172,
+      "loss": 1.092,
+      "step": 2344
+    },
+    {
+      "epoch": 0.1803707407122529,
+      "learning_rate": 0.002765554900124561,
+      "loss": 1.2404,
+      "step": 2345
+    },
+    {
+      "epoch": 0.18044765787247136,
+      "learning_rate": 0.0027653602891647357,
+      "loss": 1.1784,
+      "step": 2346
+    },
+    {
+      "epoch": 0.1805245750326898,
+      "learning_rate": 0.0027651656043192036,
+      "loss": 1.2711,
+      "step": 2347
+    },
+    {
+      "epoch": 0.18060149219290825,
+      "learning_rate": 0.002764970845599334,
+      "loss": 0.9122,
+      "step": 2348
+    },
+    {
+      "epoch": 0.18067840935312668,
+      "learning_rate": 0.002764776013016498,
+      "loss": 1.0016,
+      "step": 2349
+    },
+    {
+      "epoch": 0.18075532651334514,
+      "learning_rate": 0.0027645811065820725,
+      "loss": 1.0989,
+      "step": 2350
+    },
+    {
+      "epoch": 0.18083224367356357,
+      "learning_rate": 0.0027643861263074387,
+      "loss": 1.3739,
+      "step": 2351
+    },
+    {
+      "epoch": 0.18090916083378203,
+      "learning_rate": 0.002764191072203981,
+      "loss": 1.2372,
+      "step": 2352
+    },
+    {
+      "epoch": 0.18098607799400046,
+      "learning_rate": 0.002763995944283089,
+      "loss": 0.8442,
+      "step": 2353
+    },
+    {
+      "epoch": 0.18106299515421892,
+      "learning_rate": 0.0027638007425561566,
+      "loss": 0.8965,
+      "step": 2354
+    },
+    {
+      "epoch": 0.18113991231443735,
+      "learning_rate": 0.0027636054670345813,
+      "loss": 0.8469,
+      "step": 2355
+    },
+    {
+      "epoch": 0.18121682947465578,
+      "learning_rate": 0.002763410117729767,
+      "loss": 1.1008,
+      "step": 2356
+    },
+    {
+      "epoch": 0.18129374663487424,
+      "learning_rate": 0.002763214694653118,
+      "loss": 1.0531,
+      "step": 2357
+    },
+    {
+      "epoch": 0.18137066379509267,
+      "learning_rate": 0.002763019197816047,
+      "loss": 0.9141,
+      "step": 2358
+    },
+    {
+      "epoch": 0.18144758095531113,
+      "learning_rate": 0.0027628236272299687,
+      "loss": 0.9273,
+      "step": 2359
+    },
+    {
+      "epoch": 0.18152449811552956,
+      "learning_rate": 0.002762627982906303,
+      "loss": 1.2542,
+      "step": 2360
+    },
+    {
+      "epoch": 0.18160141527574802,
+      "learning_rate": 0.0027624322648564727,
+      "loss": 1.2319,
+      "step": 2361
+    },
+    {
+      "epoch": 0.18167833243596646,
+      "learning_rate": 0.0027622364730919076,
+      "loss": 0.8684,
+      "step": 2362
+    },
+    {
+      "epoch": 0.18175524959618491,
+      "learning_rate": 0.002762040607624039,
+      "loss": 1.0326,
+      "step": 2363
+    },
+    {
+      "epoch": 0.18183216675640335,
+      "learning_rate": 0.0027618446684643046,
+      "loss": 1.1698,
+      "step": 2364
+    },
+    {
+      "epoch": 0.1819090839166218,
+      "learning_rate": 0.002761648655624144,
+      "loss": 1.4721,
+      "step": 2365
+    },
+    {
+      "epoch": 0.18198600107684024,
+      "learning_rate": 0.002761452569115004,
+      "loss": 1.4152,
+      "step": 2366
+    },
+    {
+      "epoch": 0.1820629182370587,
+      "learning_rate": 0.002761256408948334,
+      "loss": 1.2847,
+      "step": 2367
+    },
+    {
+      "epoch": 0.18213983539727713,
+      "learning_rate": 0.002761060175135588,
+      "loss": 1.0758,
+      "step": 2368
+    },
+    {
+      "epoch": 0.1822167525574956,
+      "learning_rate": 0.002760863867688224,
+      "loss": 1.4763,
+      "step": 2369
+    },
+    {
+      "epoch": 0.18229366971771402,
+      "learning_rate": 0.002760667486617704,
+      "loss": 0.9241,
+      "step": 2370
+    },
+    {
+      "epoch": 0.18237058687793248,
+      "learning_rate": 0.002760471031935497,
+      "loss": 1.1661,
+      "step": 2371
+    },
+    {
+      "epoch": 0.1824475040381509,
+      "learning_rate": 0.0027602745036530716,
+      "loss": 0.9507,
+      "step": 2372
+    },
+    {
+      "epoch": 0.18252442119836937,
+      "learning_rate": 0.002760077901781905,
+      "loss": 1.3158,
+      "step": 2373
+    },
+    {
+      "epoch": 0.1826013383585878,
+      "learning_rate": 0.0027598812263334763,
+      "loss": 1.2181,
+      "step": 2374
+    },
+    {
+      "epoch": 0.18267825551880623,
+      "learning_rate": 0.0027596844773192697,
+      "loss": 1.3014,
+      "step": 2375
+    },
+    {
+      "epoch": 0.1827551726790247,
+      "learning_rate": 0.0027594876547507737,
+      "loss": 1.0981,
+      "step": 2376
+    },
+    {
+      "epoch": 0.18283208983924312,
+      "learning_rate": 0.0027592907586394813,
+      "loss": 1.2466,
+      "step": 2377
+    },
+    {
+      "epoch": 0.18290900699946158,
+      "learning_rate": 0.002759093788996889,
+      "loss": 1.2532,
+      "step": 2378
+    },
+    {
+      "epoch": 0.18298592415968,
+      "learning_rate": 0.0027588967458344977,
+      "loss": 1.1043,
+      "step": 2379
+    },
+    {
+      "epoch": 0.18306284131989847,
+      "learning_rate": 0.002758699629163814,
+      "loss": 1.0542,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1831397584801169,
+      "learning_rate": 0.002758502438996347,
+      "loss": 1.0792,
+      "step": 2381
+    },
+    {
+      "epoch": 0.18321667564033536,
+      "learning_rate": 0.0027583051753436107,
+      "loss": 1.3526,
+      "step": 2382
+    },
+    {
+      "epoch": 0.1832935928005538,
+      "learning_rate": 0.0027581078382171243,
+      "loss": 0.8125,
+      "step": 2383
+    },
+    {
+      "epoch": 0.18337050996077225,
+      "learning_rate": 0.0027579104276284094,
+      "loss": 1.2082,
+      "step": 2384
+    },
+    {
+      "epoch": 0.18344742712099069,
+      "learning_rate": 0.002757712943588994,
+      "loss": 1.0168,
+      "step": 2385
+    },
+    {
+      "epoch": 0.18352434428120915,
+      "learning_rate": 0.002757515386110409,
+      "loss": 1.36,
+      "step": 2386
+    },
+    {
+      "epoch": 0.18360126144142758,
+      "learning_rate": 0.002757317755204191,
+      "loss": 1.3223,
+      "step": 2387
+    },
+    {
+      "epoch": 0.18367817860164604,
+      "learning_rate": 0.0027571200508818778,
+      "loss": 1.367,
+      "step": 2388
+    },
+    {
+      "epoch": 0.18375509576186447,
+      "learning_rate": 0.0027569222731550153,
+      "loss": 1.4544,
+      "step": 2389
+    },
+    {
+      "epoch": 0.18383201292208293,
+      "learning_rate": 0.0027567244220351506,
+      "loss": 1.2253,
+      "step": 2390
+    },
+    {
+      "epoch": 0.18390893008230136,
+      "learning_rate": 0.002756526497533838,
+      "loss": 1.4035,
+      "step": 2391
+    },
+    {
+      "epoch": 0.18398584724251982,
+      "learning_rate": 0.002756328499662633,
+      "loss": 0.9734,
+      "step": 2392
+    },
+    {
+      "epoch": 0.18406276440273825,
+      "learning_rate": 0.0027561304284330983,
+      "loss": 1.2042,
+      "step": 2393
+    },
+    {
+      "epoch": 0.1841396815629567,
+      "learning_rate": 0.0027559322838567983,
+      "loss": 1.3424,
+      "step": 2394
+    },
+    {
+      "epoch": 0.18421659872317514,
+      "learning_rate": 0.002755734065945303,
+      "loss": 0.8253,
+      "step": 2395
+    },
+    {
+      "epoch": 0.18429351588339357,
+      "learning_rate": 0.0027555357747101874,
+      "loss": 1.2893,
+      "step": 2396
+    },
+    {
+      "epoch": 0.18437043304361203,
+      "learning_rate": 0.00275533741016303,
+      "loss": 1.082,
+      "step": 2397
+    },
+    {
+      "epoch": 0.18444735020383046,
+      "learning_rate": 0.002755138972315412,
+      "loss": 1.2415,
+      "step": 2398
+    },
+    {
+      "epoch": 0.18452426736404892,
+      "learning_rate": 0.002754940461178922,
+      "loss": 1.0947,
+      "step": 2399
+    },
+    {
+      "epoch": 0.18460118452426735,
+      "learning_rate": 0.00275474187676515,
+      "loss": 1.0091,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1846781016844858,
+      "learning_rate": 0.0027545432190856922,
+      "loss": 1.1793,
+      "step": 2401
+    },
+    {
+      "epoch": 0.18475501884470424,
+      "learning_rate": 0.0027543444881521486,
+      "loss": 1.3007,
+      "step": 2402
+    },
+    {
+      "epoch": 0.1848319360049227,
+      "learning_rate": 0.002754145683976123,
+      "loss": 1.3381,
+      "step": 2403
+    },
+    {
+      "epoch": 0.18490885316514113,
+      "learning_rate": 0.002753946806569224,
+      "loss": 1.0557,
+      "step": 2404
+    },
+    {
+      "epoch": 0.1849857703253596,
+      "learning_rate": 0.002753747855943064,
+      "loss": 1.3429,
+      "step": 2405
+    },
+    {
+      "epoch": 0.18506268748557803,
+      "learning_rate": 0.00275354883210926,
+      "loss": 0.9561,
+      "step": 2406
+    },
+    {
+      "epoch": 0.18513960464579649,
+      "learning_rate": 0.0027533497350794335,
+      "loss": 1.0618,
+      "step": 2407
+    },
+    {
+      "epoch": 0.18521652180601492,
+      "learning_rate": 0.0027531505648652095,
+      "loss": 1.359,
+      "step": 2408
+    },
+    {
+      "epoch": 0.18529343896623338,
+      "learning_rate": 0.0027529513214782186,
+      "loss": 1.0228,
+      "step": 2409
+    },
+    {
+      "epoch": 0.1853703561264518,
+      "learning_rate": 0.0027527520049300935,
+      "loss": 1.1886,
+      "step": 2410
+    },
+    {
+      "epoch": 0.18544727328667027,
+      "learning_rate": 0.0027525526152324736,
+      "loss": 1.165,
+      "step": 2411
+    },
+    {
+      "epoch": 0.1855241904468887,
+      "learning_rate": 0.002752353152397001,
+      "loss": 0.9727,
+      "step": 2412
+    },
+    {
+      "epoch": 0.18560110760710716,
+      "learning_rate": 0.0027521536164353227,
+      "loss": 0.8979,
+      "step": 2413
+    },
+    {
+      "epoch": 0.1856780247673256,
+      "learning_rate": 0.00275195400735909,
+      "loss": 0.971,
+      "step": 2414
+    },
+    {
+      "epoch": 0.18575494192754405,
+      "learning_rate": 0.002751754325179958,
+      "loss": 1.3337,
+      "step": 2415
+    },
+    {
+      "epoch": 0.18583185908776248,
+      "learning_rate": 0.0027515545699095863,
+      "loss": 0.9449,
+      "step": 2416
+    },
+    {
+      "epoch": 0.1859087762479809,
+      "learning_rate": 0.002751354741559639,
+      "loss": 1.4593,
+      "step": 2417
+    },
+    {
+      "epoch": 0.18598569340819937,
+      "learning_rate": 0.002751154840141784,
+      "loss": 1.0102,
+      "step": 2418
+    },
+    {
+      "epoch": 0.1860626105684178,
+      "learning_rate": 0.002750954865667694,
+      "loss": 1.1943,
+      "step": 2419
+    },
+    {
+      "epoch": 0.18613952772863626,
+      "learning_rate": 0.0027507548181490457,
+      "loss": 1.2264,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1862164448888547,
+      "learning_rate": 0.0027505546975975202,
+      "loss": 1.3027,
+      "step": 2421
+    },
+    {
+      "epoch": 0.18629336204907315,
+      "learning_rate": 0.0027503545040248024,
+      "loss": 1.3922,
+      "step": 2422
+    },
+    {
+      "epoch": 0.18637027920929158,
+      "learning_rate": 0.002750154237442582,
+      "loss": 1.252,
+      "step": 2423
+    },
+    {
+      "epoch": 0.18644719636951004,
+      "learning_rate": 0.0027499538978625533,
+      "loss": 1.1408,
+      "step": 2424
+    },
+    {
+      "epoch": 0.18652411352972847,
+      "learning_rate": 0.0027497534852964135,
+      "loss": 1.3712,
+      "step": 2425
+    },
+    {
+      "epoch": 0.18660103068994693,
+      "learning_rate": 0.002749552999755865,
+      "loss": 1.067,
+      "step": 2426
+    },
+    {
+      "epoch": 0.18667794785016537,
+      "learning_rate": 0.002749352441252615,
+      "loss": 1.0006,
+      "step": 2427
+    },
+    {
+      "epoch": 0.18675486501038382,
+      "learning_rate": 0.0027491518097983733,
+      "loss": 1.1334,
+      "step": 2428
+    },
+    {
+      "epoch": 0.18683178217060226,
+      "learning_rate": 0.002748951105404856,
+      "loss": 0.9525,
+      "step": 2429
+    },
+    {
+      "epoch": 0.18690869933082072,
+      "learning_rate": 0.0027487503280837823,
+      "loss": 0.7266,
+      "step": 2430
+    },
+    {
+      "epoch": 0.18698561649103915,
+      "learning_rate": 0.0027485494778468752,
+      "loss": 1.1159,
+      "step": 2431
+    },
+    {
+      "epoch": 0.1870625336512576,
+      "learning_rate": 0.002748348554705863,
+      "loss": 1.356,
+      "step": 2432
+    },
+    {
+      "epoch": 0.18713945081147604,
+      "learning_rate": 0.0027481475586724775,
+      "loss": 1.0046,
+      "step": 2433
+    },
+    {
+      "epoch": 0.1872163679716945,
+      "learning_rate": 0.0027479464897584554,
+      "loss": 1.0578,
+      "step": 2434
+    },
+    {
+      "epoch": 0.18729328513191293,
+      "learning_rate": 0.002747745347975537,
+      "loss": 0.8953,
+      "step": 2435
+    },
+    {
+      "epoch": 0.1873702022921314,
+      "learning_rate": 0.002747544133335468,
+      "loss": 1.1441,
+      "step": 2436
+    },
+    {
+      "epoch": 0.18744711945234982,
+      "learning_rate": 0.0027473428458499967,
+      "loss": 1.2491,
+      "step": 2437
+    },
+    {
+      "epoch": 0.18752403661256825,
+      "learning_rate": 0.002747141485530876,
+      "loss": 1.2872,
+      "step": 2438
+    },
+    {
+      "epoch": 0.1876009537727867,
+      "learning_rate": 0.002746940052389865,
+      "loss": 1.0918,
+      "step": 2439
+    },
+    {
+      "epoch": 0.18767787093300514,
+      "learning_rate": 0.0027467385464387244,
+      "loss": 1.0533,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1877547880932236,
+      "learning_rate": 0.0027465369676892204,
+      "loss": 1.1003,
+      "step": 2441
+    },
+    {
+      "epoch": 0.18783170525344203,
+      "learning_rate": 0.0027463353161531244,
+      "loss": 1.1003,
+      "step": 2442
+    },
+    {
+      "epoch": 0.1879086224136605,
+      "learning_rate": 0.00274613359184221,
+      "loss": 1.1075,
+      "step": 2443
+    },
+    {
+      "epoch": 0.18798553957387892,
+      "learning_rate": 0.002745931794768257,
+      "loss": 0.9396,
+      "step": 2444
+    },
+    {
+      "epoch": 0.18806245673409738,
+      "learning_rate": 0.0027457299249430476,
+      "loss": 1.3618,
+      "step": 2445
+    },
+    {
+      "epoch": 0.18813937389431581,
+      "learning_rate": 0.0027455279823783694,
+      "loss": 1.0066,
+      "step": 2446
+    },
+    {
+      "epoch": 0.18821629105453427,
+      "learning_rate": 0.0027453259670860144,
+      "loss": 1.2253,
+      "step": 2447
+    },
+    {
+      "epoch": 0.1882932082147527,
+      "learning_rate": 0.002745123879077779,
+      "loss": 1.2503,
+      "step": 2448
+    },
+    {
+      "epoch": 0.18837012537497116,
+      "learning_rate": 0.0027449217183654617,
+      "loss": 0.852,
+      "step": 2449
+    },
+    {
+      "epoch": 0.1884470425351896,
+      "learning_rate": 0.0027447194849608684,
+      "loss": 1.2111,
+      "step": 2450
+    },
+    {
+      "epoch": 0.18852395969540806,
+      "learning_rate": 0.0027445171788758067,
+      "loss": 0.8599,
+      "step": 2451
+    },
+    {
+      "epoch": 0.1886008768556265,
+      "learning_rate": 0.0027443148001220904,
+      "loss": 1.1196,
+      "step": 2452
+    },
+    {
+      "epoch": 0.18867779401584495,
+      "learning_rate": 0.002744112348711536,
+      "loss": 0.9139,
+      "step": 2453
+    },
+    {
+      "epoch": 0.18875471117606338,
+      "learning_rate": 0.002743909824655965,
+      "loss": 1.205,
+      "step": 2454
+    },
+    {
+      "epoch": 0.18883162833628184,
+      "learning_rate": 0.002743707227967203,
+      "loss": 1.1818,
+      "step": 2455
+    },
+    {
+      "epoch": 0.18890854549650027,
+      "learning_rate": 0.0027435045586570793,
+      "loss": 1.8483,
+      "step": 2456
+    },
+    {
+      "epoch": 0.1889854626567187,
+      "learning_rate": 0.002743301816737429,
+      "loss": 0.9127,
+      "step": 2457
+    },
+    {
+      "epoch": 0.18906237981693716,
+      "learning_rate": 0.0027430990022200897,
+      "loss": 1.0292,
+      "step": 2458
+    },
+    {
+      "epoch": 0.1891392969771556,
+      "learning_rate": 0.0027428961151169046,
+      "loss": 0.9004,
+      "step": 2459
+    },
+    {
+      "epoch": 0.18921621413737405,
+      "learning_rate": 0.0027426931554397193,
+      "loss": 1.0631,
+      "step": 2460
+    },
+    {
+      "epoch": 0.18929313129759248,
+      "learning_rate": 0.002742490123200386,
+      "loss": 1.1435,
+      "step": 2461
+    },
+    {
+      "epoch": 0.18937004845781094,
+      "learning_rate": 0.0027422870184107594,
+      "loss": 1.2402,
+      "step": 2462
+    },
+    {
+      "epoch": 0.18944696561802937,
+      "learning_rate": 0.002742083841082699,
+      "loss": 1.1791,
+      "step": 2463
+    },
+    {
+      "epoch": 0.18952388277824783,
+      "learning_rate": 0.002741880591228069,
+      "loss": 1.0982,
+      "step": 2464
+    },
+    {
+      "epoch": 0.18960079993846626,
+      "learning_rate": 0.0027416772688587366,
+      "loss": 1.0543,
+      "step": 2465
+    },
+    {
+      "epoch": 0.18967771709868472,
+      "learning_rate": 0.0027414738739865744,
+      "loss": 1.6758,
+      "step": 2466
+    },
+    {
+      "epoch": 0.18975463425890315,
+      "learning_rate": 0.0027412704066234594,
+      "loss": 1.0502,
+      "step": 2467
+    },
+    {
+      "epoch": 0.1898315514191216,
+      "learning_rate": 0.0027410668667812712,
+      "loss": 1.0975,
+      "step": 2468
+    },
+    {
+      "epoch": 0.18990846857934005,
+      "learning_rate": 0.0027408632544718954,
+      "loss": 1.3923,
+      "step": 2469
+    },
+    {
+      "epoch": 0.1899853857395585,
+      "learning_rate": 0.002740659569707221,
+      "loss": 1.1496,
+      "step": 2470
+    },
+    {
+      "epoch": 0.19006230289977694,
+      "learning_rate": 0.0027404558124991415,
+      "loss": 1.1202,
+      "step": 2471
+    },
+    {
+      "epoch": 0.1901392200599954,
+      "learning_rate": 0.0027402519828595537,
+      "loss": 1.1255,
+      "step": 2472
+    },
+    {
+      "epoch": 0.19021613722021383,
+      "learning_rate": 0.002740048080800361,
+      "loss": 0.8023,
+      "step": 2473
+    },
+    {
+      "epoch": 0.19029305438043229,
+      "learning_rate": 0.002739844106333468,
+      "loss": 1.0702,
+      "step": 2474
+    },
+    {
+      "epoch": 0.19036997154065072,
+      "learning_rate": 0.0027396400594707853,
+      "loss": 1.4331,
+      "step": 2475
+    },
+    {
+      "epoch": 0.19044688870086918,
+      "learning_rate": 0.002739435940224228,
+      "loss": 0.8715,
+      "step": 2476
+    },
+    {
+      "epoch": 0.1905238058610876,
+      "learning_rate": 0.0027392317486057147,
+      "loss": 1.0336,
+      "step": 2477
+    },
+    {
+      "epoch": 0.19060072302130604,
+      "learning_rate": 0.0027390274846271676,
+      "loss": 1.3031,
+      "step": 2478
+    },
+    {
+      "epoch": 0.1906776401815245,
+      "learning_rate": 0.0027388231483005143,
+      "loss": 1.8521,
+      "step": 2479
+    },
+    {
+      "epoch": 0.19075455734174293,
+      "learning_rate": 0.0027386187396376866,
+      "loss": 1.0224,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1908314745019614,
+      "learning_rate": 0.00273841425865062,
+      "loss": 1.1065,
+      "step": 2481
+    },
+    {
+      "epoch": 0.19090839166217982,
+      "learning_rate": 0.0027382097053512537,
+      "loss": 1.4644,
+      "step": 2482
+    },
+    {
+      "epoch": 0.19098530882239828,
+      "learning_rate": 0.002738005079751532,
+      "loss": 1.2938,
+      "step": 2483
+    },
+    {
+      "epoch": 0.1910622259826167,
+      "learning_rate": 0.0027378003818634045,
+      "loss": 1.2478,
+      "step": 2484
+    },
+    {
+      "epoch": 0.19113914314283517,
+      "learning_rate": 0.002737595611698822,
+      "loss": 1.1379,
+      "step": 2485
+    },
+    {
+      "epoch": 0.1912160603030536,
+      "learning_rate": 0.0027373907692697422,
+      "loss": 1.4759,
+      "step": 2486
+    },
+    {
+      "epoch": 0.19129297746327206,
+      "learning_rate": 0.0027371858545881256,
+      "loss": 0.9392,
+      "step": 2487
+    },
+    {
+      "epoch": 0.1913698946234905,
+      "learning_rate": 0.0027369808676659378,
+      "loss": 1.0521,
+      "step": 2488
+    },
+    {
+      "epoch": 0.19144681178370895,
+      "learning_rate": 0.002736775808515148,
+      "loss": 1.143,
+      "step": 2489
+    },
+    {
+      "epoch": 0.19152372894392738,
+      "learning_rate": 0.0027365706771477296,
+      "loss": 1.0759,
+      "step": 2490
+    },
+    {
+      "epoch": 0.19160064610414584,
+      "learning_rate": 0.00273636547357566,
+      "loss": 1.4833,
+      "step": 2491
+    },
+    {
+      "epoch": 0.19167756326436428,
+      "learning_rate": 0.002736160197810923,
+      "loss": 1.121,
+      "step": 2492
+    },
+    {
+      "epoch": 0.19175448042458274,
+      "learning_rate": 0.0027359548498655036,
+      "loss": 0.9887,
+      "step": 2493
+    },
+    {
+      "epoch": 0.19183139758480117,
+      "learning_rate": 0.002735749429751392,
+      "loss": 1.022,
+      "step": 2494
+    },
+    {
+      "epoch": 0.19190831474501963,
+      "learning_rate": 0.0027355439374805836,
+      "loss": 1.4389,
+      "step": 2495
+    },
+    {
+      "epoch": 0.19198523190523806,
+      "learning_rate": 0.002735338373065077,
+      "loss": 0.9638,
+      "step": 2496
+    },
+    {
+      "epoch": 0.19206214906545652,
+      "learning_rate": 0.002735132736516875,
+      "loss": 1.2374,
+      "step": 2497
+    },
+    {
+      "epoch": 0.19213906622567495,
+      "learning_rate": 0.002734927027847986,
+      "loss": 1.0329,
+      "step": 2498
+    },
+    {
+      "epoch": 0.19221598338589338,
+      "learning_rate": 0.0027347212470704204,
+      "loss": 0.9132,
+      "step": 2499
+    },
+    {
+      "epoch": 0.19229290054611184,
+      "learning_rate": 0.0027345153941961946,
+      "loss": 1.1657,
+      "step": 2500
+    },
+    {
+      "epoch": 0.19236981770633027,
+      "learning_rate": 0.002734309469237328,
+      "loss": 1.459,
+      "step": 2501
+    },
+    {
+      "epoch": 0.19244673486654873,
+      "learning_rate": 0.0027341034722058455,
+      "loss": 1.1164,
+      "step": 2502
+    },
+    {
+      "epoch": 0.19252365202676716,
+      "learning_rate": 0.0027338974031137745,
+      "loss": 1.3067,
+      "step": 2503
+    },
+    {
+      "epoch": 0.19260056918698562,
+      "learning_rate": 0.002733691261973149,
+      "loss": 1.2746,
+      "step": 2504
+    },
+    {
+      "epoch": 0.19267748634720405,
+      "learning_rate": 0.0027334850487960045,
+      "loss": 0.8684,
+      "step": 2505
+    },
+    {
+      "epoch": 0.1927544035074225,
+      "learning_rate": 0.0027332787635943826,
+      "loss": 1.1647,
+      "step": 2506
+    },
+    {
+      "epoch": 0.19283132066764094,
+      "learning_rate": 0.0027330724063803287,
+      "loss": 1.6763,
+      "step": 2507
+    },
+    {
+      "epoch": 0.1929082378278594,
+      "learning_rate": 0.0027328659771658913,
+      "loss": 0.8732,
+      "step": 2508
+    },
+    {
+      "epoch": 0.19298515498807783,
+      "learning_rate": 0.0027326594759631254,
+      "loss": 1.2196,
+      "step": 2509
+    },
+    {
+      "epoch": 0.1930620721482963,
+      "learning_rate": 0.0027324529027840873,
+      "loss": 1.0564,
+      "step": 2510
+    },
+    {
+      "epoch": 0.19313898930851472,
+      "learning_rate": 0.00273224625764084,
+      "loss": 0.8968,
+      "step": 2511
+    },
+    {
+      "epoch": 0.19321590646873318,
+      "learning_rate": 0.0027320395405454495,
+      "loss": 1.2223,
+      "step": 2512
+    },
+    {
+      "epoch": 0.19329282362895162,
+      "learning_rate": 0.002731832751509986,
+      "loss": 0.7521,
+      "step": 2513
+    },
+    {
+      "epoch": 0.19336974078917007,
+      "learning_rate": 0.002731625890546525,
+      "loss": 1.1269,
+      "step": 2514
+    },
+    {
+      "epoch": 0.1934466579493885,
+      "learning_rate": 0.0027314189576671446,
+      "loss": 1.0745,
+      "step": 2515
+    },
+    {
+      "epoch": 0.19352357510960697,
+      "learning_rate": 0.0027312119528839275,
+      "loss": 1.1512,
+      "step": 2516
+    },
+    {
+      "epoch": 0.1936004922698254,
+      "learning_rate": 0.0027310048762089617,
+      "loss": 1.1137,
+      "step": 2517
+    },
+    {
+      "epoch": 0.19367740943004386,
+      "learning_rate": 0.002730797727654338,
+      "loss": 1.2492,
+      "step": 2518
+    },
+    {
+      "epoch": 0.1937543265902623,
+      "learning_rate": 0.002730590507232152,
+      "loss": 1.2935,
+      "step": 2519
+    },
+    {
+      "epoch": 0.19383124375048072,
+      "learning_rate": 0.002730383214954504,
+      "loss": 1.3263,
+      "step": 2520
+    },
+    {
+      "epoch": 0.19390816091069918,
+      "learning_rate": 0.002730175850833498,
+      "loss": 1.1364,
+      "step": 2521
+    },
+    {
+      "epoch": 0.1939850780709176,
+      "learning_rate": 0.002729968414881242,
+      "loss": 1.0328,
+      "step": 2522
+    },
+    {
+      "epoch": 0.19406199523113607,
+      "learning_rate": 0.0027297609071098485,
+      "loss": 1.3088,
+      "step": 2523
+    },
+    {
+      "epoch": 0.1941389123913545,
+      "learning_rate": 0.002729553327531434,
+      "loss": 1.1823,
+      "step": 2524
+    },
+    {
+      "epoch": 0.19421582955157296,
+      "learning_rate": 0.002729345676158119,
+      "loss": 1.1516,
+      "step": 2525
+    },
+    {
+      "epoch": 0.1942927467117914,
+      "learning_rate": 0.0027291379530020293,
+      "loss": 1.0633,
+      "step": 2526
+    },
+    {
+      "epoch": 0.19436966387200985,
+      "learning_rate": 0.0027289301580752925,
+      "loss": 1.0713,
+      "step": 2527
+    },
+    {
+      "epoch": 0.19444658103222828,
+      "learning_rate": 0.002728722291390044,
+      "loss": 1.4266,
+      "step": 2528
+    },
+    {
+      "epoch": 0.19452349819244674,
+      "learning_rate": 0.00272851435295842,
+      "loss": 0.9635,
+      "step": 2529
+    },
+    {
+      "epoch": 0.19460041535266517,
+      "learning_rate": 0.0027283063427925624,
+      "loss": 1.653,
+      "step": 2530
+    },
+    {
+      "epoch": 0.19467733251288363,
+      "learning_rate": 0.0027280982609046178,
+      "loss": 0.9572,
+      "step": 2531
+    },
+    {
+      "epoch": 0.19475424967310206,
+      "learning_rate": 0.0027278901073067353,
+      "loss": 1.4323,
+      "step": 2532
+    },
+    {
+      "epoch": 0.19483116683332052,
+      "learning_rate": 0.00272768188201107,
+      "loss": 1.4142,
+      "step": 2533
+    },
+    {
+      "epoch": 0.19490808399353896,
+      "learning_rate": 0.00272747358502978,
+      "loss": 1.173,
+      "step": 2534
+    },
+    {
+      "epoch": 0.19498500115375741,
+      "learning_rate": 0.002727265216375029,
+      "loss": 1.0115,
+      "step": 2535
+    },
+    {
+      "epoch": 0.19506191831397585,
+      "learning_rate": 0.0027270567760589825,
+      "loss": 1.2721,
+      "step": 2536
+    },
+    {
+      "epoch": 0.1951388354741943,
+      "learning_rate": 0.0027268482640938115,
+      "loss": 1.1139,
+      "step": 2537
+    },
+    {
+      "epoch": 0.19521575263441274,
+      "learning_rate": 0.0027266396804916924,
+      "loss": 1.286,
+      "step": 2538
+    },
+    {
+      "epoch": 0.19529266979463117,
+      "learning_rate": 0.0027264310252648034,
+      "loss": 0.7293,
+      "step": 2539
+    },
+    {
+      "epoch": 0.19536958695484963,
+      "learning_rate": 0.002726222298425329,
+      "loss": 1.8487,
+      "step": 2540
+    },
+    {
+      "epoch": 0.19544650411506806,
+      "learning_rate": 0.002726013499985457,
+      "loss": 1.2382,
+      "step": 2541
+    },
+    {
+      "epoch": 0.19552342127528652,
+      "learning_rate": 0.0027258046299573784,
+      "loss": 1.0999,
+      "step": 2542
+    },
+    {
+      "epoch": 0.19560033843550495,
+      "learning_rate": 0.0027255956883532905,
+      "loss": 1.1137,
+      "step": 2543
+    },
+    {
+      "epoch": 0.1956772555957234,
+      "learning_rate": 0.0027253866751853926,
+      "loss": 0.9086,
+      "step": 2544
+    },
+    {
+      "epoch": 0.19575417275594184,
+      "learning_rate": 0.00272517759046589,
+      "loss": 1.0784,
+      "step": 2545
+    },
+    {
+      "epoch": 0.1958310899161603,
+      "learning_rate": 0.002724968434206991,
+      "loss": 1.2894,
+      "step": 2546
+    },
+    {
+      "epoch": 0.19590800707637873,
+      "learning_rate": 0.0027247592064209084,
+      "loss": 1.22,
+      "step": 2547
+    },
+    {
+      "epoch": 0.1959849242365972,
+      "learning_rate": 0.002724549907119859,
+      "loss": 0.5908,
+      "step": 2548
+    },
+    {
+      "epoch": 0.19606184139681562,
+      "learning_rate": 0.002724340536316065,
+      "loss": 0.9553,
+      "step": 2549
+    },
+    {
+      "epoch": 0.19613875855703408,
+      "learning_rate": 0.0027241310940217507,
+      "loss": 1.5002,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1962156757172525,
+      "learning_rate": 0.0027239215802491463,
+      "loss": 1.4841,
+      "step": 2551
+    },
+    {
+      "epoch": 0.19629259287747097,
+      "learning_rate": 0.0027237119950104852,
+      "loss": 1.0263,
+      "step": 2552
+    },
+    {
+      "epoch": 0.1963695100376894,
+      "learning_rate": 0.002723502338318005,
+      "loss": 0.772,
+      "step": 2553
+    },
+    {
+      "epoch": 0.19644642719790786,
+      "learning_rate": 0.002723292610183949,
+      "loss": 1.1455,
+      "step": 2554
+    },
+    {
+      "epoch": 0.1965233443581263,
+      "learning_rate": 0.002723082810620562,
+      "loss": 1.0296,
+      "step": 2555
+    },
+    {
+      "epoch": 0.19660026151834475,
+      "learning_rate": 0.002722872939640095,
+      "loss": 1.2724,
+      "step": 2556
+    },
+    {
+      "epoch": 0.19667717867856319,
+      "learning_rate": 0.002722662997254803,
+      "loss": 1.2582,
+      "step": 2557
+    },
+    {
+      "epoch": 0.19675409583878165,
+      "learning_rate": 0.0027224529834769438,
+      "loss": 1.3994,
+      "step": 2558
+    },
+    {
+      "epoch": 0.19683101299900008,
+      "learning_rate": 0.002722242898318781,
+      "loss": 1.1654,
+      "step": 2559
+    },
+    {
+      "epoch": 0.1969079301592185,
+      "learning_rate": 0.002722032741792582,
+      "loss": 1.2423,
+      "step": 2560
+    },
+    {
+      "epoch": 0.19698484731943697,
+      "learning_rate": 0.0027218225139106172,
+      "loss": 0.9579,
+      "step": 2561
+    },
+    {
+      "epoch": 0.1970617644796554,
+      "learning_rate": 0.002721612214685163,
+      "loss": 1.0188,
+      "step": 2562
+    },
+    {
+      "epoch": 0.19713868163987386,
+      "learning_rate": 0.0027214018441284976,
+      "loss": 1.2168,
+      "step": 2563
+    },
+    {
+      "epoch": 0.1972155988000923,
+      "learning_rate": 0.0027211914022529063,
+      "loss": 0.9237,
+      "step": 2564
+    },
+    {
+      "epoch": 0.19729251596031075,
+      "learning_rate": 0.002720980889070676,
+      "loss": 1.4665,
+      "step": 2565
+    },
+    {
+      "epoch": 0.19736943312052918,
+      "learning_rate": 0.002720770304594099,
+      "loss": 1.1105,
+      "step": 2566
+    },
+    {
+      "epoch": 0.19744635028074764,
+      "learning_rate": 0.0027205596488354723,
+      "loss": 1.3942,
+      "step": 2567
+    },
+    {
+      "epoch": 0.19752326744096607,
+      "learning_rate": 0.002720348921807095,
+      "loss": 1.0266,
+      "step": 2568
+    },
+    {
+      "epoch": 0.19760018460118453,
+      "learning_rate": 0.0027201381235212726,
+      "loss": 1.2189,
+      "step": 2569
+    },
+    {
+      "epoch": 0.19767710176140296,
+      "learning_rate": 0.0027199272539903136,
+      "loss": 1.0854,
+      "step": 2570
+    },
+    {
+      "epoch": 0.19775401892162142,
+      "learning_rate": 0.0027197163132265307,
+      "loss": 1.4957,
+      "step": 2571
+    },
+    {
+      "epoch": 0.19783093608183985,
+      "learning_rate": 0.0027195053012422413,
+      "loss": 1.0693,
+      "step": 2572
+    },
+    {
+      "epoch": 0.1979078532420583,
+      "learning_rate": 0.0027192942180497663,
+      "loss": 0.7972,
+      "step": 2573
+    },
+    {
+      "epoch": 0.19798477040227674,
+      "learning_rate": 0.0027190830636614313,
+      "loss": 1.1248,
+      "step": 2574
+    },
+    {
+      "epoch": 0.1980616875624952,
+      "learning_rate": 0.002718871838089566,
+      "loss": 1.4307,
+      "step": 2575
+    },
+    {
+      "epoch": 0.19813860472271363,
+      "learning_rate": 0.002718660541346503,
+      "loss": 1.168,
+      "step": 2576
+    },
+    {
+      "epoch": 0.1982155218829321,
+      "learning_rate": 0.0027184491734445816,
+      "loss": 1.1479,
+      "step": 2577
+    },
+    {
+      "epoch": 0.19829243904315053,
+      "learning_rate": 0.002718237734396143,
+      "loss": 1.3318,
+      "step": 2578
+    },
+    {
+      "epoch": 0.19836935620336899,
+      "learning_rate": 0.0027180262242135333,
+      "loss": 0.8992,
+      "step": 2579
+    },
+    {
+      "epoch": 0.19844627336358742,
+      "learning_rate": 0.002717814642909103,
+      "loss": 1.1816,
+      "step": 2580
+    },
+    {
+      "epoch": 0.19852319052380585,
+      "learning_rate": 0.0027176029904952066,
+      "loss": 1.1761,
+      "step": 2581
+    },
+    {
+      "epoch": 0.1986001076840243,
+      "learning_rate": 0.0027173912669842025,
+      "loss": 1.6291,
+      "step": 2582
+    },
+    {
+      "epoch": 0.19867702484424274,
+      "learning_rate": 0.002717179472388454,
+      "loss": 1.3271,
+      "step": 2583
+    },
+    {
+      "epoch": 0.1987539420044612,
+      "learning_rate": 0.0027169676067203274,
+      "loss": 1.136,
+      "step": 2584
+    },
+    {
+      "epoch": 0.19883085916467963,
+      "learning_rate": 0.0027167556699921936,
+      "loss": 1.3296,
+      "step": 2585
+    },
+    {
+      "epoch": 0.1989077763248981,
+      "learning_rate": 0.0027165436622164285,
+      "loss": 1.1308,
+      "step": 2586
+    },
+    {
+      "epoch": 0.19898469348511652,
+      "learning_rate": 0.002716331583405411,
+      "loss": 1.2838,
+      "step": 2587
+    },
+    {
+      "epoch": 0.19906161064533498,
+      "learning_rate": 0.0027161194335715245,
+      "loss": 1.2179,
+      "step": 2588
+    },
+    {
+      "epoch": 0.1991385278055534,
+      "learning_rate": 0.0027159072127271575,
+      "loss": 1.3697,
+      "step": 2589
+    },
+    {
+      "epoch": 0.19921544496577187,
+      "learning_rate": 0.0027156949208847004,
+      "loss": 1.3041,
+      "step": 2590
+    },
+    {
+      "epoch": 0.1992923621259903,
+      "learning_rate": 0.0027154825580565506,
+      "loss": 0.8961,
+      "step": 2591
+    },
+    {
+      "epoch": 0.19936927928620876,
+      "learning_rate": 0.0027152701242551067,
+      "loss": 1.1366,
+      "step": 2592
+    },
+    {
+      "epoch": 0.1994461964464272,
+      "learning_rate": 0.0027150576194927747,
+      "loss": 1.0879,
+      "step": 2593
+    },
+    {
+      "epoch": 0.19952311360664565,
+      "learning_rate": 0.002714845043781961,
+      "loss": 1.441,
+      "step": 2594
+    },
+    {
+      "epoch": 0.19960003076686408,
+      "learning_rate": 0.00271463239713508,
+      "loss": 1.2458,
+      "step": 2595
+    },
+    {
+      "epoch": 0.19967694792708254,
+      "learning_rate": 0.002714419679564547,
+      "loss": 0.899,
+      "step": 2596
+    },
+    {
+      "epoch": 0.19975386508730097,
+      "learning_rate": 0.0027142068910827833,
+      "loss": 1.0908,
+      "step": 2597
+    },
+    {
+      "epoch": 0.19983078224751943,
+      "learning_rate": 0.0027139940317022137,
+      "loss": 1.3489,
+      "step": 2598
+    },
+    {
+      "epoch": 0.19990769940773787,
+      "learning_rate": 0.002713781101435268,
+      "loss": 1.4459,
+      "step": 2599
+    },
+    {
+      "epoch": 0.19998461656795632,
+      "learning_rate": 0.002713568100294378,
+      "loss": 1.1666,
+      "step": 2600
+    },
+    {
+      "epoch": 0.20006153372817476,
+      "learning_rate": 0.0027133550282919822,
+      "loss": 1.4039,
+      "step": 2601
+    },
+    {
+      "epoch": 0.2001384508883932,
+      "learning_rate": 0.0027131418854405218,
+      "loss": 1.2346,
+      "step": 2602
+    },
+    {
+      "epoch": 0.20021536804861165,
+      "learning_rate": 0.002712928671752442,
+      "loss": 0.8976,
+      "step": 2603
+    },
+    {
+      "epoch": 0.20029228520883008,
+      "learning_rate": 0.0027127153872401936,
+      "loss": 1.128,
+      "step": 2604
+    },
+    {
+      "epoch": 0.20036920236904854,
+      "learning_rate": 0.0027125020319162295,
+      "loss": 0.9785,
+      "step": 2605
+    },
+    {
+      "epoch": 0.20044611952926697,
+      "learning_rate": 0.0027122886057930084,
+      "loss": 1.1928,
+      "step": 2606
+    },
+    {
+      "epoch": 0.20052303668948543,
+      "learning_rate": 0.0027120751088829925,
+      "loss": 0.9113,
+      "step": 2607
+    },
+    {
+      "epoch": 0.20059995384970386,
+      "learning_rate": 0.002711861541198647,
+      "loss": 0.9073,
+      "step": 2608
+    },
+    {
+      "epoch": 0.20067687100992232,
+      "learning_rate": 0.0027116479027524437,
+      "loss": 1.063,
+      "step": 2609
+    },
+    {
+      "epoch": 0.20075378817014075,
+      "learning_rate": 0.0027114341935568563,
+      "loss": 1.3355,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2008307053303592,
+      "learning_rate": 0.002711220413624364,
+      "loss": 1.2035,
+      "step": 2611
+    },
+    {
+      "epoch": 0.20090762249057764,
+      "learning_rate": 0.002711006562967449,
+      "loss": 1.4746,
+      "step": 2612
+    },
+    {
+      "epoch": 0.2009845396507961,
+      "learning_rate": 0.002710792641598599,
+      "loss": 1.3634,
+      "step": 2613
+    },
+    {
+      "epoch": 0.20106145681101453,
+      "learning_rate": 0.002710578649530305,
+      "loss": 1.2459,
+      "step": 2614
+    },
+    {
+      "epoch": 0.201138373971233,
+      "learning_rate": 0.002710364586775062,
+      "loss": 1.4056,
+      "step": 2615
+    },
+    {
+      "epoch": 0.20121529113145142,
+      "learning_rate": 0.002710150453345369,
+      "loss": 1.4892,
+      "step": 2616
+    },
+    {
+      "epoch": 0.20129220829166988,
+      "learning_rate": 0.00270993624925373,
+      "loss": 1.1893,
+      "step": 2617
+    },
+    {
+      "epoch": 0.20136912545188831,
+      "learning_rate": 0.002709721974512653,
+      "loss": 1.1919,
+      "step": 2618
+    },
+    {
+      "epoch": 0.20144604261210677,
+      "learning_rate": 0.0027095076291346483,
+      "loss": 1.3314,
+      "step": 2619
+    },
+    {
+      "epoch": 0.2015229597723252,
+      "learning_rate": 0.0027092932131322334,
+      "loss": 1.2867,
+      "step": 2620
+    },
+    {
+      "epoch": 0.20159987693254364,
+      "learning_rate": 0.0027090787265179267,
+      "loss": 1.4829,
+      "step": 2621
+    },
+    {
+      "epoch": 0.2016767940927621,
+      "learning_rate": 0.002708864169304254,
+      "loss": 0.7613,
+      "step": 2622
+    },
+    {
+      "epoch": 0.20175371125298053,
+      "learning_rate": 0.0027086495415037416,
+      "loss": 1.1193,
+      "step": 2623
+    },
+    {
+      "epoch": 0.201830628413199,
+      "learning_rate": 0.0027084348431289235,
+      "loss": 1.1944,
+      "step": 2624
+    },
+    {
+      "epoch": 0.20190754557341742,
+      "learning_rate": 0.0027082200741923353,
+      "loss": 1.3288,
+      "step": 2625
+    },
+    {
+      "epoch": 0.20198446273363588,
+      "learning_rate": 0.002708005234706518,
+      "loss": 1.0715,
+      "step": 2626
+    },
+    {
+      "epoch": 0.2020613798938543,
+      "learning_rate": 0.002707790324684016,
+      "loss": 1.121,
+      "step": 2627
+    },
+    {
+      "epoch": 0.20213829705407277,
+      "learning_rate": 0.002707575344137378,
+      "loss": 0.808,
+      "step": 2628
+    },
+    {
+      "epoch": 0.2022152142142912,
+      "learning_rate": 0.0027073602930791575,
+      "loss": 1.3753,
+      "step": 2629
+    },
+    {
+      "epoch": 0.20229213137450966,
+      "learning_rate": 0.0027071451715219106,
+      "loss": 0.9612,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2023690485347281,
+      "learning_rate": 0.0027069299794781998,
+      "loss": 1.4398,
+      "step": 2631
+    },
+    {
+      "epoch": 0.20244596569494655,
+      "learning_rate": 0.0027067147169605892,
+      "loss": 1.1476,
+      "step": 2632
+    },
+    {
+      "epoch": 0.20252288285516498,
+      "learning_rate": 0.002706499383981649,
+      "loss": 1.1654,
+      "step": 2633
+    },
+    {
+      "epoch": 0.20259980001538344,
+      "learning_rate": 0.002706283980553952,
+      "loss": 0.8518,
+      "step": 2634
+    },
+    {
+      "epoch": 0.20267671717560187,
+      "learning_rate": 0.0027060685066900765,
+      "loss": 1.1888,
+      "step": 2635
+    },
+    {
+      "epoch": 0.20275363433582033,
+      "learning_rate": 0.0027058529624026037,
+      "loss": 1.0883,
+      "step": 2636
+    },
+    {
+      "epoch": 0.20283055149603876,
+      "learning_rate": 0.0027056373477041197,
+      "loss": 0.9727,
+      "step": 2637
+    },
+    {
+      "epoch": 0.20290746865625722,
+      "learning_rate": 0.0027054216626072147,
+      "loss": 1.0949,
+      "step": 2638
+    },
+    {
+      "epoch": 0.20298438581647565,
+      "learning_rate": 0.0027052059071244828,
+      "loss": 1.2156,
+      "step": 2639
+    },
+    {
+      "epoch": 0.2030613029766941,
+      "learning_rate": 0.002704990081268522,
+      "loss": 1.4779,
+      "step": 2640
+    },
+    {
+      "epoch": 0.20313822013691255,
+      "learning_rate": 0.002704774185051934,
+      "loss": 1.258,
+      "step": 2641
+    },
+    {
+      "epoch": 0.20321513729713098,
+      "learning_rate": 0.0027045582184873256,
+      "loss": 0.8716,
+      "step": 2642
+    },
+    {
+      "epoch": 0.20329205445734944,
+      "learning_rate": 0.002704342181587308,
+      "loss": 1.1194,
+      "step": 2643
+    },
+    {
+      "epoch": 0.20336897161756787,
+      "learning_rate": 0.0027041260743644953,
+      "loss": 1.3263,
+      "step": 2644
+    },
+    {
+      "epoch": 0.20344588877778633,
+      "learning_rate": 0.0027039098968315058,
+      "loss": 1.1777,
+      "step": 2645
+    },
+    {
+      "epoch": 0.20352280593800476,
+      "learning_rate": 0.0027036936490009633,
+      "loss": 1.062,
+      "step": 2646
+    },
+    {
+      "epoch": 0.20359972309822322,
+      "learning_rate": 0.002703477330885494,
+      "loss": 1.0762,
+      "step": 2647
+    },
+    {
+      "epoch": 0.20367664025844165,
+      "learning_rate": 0.0027032609424977286,
+      "loss": 0.9619,
+      "step": 2648
+    },
+    {
+      "epoch": 0.2037535574186601,
+      "learning_rate": 0.002703044483850303,
+      "loss": 0.9424,
+      "step": 2649
+    },
+    {
+      "epoch": 0.20383047457887854,
+      "learning_rate": 0.0027028279549558562,
+      "loss": 1.333,
+      "step": 2650
+    },
+    {
+      "epoch": 0.203907391739097,
+      "learning_rate": 0.002702611355827032,
+      "loss": 1.2799,
+      "step": 2651
+    },
+    {
+      "epoch": 0.20398430889931543,
+      "learning_rate": 0.002702394686476477,
+      "loss": 1.186,
+      "step": 2652
+    },
+    {
+      "epoch": 0.2040612260595339,
+      "learning_rate": 0.002702177946916843,
+      "loss": 1.1085,
+      "step": 2653
+    },
+    {
+      "epoch": 0.20413814321975232,
+      "learning_rate": 0.0027019611371607863,
+      "loss": 1.0352,
+      "step": 2654
+    },
+    {
+      "epoch": 0.20421506037997078,
+      "learning_rate": 0.0027017442572209657,
+      "loss": 1.4193,
+      "step": 2655
+    },
+    {
+      "epoch": 0.2042919775401892,
+      "learning_rate": 0.0027015273071100457,
+      "loss": 1.2035,
+      "step": 2656
+    },
+    {
+      "epoch": 0.20436889470040767,
+      "learning_rate": 0.002701310286840694,
+      "loss": 1.5892,
+      "step": 2657
+    },
+    {
+      "epoch": 0.2044458118606261,
+      "learning_rate": 0.0027010931964255827,
+      "loss": 0.9946,
+      "step": 2658
+    },
+    {
+      "epoch": 0.20452272902084456,
+      "learning_rate": 0.002700876035877388,
+      "loss": 1.1391,
+      "step": 2659
+    },
+    {
+      "epoch": 0.204599646181063,
+      "learning_rate": 0.00270065880520879,
+      "loss": 0.9078,
+      "step": 2660
+    },
+    {
+      "epoch": 0.20467656334128145,
+      "learning_rate": 0.002700441504432473,
+      "loss": 1.0461,
+      "step": 2661
+    },
+    {
+      "epoch": 0.20475348050149988,
+      "learning_rate": 0.0027002241335611255,
+      "loss": 0.9501,
+      "step": 2662
+    },
+    {
+      "epoch": 0.20483039766171832,
+      "learning_rate": 0.00270000669260744,
+      "loss": 1.4062,
+      "step": 2663
+    },
+    {
+      "epoch": 0.20490731482193678,
+      "learning_rate": 0.002699789181584113,
+      "loss": 1.2921,
+      "step": 2664
+    },
+    {
+      "epoch": 0.2049842319821552,
+      "learning_rate": 0.0026995716005038454,
+      "loss": 1.4167,
+      "step": 2665
+    },
+    {
+      "epoch": 0.20506114914237367,
+      "learning_rate": 0.002699353949379342,
+      "loss": 1.1758,
+      "step": 2666
+    },
+    {
+      "epoch": 0.2051380663025921,
+      "learning_rate": 0.0026991362282233113,
+      "loss": 1.2451,
+      "step": 2667
+    },
+    {
+      "epoch": 0.20521498346281056,
+      "learning_rate": 0.0026989184370484667,
+      "loss": 1.105,
+      "step": 2668
+    },
+    {
+      "epoch": 0.205291900623029,
+      "learning_rate": 0.0026987005758675247,
+      "loss": 0.9573,
+      "step": 2669
+    },
+    {
+      "epoch": 0.20536881778324745,
+      "learning_rate": 0.002698482644693207,
+      "loss": 1.0918,
+      "step": 2670
+    },
+    {
+      "epoch": 0.20544573494346588,
+      "learning_rate": 0.002698264643538239,
+      "loss": 1.137,
+      "step": 2671
+    },
+    {
+      "epoch": 0.20552265210368434,
+      "learning_rate": 0.0026980465724153494,
+      "loss": 1.3692,
+      "step": 2672
+    },
+    {
+      "epoch": 0.20559956926390277,
+      "learning_rate": 0.002697828431337272,
+      "loss": 1.614,
+      "step": 2673
+    },
+    {
+      "epoch": 0.20567648642412123,
+      "learning_rate": 0.0026976102203167445,
+      "loss": 1.3403,
+      "step": 2674
+    },
+    {
+      "epoch": 0.20575340358433966,
+      "learning_rate": 0.0026973919393665073,
+      "loss": 1.0917,
+      "step": 2675
+    },
+    {
+      "epoch": 0.20583032074455812,
+      "learning_rate": 0.002697173588499307,
+      "loss": 1.7301,
+      "step": 2676
+    },
+    {
+      "epoch": 0.20590723790477655,
+      "learning_rate": 0.0026969551677278937,
+      "loss": 1.4602,
+      "step": 2677
+    },
+    {
+      "epoch": 0.205984155064995,
+      "learning_rate": 0.0026967366770650207,
+      "loss": 1.1644,
+      "step": 2678
+    },
+    {
+      "epoch": 0.20606107222521344,
+      "learning_rate": 0.0026965181165234456,
+      "loss": 1.2144,
+      "step": 2679
+    },
+    {
+      "epoch": 0.2061379893854319,
+      "learning_rate": 0.002696299486115931,
+      "loss": 1.0635,
+      "step": 2680
+    },
+    {
+      "epoch": 0.20621490654565033,
+      "learning_rate": 0.0026960807858552424,
+      "loss": 1.1161,
+      "step": 2681
+    },
+    {
+      "epoch": 0.2062918237058688,
+      "learning_rate": 0.00269586201575415,
+      "loss": 1.135,
+      "step": 2682
+    },
+    {
+      "epoch": 0.20636874086608722,
+      "learning_rate": 0.0026956431758254286,
+      "loss": 1.1538,
+      "step": 2683
+    },
+    {
+      "epoch": 0.20644565802630566,
+      "learning_rate": 0.002695424266081856,
+      "loss": 1.3421,
+      "step": 2684
+    },
+    {
+      "epoch": 0.20652257518652412,
+      "learning_rate": 0.0026952052865362145,
+      "loss": 0.6206,
+      "step": 2685
+    },
+    {
+      "epoch": 0.20659949234674255,
+      "learning_rate": 0.0026949862372012906,
+      "loss": 1.226,
+      "step": 2686
+    },
+    {
+      "epoch": 0.206676409506961,
+      "learning_rate": 0.002694767118089875,
+      "loss": 1.194,
+      "step": 2687
+    },
+    {
+      "epoch": 0.20675332666717944,
+      "learning_rate": 0.0026945479292147623,
+      "loss": 1.1061,
+      "step": 2688
+    },
+    {
+      "epoch": 0.2068302438273979,
+      "learning_rate": 0.0026943286705887514,
+      "loss": 1.2629,
+      "step": 2689
+    },
+    {
+      "epoch": 0.20690716098761633,
+      "learning_rate": 0.002694109342224644,
+      "loss": 0.9402,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2069840781478348,
+      "learning_rate": 0.0026938899441352484,
+      "loss": 1.1339,
+      "step": 2691
+    },
+    {
+      "epoch": 0.20706099530805322,
+      "learning_rate": 0.0026936704763333742,
+      "loss": 1.0795,
+      "step": 2692
+    },
+    {
+      "epoch": 0.20713791246827168,
+      "learning_rate": 0.002693450938831837,
+      "loss": 1.2599,
+      "step": 2693
+    },
+    {
+      "epoch": 0.2072148296284901,
+      "learning_rate": 0.002693231331643456,
+      "loss": 1.2991,
+      "step": 2694
+    },
+    {
+      "epoch": 0.20729174678870857,
+      "learning_rate": 0.002693011654781054,
+      "loss": 0.9029,
+      "step": 2695
+    },
+    {
+      "epoch": 0.207368663948927,
+      "learning_rate": 0.002692791908257458,
+      "loss": 1.1686,
+      "step": 2696
+    },
+    {
+      "epoch": 0.20744558110914546,
+      "learning_rate": 0.0026925720920854992,
+      "loss": 1.1356,
+      "step": 2697
+    },
+    {
+      "epoch": 0.2075224982693639,
+      "learning_rate": 0.0026923522062780128,
+      "loss": 1.0119,
+      "step": 2698
+    },
+    {
+      "epoch": 0.20759941542958235,
+      "learning_rate": 0.002692132250847839,
+      "loss": 1.5194,
+      "step": 2699
+    },
+    {
+      "epoch": 0.20767633258980078,
+      "learning_rate": 0.0026919122258078207,
+      "loss": 0.8539,
+      "step": 2700
+    },
+    {
+      "epoch": 0.20775324975001924,
+      "learning_rate": 0.002691692131170805,
+      "loss": 1.1621,
+      "step": 2701
+    },
+    {
+      "epoch": 0.20783016691023767,
+      "learning_rate": 0.0026914719669496436,
+      "loss": 1.0978,
+      "step": 2702
+    },
+    {
+      "epoch": 0.2079070840704561,
+      "learning_rate": 0.002691251733157193,
+      "loss": 1.0201,
+      "step": 2703
+    },
+    {
+      "epoch": 0.20798400123067456,
+      "learning_rate": 0.0026910314298063116,
+      "loss": 1.3523,
+      "step": 2704
+    },
+    {
+      "epoch": 0.208060918390893,
+      "learning_rate": 0.002690811056909864,
+      "loss": 1.374,
+      "step": 2705
+    },
+    {
+      "epoch": 0.20813783555111146,
+      "learning_rate": 0.0026905906144807177,
+      "loss": 1.2682,
+      "step": 2706
+    },
+    {
+      "epoch": 0.2082147527113299,
+      "learning_rate": 0.002690370102531745,
+      "loss": 1.0675,
+      "step": 2707
+    },
+    {
+      "epoch": 0.20829166987154835,
+      "learning_rate": 0.0026901495210758207,
+      "loss": 1.0536,
+      "step": 2708
+    },
+    {
+      "epoch": 0.20836858703176678,
+      "learning_rate": 0.0026899288701258263,
+      "loss": 1.2805,
+      "step": 2709
+    },
+    {
+      "epoch": 0.20844550419198524,
+      "learning_rate": 0.0026897081496946446,
+      "loss": 0.9618,
+      "step": 2710
+    },
+    {
+      "epoch": 0.20852242135220367,
+      "learning_rate": 0.002689487359795164,
+      "loss": 0.9159,
+      "step": 2711
+    },
+    {
+      "epoch": 0.20859933851242213,
+      "learning_rate": 0.0026892665004402774,
+      "loss": 0.8268,
+      "step": 2712
+    },
+    {
+      "epoch": 0.20867625567264056,
+      "learning_rate": 0.00268904557164288,
+      "loss": 1.5303,
+      "step": 2713
+    },
+    {
+      "epoch": 0.20875317283285902,
+      "learning_rate": 0.0026888245734158723,
+      "loss": 0.9274,
+      "step": 2714
+    },
+    {
+      "epoch": 0.20883008999307745,
+      "learning_rate": 0.0026886035057721593,
+      "loss": 1.3888,
+      "step": 2715
+    },
+    {
+      "epoch": 0.2089070071532959,
+      "learning_rate": 0.0026883823687246485,
+      "loss": 0.9998,
+      "step": 2716
+    },
+    {
+      "epoch": 0.20898392431351434,
+      "learning_rate": 0.0026881611622862527,
+      "loss": 1.1488,
+      "step": 2717
+    },
+    {
+      "epoch": 0.2090608414737328,
+      "learning_rate": 0.0026879398864698887,
+      "loss": 1.058,
+      "step": 2718
+    },
+    {
+      "epoch": 0.20913775863395123,
+      "learning_rate": 0.0026877185412884762,
+      "loss": 1.1346,
+      "step": 2719
+    },
+    {
+      "epoch": 0.2092146757941697,
+      "learning_rate": 0.002687497126754941,
+      "loss": 0.9705,
+      "step": 2720
+    },
+    {
+      "epoch": 0.20929159295438812,
+      "learning_rate": 0.00268727564288221,
+      "loss": 1.2282,
+      "step": 2721
+    },
+    {
+      "epoch": 0.20936851011460658,
+      "learning_rate": 0.002687054089683218,
+      "loss": 1.4496,
+      "step": 2722
+    },
+    {
+      "epoch": 0.209445427274825,
+      "learning_rate": 0.0026868324671708997,
+      "loss": 1.2114,
+      "step": 2723
+    },
+    {
+      "epoch": 0.20952234443504344,
+      "learning_rate": 0.002686610775358197,
+      "loss": 1.2715,
+      "step": 2724
+    },
+    {
+      "epoch": 0.2095992615952619,
+      "learning_rate": 0.0026863890142580543,
+      "loss": 1.1555,
+      "step": 2725
+    },
+    {
+      "epoch": 0.20967617875548034,
+      "learning_rate": 0.0026861671838834212,
+      "loss": 1.3641,
+      "step": 2726
+    },
+    {
+      "epoch": 0.2097530959156988,
+      "learning_rate": 0.0026859452842472493,
+      "loss": 0.9379,
+      "step": 2727
+    },
+    {
+      "epoch": 0.20983001307591723,
+      "learning_rate": 0.0026857233153624966,
+      "loss": 1.3668,
+      "step": 2728
+    },
+    {
+      "epoch": 0.20990693023613569,
+      "learning_rate": 0.002685501277242124,
+      "loss": 1.1526,
+      "step": 2729
+    },
+    {
+      "epoch": 0.20998384739635412,
+      "learning_rate": 0.002685279169899096,
+      "loss": 1.4391,
+      "step": 2730
+    },
+    {
+      "epoch": 0.21006076455657258,
+      "learning_rate": 0.002685056993346382,
+      "loss": 1.3306,
+      "step": 2731
+    },
+    {
+      "epoch": 0.210137681716791,
+      "learning_rate": 0.0026848347475969554,
+      "loss": 1.1268,
+      "step": 2732
+    },
+    {
+      "epoch": 0.21021459887700947,
+      "learning_rate": 0.002684612432663793,
+      "loss": 1.0123,
+      "step": 2733
+    },
+    {
+      "epoch": 0.2102915160372279,
+      "learning_rate": 0.002684390048559876,
+      "loss": 0.7296,
+      "step": 2734
+    },
+    {
+      "epoch": 0.21036843319744636,
+      "learning_rate": 0.0026841675952981895,
+      "loss": 1.0813,
+      "step": 2735
+    },
+    {
+      "epoch": 0.2104453503576648,
+      "learning_rate": 0.002683945072891724,
+      "loss": 1.1279,
+      "step": 2736
+    },
+    {
+      "epoch": 0.21052226751788325,
+      "learning_rate": 0.0026837224813534705,
+      "loss": 1.2752,
+      "step": 2737
+    },
+    {
+      "epoch": 0.21059918467810168,
+      "learning_rate": 0.002683499820696428,
+      "loss": 1.0637,
+      "step": 2738
+    },
+    {
+      "epoch": 0.21067610183832014,
+      "learning_rate": 0.0026832770909335983,
+      "loss": 1.0091,
+      "step": 2739
+    },
+    {
+      "epoch": 0.21075301899853857,
+      "learning_rate": 0.0026830542920779858,
+      "loss": 1.1888,
+      "step": 2740
+    },
+    {
+      "epoch": 0.21082993615875703,
+      "learning_rate": 0.0026828314241425996,
+      "loss": 0.8736,
+      "step": 2741
+    },
+    {
+      "epoch": 0.21090685331897546,
+      "learning_rate": 0.0026826084871404542,
+      "loss": 1.1327,
+      "step": 2742
+    },
+    {
+      "epoch": 0.21098377047919392,
+      "learning_rate": 0.002682385481084567,
+      "loss": 1.0802,
+      "step": 2743
+    },
+    {
+      "epoch": 0.21106068763941235,
+      "learning_rate": 0.002682162405987959,
+      "loss": 1.1862,
+      "step": 2744
+    },
+    {
+      "epoch": 0.21113760479963078,
+      "learning_rate": 0.002681939261863656,
+      "loss": 1.0259,
+      "step": 2745
+    },
+    {
+      "epoch": 0.21121452195984924,
+      "learning_rate": 0.002681716048724688,
+      "loss": 1.4125,
+      "step": 2746
+    },
+    {
+      "epoch": 0.21129143912006768,
+      "learning_rate": 0.002681492766584088,
+      "loss": 1.3,
+      "step": 2747
+    },
+    {
+      "epoch": 0.21136835628028613,
+      "learning_rate": 0.0026812694154548945,
+      "loss": 1.0045,
+      "step": 2748
+    },
+    {
+      "epoch": 0.21144527344050457,
+      "learning_rate": 0.0026810459953501486,
+      "loss": 1.0391,
+      "step": 2749
+    },
+    {
+      "epoch": 0.21152219060072303,
+      "learning_rate": 0.002680822506282896,
+      "loss": 1.1679,
+      "step": 2750
+    },
+    {
+      "epoch": 0.21159910776094146,
+      "learning_rate": 0.002680598948266187,
+      "loss": 0.96,
+      "step": 2751
+    },
+    {
+      "epoch": 0.21167602492115992,
+      "learning_rate": 0.0026803753213130747,
+      "loss": 1.0611,
+      "step": 2752
+    },
+    {
+      "epoch": 0.21175294208137835,
+      "learning_rate": 0.0026801516254366174,
+      "loss": 1.3594,
+      "step": 2753
+    },
+    {
+      "epoch": 0.2118298592415968,
+      "learning_rate": 0.0026799278606498764,
+      "loss": 0.9369,
+      "step": 2754
+    },
+    {
+      "epoch": 0.21190677640181524,
+      "learning_rate": 0.0026797040269659184,
+      "loss": 1.237,
+      "step": 2755
+    },
+    {
+      "epoch": 0.2119836935620337,
+      "learning_rate": 0.0026794801243978125,
+      "loss": 0.8864,
+      "step": 2756
+    },
+    {
+      "epoch": 0.21206061072225213,
+      "learning_rate": 0.002679256152958633,
+      "loss": 1.373,
+      "step": 2757
+    },
+    {
+      "epoch": 0.2121375278824706,
+      "learning_rate": 0.002679032112661458,
+      "loss": 1.2083,
+      "step": 2758
+    },
+    {
+      "epoch": 0.21221444504268902,
+      "learning_rate": 0.002678808003519369,
+      "loss": 1.1763,
+      "step": 2759
+    },
+    {
+      "epoch": 0.21229136220290748,
+      "learning_rate": 0.002678583825545452,
+      "loss": 1.2056,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2123682793631259,
+      "learning_rate": 0.0026783595787527974,
+      "loss": 1.6132,
+      "step": 2761
+    },
+    {
+      "epoch": 0.21244519652334437,
+      "learning_rate": 0.0026781352631544985,
+      "loss": 1.074,
+      "step": 2762
+    },
+    {
+      "epoch": 0.2125221136835628,
+      "learning_rate": 0.0026779108787636542,
+      "loss": 1.371,
+      "step": 2763
+    },
+    {
+      "epoch": 0.21259903084378123,
+      "learning_rate": 0.002677686425593366,
+      "loss": 1.1205,
+      "step": 2764
+    },
+    {
+      "epoch": 0.2126759480039997,
+      "learning_rate": 0.0026774619036567396,
+      "loss": 1.0216,
+      "step": 2765
+    },
+    {
+      "epoch": 0.21275286516421812,
+      "learning_rate": 0.0026772373129668858,
+      "loss": 1.3456,
+      "step": 2766
+    },
+    {
+      "epoch": 0.21282978232443658,
+      "learning_rate": 0.0026770126535369185,
+      "loss": 1.093,
+      "step": 2767
+    },
+    {
+      "epoch": 0.21290669948465502,
+      "learning_rate": 0.0026767879253799557,
+      "loss": 1.3328,
+      "step": 2768
+    },
+    {
+      "epoch": 0.21298361664487347,
+      "learning_rate": 0.00267656312850912,
+      "loss": 1.1032,
+      "step": 2769
+    },
+    {
+      "epoch": 0.2130605338050919,
+      "learning_rate": 0.0026763382629375367,
+      "loss": 1.1695,
+      "step": 2770
+    },
+    {
+      "epoch": 0.21313745096531037,
+      "learning_rate": 0.002676113328678336,
+      "loss": 1.0566,
+      "step": 2771
+    },
+    {
+      "epoch": 0.2132143681255288,
+      "learning_rate": 0.0026758883257446526,
+      "loss": 1.3098,
+      "step": 2772
+    },
+    {
+      "epoch": 0.21329128528574726,
+      "learning_rate": 0.002675663254149624,
+      "loss": 1.3136,
+      "step": 2773
+    },
+    {
+      "epoch": 0.2133682024459657,
+      "learning_rate": 0.0026754381139063926,
+      "loss": 1.2384,
+      "step": 2774
+    },
+    {
+      "epoch": 0.21344511960618415,
+      "learning_rate": 0.002675212905028105,
+      "loss": 0.8283,
+      "step": 2775
+    },
+    {
+      "epoch": 0.21352203676640258,
+      "learning_rate": 0.0026749876275279118,
+      "loss": 1.0777,
+      "step": 2776
+    },
+    {
+      "epoch": 0.21359895392662104,
+      "learning_rate": 0.002674762281418966,
+      "loss": 1.1138,
+      "step": 2777
+    },
+    {
+      "epoch": 0.21367587108683947,
+      "learning_rate": 0.002674536866714426,
+      "loss": 0.8933,
+      "step": 2778
+    },
+    {
+      "epoch": 0.21375278824705793,
+      "learning_rate": 0.0026743113834274547,
+      "loss": 1.2152,
+      "step": 2779
+    },
+    {
+      "epoch": 0.21382970540727636,
+      "learning_rate": 0.0026740858315712174,
+      "loss": 1.2671,
+      "step": 2780
+    },
+    {
+      "epoch": 0.21390662256749482,
+      "learning_rate": 0.002673860211158885,
+      "loss": 1.1557,
+      "step": 2781
+    },
+    {
+      "epoch": 0.21398353972771325,
+      "learning_rate": 0.0026736345222036317,
+      "loss": 0.9709,
+      "step": 2782
+    },
+    {
+      "epoch": 0.2140604568879317,
+      "learning_rate": 0.002673408764718635,
+      "loss": 1.2213,
+      "step": 2783
+    },
+    {
+      "epoch": 0.21413737404815014,
+      "learning_rate": 0.002673182938717078,
+      "loss": 0.9639,
+      "step": 2784
+    },
+    {
+      "epoch": 0.21421429120836857,
+      "learning_rate": 0.0026729570442121466,
+      "loss": 1.2288,
+      "step": 2785
+    },
+    {
+      "epoch": 0.21429120836858703,
+      "learning_rate": 0.0026727310812170307,
+      "loss": 1.1893,
+      "step": 2786
+    },
+    {
+      "epoch": 0.21436812552880546,
+      "learning_rate": 0.0026725050497449256,
+      "loss": 1.0382,
+      "step": 2787
+    },
+    {
+      "epoch": 0.21444504268902392,
+      "learning_rate": 0.002672278949809028,
+      "loss": 1.1415,
+      "step": 2788
+    },
+    {
+      "epoch": 0.21452195984924236,
+      "learning_rate": 0.0026720527814225407,
+      "loss": 1.0127,
+      "step": 2789
+    },
+    {
+      "epoch": 0.21459887700946081,
+      "learning_rate": 0.00267182654459867,
+      "loss": 1.2729,
+      "step": 2790
+    },
+    {
+      "epoch": 0.21467579416967925,
+      "learning_rate": 0.0026716002393506262,
+      "loss": 1.2256,
+      "step": 2791
+    },
+    {
+      "epoch": 0.2147527113298977,
+      "learning_rate": 0.0026713738656916237,
+      "loss": 1.4644,
+      "step": 2792
+    },
+    {
+      "epoch": 0.21482962849011614,
+      "learning_rate": 0.0026711474236348802,
+      "loss": 0.797,
+      "step": 2793
+    },
+    {
+      "epoch": 0.2149065456503346,
+      "learning_rate": 0.002670920913193618,
+      "loss": 1.2909,
+      "step": 2794
+    },
+    {
+      "epoch": 0.21498346281055303,
+      "learning_rate": 0.0026706943343810636,
+      "loss": 1.3884,
+      "step": 2795
+    },
+    {
+      "epoch": 0.2150603799707715,
+      "learning_rate": 0.0026704676872104467,
+      "loss": 1.1233,
+      "step": 2796
+    },
+    {
+      "epoch": 0.21513729713098992,
+      "learning_rate": 0.0026702409716950017,
+      "loss": 1.3233,
+      "step": 2797
+    },
+    {
+      "epoch": 0.21521421429120838,
+      "learning_rate": 0.002670014187847967,
+      "loss": 1.0681,
+      "step": 2798
+    },
+    {
+      "epoch": 0.2152911314514268,
+      "learning_rate": 0.0026697873356825844,
+      "loss": 1.3902,
+      "step": 2799
+    },
+    {
+      "epoch": 0.21536804861164527,
+      "learning_rate": 0.0026695604152121005,
+      "loss": 0.9033,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2154449657718637,
+      "learning_rate": 0.0026693334264497647,
+      "loss": 1.2176,
+      "step": 2801
+    },
+    {
+      "epoch": 0.21552188293208216,
+      "learning_rate": 0.0026691063694088313,
+      "loss": 1.4218,
+      "step": 2802
+    },
+    {
+      "epoch": 0.2155988000923006,
+      "learning_rate": 0.0026688792441025588,
+      "loss": 1.2279,
+      "step": 2803
+    },
+    {
+      "epoch": 0.21567571725251905,
+      "learning_rate": 0.002668652050544209,
+      "loss": 1.4316,
+      "step": 2804
+    },
+    {
+      "epoch": 0.21575263441273748,
+      "learning_rate": 0.002668424788747048,
+      "loss": 0.9718,
+      "step": 2805
+    },
+    {
+      "epoch": 0.2158295515729559,
+      "learning_rate": 0.002668197458724346,
+      "loss": 1.4095,
+      "step": 2806
+    },
+    {
+      "epoch": 0.21590646873317437,
+      "learning_rate": 0.0026679700604893765,
+      "loss": 1.143,
+      "step": 2807
+    },
+    {
+      "epoch": 0.2159833858933928,
+      "learning_rate": 0.002667742594055419,
+      "loss": 1.2975,
+      "step": 2808
+    },
+    {
+      "epoch": 0.21606030305361126,
+      "learning_rate": 0.0026675150594357533,
+      "loss": 0.6969,
+      "step": 2809
+    },
+    {
+      "epoch": 0.2161372202138297,
+      "learning_rate": 0.0026672874566436673,
+      "loss": 1.015,
+      "step": 2810
+    },
+    {
+      "epoch": 0.21621413737404815,
+      "learning_rate": 0.0026670597856924493,
+      "loss": 1.2185,
+      "step": 2811
+    },
+    {
+      "epoch": 0.21629105453426659,
+      "learning_rate": 0.0026668320465953947,
+      "loss": 1.2983,
+      "step": 2812
+    },
+    {
+      "epoch": 0.21636797169448505,
+      "learning_rate": 0.0026666042393658014,
+      "loss": 1.0997,
+      "step": 2813
+    },
+    {
+      "epoch": 0.21644488885470348,
+      "learning_rate": 0.00266637636401697,
+      "loss": 1.1781,
+      "step": 2814
+    },
+    {
+      "epoch": 0.21652180601492194,
+      "learning_rate": 0.002666148420562207,
+      "loss": 1.0728,
+      "step": 2815
+    },
+    {
+      "epoch": 0.21659872317514037,
+      "learning_rate": 0.002665920409014823,
+      "loss": 1.3593,
+      "step": 2816
+    },
+    {
+      "epoch": 0.21667564033535883,
+      "learning_rate": 0.0026656923293881312,
+      "loss": 0.9013,
+      "step": 2817
+    },
+    {
+      "epoch": 0.21675255749557726,
+      "learning_rate": 0.002665464181695449,
+      "loss": 2.0623,
+      "step": 2818
+    },
+    {
+      "epoch": 0.21682947465579572,
+      "learning_rate": 0.0026652359659500987,
+      "loss": 0.9032,
+      "step": 2819
+    },
+    {
+      "epoch": 0.21690639181601415,
+      "learning_rate": 0.0026650076821654062,
+      "loss": 1.2291,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2169833089762326,
+      "learning_rate": 0.002664779330354701,
+      "loss": 1.515,
+      "step": 2821
+    },
+    {
+      "epoch": 0.21706022613645104,
+      "learning_rate": 0.0026645509105313165,
+      "loss": 1.4406,
+      "step": 2822
+    },
+    {
+      "epoch": 0.2171371432966695,
+      "learning_rate": 0.002664322422708591,
+      "loss": 1.3025,
+      "step": 2823
+    },
+    {
+      "epoch": 0.21721406045688793,
+      "learning_rate": 0.0026640938668998655,
+      "loss": 1.0984,
+      "step": 2824
+    },
+    {
+      "epoch": 0.2172909776171064,
+      "learning_rate": 0.0026638652431184863,
+      "loss": 1.2292,
+      "step": 2825
+    },
+    {
+      "epoch": 0.21736789477732482,
+      "learning_rate": 0.0026636365513778024,
+      "loss": 1.1934,
+      "step": 2826
+    },
+    {
+      "epoch": 0.21744481193754325,
+      "learning_rate": 0.0026634077916911675,
+      "loss": 1.1607,
+      "step": 2827
+    },
+    {
+      "epoch": 0.2175217290977617,
+      "learning_rate": 0.002663178964071939,
+      "loss": 1.6558,
+      "step": 2828
+    },
+    {
+      "epoch": 0.21759864625798014,
+      "learning_rate": 0.0026629500685334782,
+      "loss": 1.4741,
+      "step": 2829
+    },
+    {
+      "epoch": 0.2176755634181986,
+      "learning_rate": 0.002662721105089152,
+      "loss": 1.2368,
+      "step": 2830
+    },
+    {
+      "epoch": 0.21775248057841703,
+      "learning_rate": 0.002662492073752328,
+      "loss": 0.9789,
+      "step": 2831
+    },
+    {
+      "epoch": 0.2178293977386355,
+      "learning_rate": 0.0026622629745363803,
+      "loss": 1.1643,
+      "step": 2832
+    },
+    {
+      "epoch": 0.21790631489885393,
+      "learning_rate": 0.0026620338074546863,
+      "loss": 1.3747,
+      "step": 2833
+    },
+    {
+      "epoch": 0.21798323205907238,
+      "learning_rate": 0.0026618045725206275,
+      "loss": 0.9163,
+      "step": 2834
+    },
+    {
+      "epoch": 0.21806014921929082,
+      "learning_rate": 0.0026615752697475886,
+      "loss": 1.019,
+      "step": 2835
+    },
+    {
+      "epoch": 0.21813706637950928,
+      "learning_rate": 0.0026613458991489596,
+      "loss": 0.892,
+      "step": 2836
+    },
+    {
+      "epoch": 0.2182139835397277,
+      "learning_rate": 0.002661116460738133,
+      "loss": 1.0551,
+      "step": 2837
+    },
+    {
+      "epoch": 0.21829090069994617,
+      "learning_rate": 0.0026608869545285056,
+      "loss": 1.1778,
+      "step": 2838
+    },
+    {
+      "epoch": 0.2183678178601646,
+      "learning_rate": 0.00266065738053348,
+      "loss": 1.3514,
+      "step": 2839
+    },
+    {
+      "epoch": 0.21844473502038306,
+      "learning_rate": 0.0026604277387664594,
+      "loss": 0.8138,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2185216521806015,
+      "learning_rate": 0.0026601980292408544,
+      "loss": 1.3814,
+      "step": 2841
+    },
+    {
+      "epoch": 0.21859856934081995,
+      "learning_rate": 0.002659968251970077,
+      "loss": 1.3129,
+      "step": 2842
+    },
+    {
+      "epoch": 0.21867548650103838,
+      "learning_rate": 0.002659738406967545,
+      "loss": 1.1769,
+      "step": 2843
+    },
+    {
+      "epoch": 0.21875240366125684,
+      "learning_rate": 0.0026595084942466784,
+      "loss": 1.0289,
+      "step": 2844
+    },
+    {
+      "epoch": 0.21882932082147527,
+      "learning_rate": 0.0026592785138209022,
+      "loss": 1.1511,
+      "step": 2845
+    },
+    {
+      "epoch": 0.2189062379816937,
+      "learning_rate": 0.002659048465703646,
+      "loss": 1.2311,
+      "step": 2846
+    },
+    {
+      "epoch": 0.21898315514191216,
+      "learning_rate": 0.0026588183499083413,
+      "loss": 1.0108,
+      "step": 2847
+    },
+    {
+      "epoch": 0.2190600723021306,
+      "learning_rate": 0.002658588166448426,
+      "loss": 1.1901,
+      "step": 2848
+    },
+    {
+      "epoch": 0.21913698946234905,
+      "learning_rate": 0.00265835791533734,
+      "loss": 0.9988,
+      "step": 2849
+    },
+    {
+      "epoch": 0.21921390662256748,
+      "learning_rate": 0.0026581275965885283,
+      "loss": 1.0638,
+      "step": 2850
+    },
+    {
+      "epoch": 0.21929082378278594,
+      "learning_rate": 0.0026578972102154386,
+      "loss": 1.2478,
+      "step": 2851
+    },
+    {
+      "epoch": 0.21936774094300437,
+      "learning_rate": 0.002657666756231525,
+      "loss": 1.3969,
+      "step": 2852
+    },
+    {
+      "epoch": 0.21944465810322283,
+      "learning_rate": 0.002657436234650242,
+      "loss": 1.2669,
+      "step": 2853
+    },
+    {
+      "epoch": 0.21952157526344127,
+      "learning_rate": 0.002657205645485052,
+      "loss": 1.2133,
+      "step": 2854
+    },
+    {
+      "epoch": 0.21959849242365972,
+      "learning_rate": 0.0026569749887494177,
+      "loss": 0.8658,
+      "step": 2855
+    },
+    {
+      "epoch": 0.21967540958387816,
+      "learning_rate": 0.002656744264456808,
+      "loss": 1.2947,
+      "step": 2856
+    },
+    {
+      "epoch": 0.21975232674409662,
+      "learning_rate": 0.0026565134726206957,
+      "loss": 1.3169,
+      "step": 2857
+    },
+    {
+      "epoch": 0.21982924390431505,
+      "learning_rate": 0.002656282613254556,
+      "loss": 1.1669,
+      "step": 2858
+    },
+    {
+      "epoch": 0.2199061610645335,
+      "learning_rate": 0.00265605168637187,
+      "loss": 1.3458,
+      "step": 2859
+    },
+    {
+      "epoch": 0.21998307822475194,
+      "learning_rate": 0.0026558206919861205,
+      "loss": 1.2217,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2200599953849704,
+      "learning_rate": 0.0026555896301107967,
+      "loss": 1.4341,
+      "step": 2861
+    },
+    {
+      "epoch": 0.22013691254518883,
+      "learning_rate": 0.0026553585007593893,
+      "loss": 0.9718,
+      "step": 2862
+    },
+    {
+      "epoch": 0.2202138297054073,
+      "learning_rate": 0.002655127303945396,
+      "loss": 1.3726,
+      "step": 2863
+    },
+    {
+      "epoch": 0.22029074686562572,
+      "learning_rate": 0.002654896039682315,
+      "loss": 1.0828,
+      "step": 2864
+    },
+    {
+      "epoch": 0.22036766402584418,
+      "learning_rate": 0.0026546647079836506,
+      "loss": 1.177,
+      "step": 2865
+    },
+    {
+      "epoch": 0.2204445811860626,
+      "learning_rate": 0.002654433308862911,
+      "loss": 1.1054,
+      "step": 2866
+    },
+    {
+      "epoch": 0.22052149834628104,
+      "learning_rate": 0.002654201842333607,
+      "loss": 1.4729,
+      "step": 2867
+    },
+    {
+      "epoch": 0.2205984155064995,
+      "learning_rate": 0.0026539703084092546,
+      "loss": 1.4224,
+      "step": 2868
+    },
+    {
+      "epoch": 0.22067533266671793,
+      "learning_rate": 0.0026537387071033734,
+      "loss": 0.9926,
+      "step": 2869
+    },
+    {
+      "epoch": 0.2207522498269364,
+      "learning_rate": 0.002653507038429487,
+      "loss": 1.4223,
+      "step": 2870
+    },
+    {
+      "epoch": 0.22082916698715482,
+      "learning_rate": 0.0026532753024011216,
+      "loss": 1.1191,
+      "step": 2871
+    },
+    {
+      "epoch": 0.22090608414737328,
+      "learning_rate": 0.0026530434990318097,
+      "loss": 1.4377,
+      "step": 2872
+    },
+    {
+      "epoch": 0.22098300130759171,
+      "learning_rate": 0.0026528116283350864,
+      "loss": 1.1086,
+      "step": 2873
+    },
+    {
+      "epoch": 0.22105991846781017,
+      "learning_rate": 0.002652579690324491,
+      "loss": 1.0963,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2211368356280286,
+      "learning_rate": 0.0026523476850135656,
+      "loss": 1.1156,
+      "step": 2875
+    },
+    {
+      "epoch": 0.22121375278824706,
+      "learning_rate": 0.0026521156124158585,
+      "loss": 0.9587,
+      "step": 2876
+    },
+    {
+      "epoch": 0.2212906699484655,
+      "learning_rate": 0.0026518834725449197,
+      "loss": 1.466,
+      "step": 2877
+    },
+    {
+      "epoch": 0.22136758710868396,
+      "learning_rate": 0.0026516512654143048,
+      "loss": 1.2132,
+      "step": 2878
+    },
+    {
+      "epoch": 0.2214445042689024,
+      "learning_rate": 0.0026514189910375723,
+      "loss": 1.0236,
+      "step": 2879
+    },
+    {
+      "epoch": 0.22152142142912085,
+      "learning_rate": 0.0026511866494282844,
+      "loss": 1.3139,
+      "step": 2880
+    },
+    {
+      "epoch": 0.22159833858933928,
+      "learning_rate": 0.0026509542406000087,
+      "loss": 1.0677,
+      "step": 2881
+    },
+    {
+      "epoch": 0.22167525574955774,
+      "learning_rate": 0.0026507217645663156,
+      "loss": 1.5382,
+      "step": 2882
+    },
+    {
+      "epoch": 0.22175217290977617,
+      "learning_rate": 0.002650489221340779,
+      "loss": 1.2255,
+      "step": 2883
+    },
+    {
+      "epoch": 0.22182909006999463,
+      "learning_rate": 0.002650256610936978,
+      "loss": 1.1327,
+      "step": 2884
+    },
+    {
+      "epoch": 0.22190600723021306,
+      "learning_rate": 0.0026500239333684945,
+      "loss": 0.9811,
+      "step": 2885
+    },
+    {
+      "epoch": 0.22198292439043152,
+      "learning_rate": 0.0026497911886489154,
+      "loss": 1.2869,
+      "step": 2886
+    },
+    {
+      "epoch": 0.22205984155064995,
+      "learning_rate": 0.0026495583767918297,
+      "loss": 1.4157,
+      "step": 2887
+    },
+    {
+      "epoch": 0.22213675871086838,
+      "learning_rate": 0.0026493254978108335,
+      "loss": 1.3822,
+      "step": 2888
+    },
+    {
+      "epoch": 0.22221367587108684,
+      "learning_rate": 0.002649092551719523,
+      "loss": 1.158,
+      "step": 2889
+    },
+    {
+      "epoch": 0.22229059303130527,
+      "learning_rate": 0.0026488595385315006,
+      "loss": 1.387,
+      "step": 2890
+    },
+    {
+      "epoch": 0.22236751019152373,
+      "learning_rate": 0.0026486264582603726,
+      "loss": 1.2149,
+      "step": 2891
+    },
+    {
+      "epoch": 0.22244442735174216,
+      "learning_rate": 0.002648393310919749,
+      "loss": 1.0277,
+      "step": 2892
+    },
+    {
+      "epoch": 0.22252134451196062,
+      "learning_rate": 0.002648160096523243,
+      "loss": 1.1177,
+      "step": 2893
+    },
+    {
+      "epoch": 0.22259826167217905,
+      "learning_rate": 0.0026479268150844722,
+      "loss": 1.1437,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2226751788323975,
+      "learning_rate": 0.0026476934666170584,
+      "loss": 1.0757,
+      "step": 2895
+    },
+    {
+      "epoch": 0.22275209599261594,
+      "learning_rate": 0.0026474600511346273,
+      "loss": 1.2584,
+      "step": 2896
+    },
+    {
+      "epoch": 0.2228290131528344,
+      "learning_rate": 0.002647226568650808,
+      "loss": 1.3191,
+      "step": 2897
+    },
+    {
+      "epoch": 0.22290593031305284,
+      "learning_rate": 0.0026469930191792327,
+      "loss": 1.0578,
+      "step": 2898
+    },
+    {
+      "epoch": 0.2229828474732713,
+      "learning_rate": 0.0026467594027335404,
+      "loss": 1.3145,
+      "step": 2899
+    },
+    {
+      "epoch": 0.22305976463348973,
+      "learning_rate": 0.0026465257193273714,
+      "loss": 1.1279,
+      "step": 2900
+    },
+    {
+      "epoch": 0.22313668179370819,
+      "learning_rate": 0.002646291968974371,
+      "loss": 1.2705,
+      "step": 2901
+    },
+    {
+      "epoch": 0.22321359895392662,
+      "learning_rate": 0.002646058151688188,
+      "loss": 1.0089,
+      "step": 2902
+    },
+    {
+      "epoch": 0.22329051611414508,
+      "learning_rate": 0.0026458242674824747,
+      "loss": 1.2941,
+      "step": 2903
+    },
+    {
+      "epoch": 0.2233674332743635,
+      "learning_rate": 0.0026455903163708884,
+      "loss": 0.9568,
+      "step": 2904
+    },
+    {
+      "epoch": 0.22344435043458197,
+      "learning_rate": 0.0026453562983670895,
+      "loss": 1.423,
+      "step": 2905
+    },
+    {
+      "epoch": 0.2235212675948004,
+      "learning_rate": 0.0026451222134847427,
+      "loss": 0.7131,
+      "step": 2906
+    },
+    {
+      "epoch": 0.22359818475501886,
+      "learning_rate": 0.0026448880617375168,
+      "loss": 0.8101,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2236751019152373,
+      "learning_rate": 0.002644653843139083,
+      "loss": 0.948,
+      "step": 2908
+    },
+    {
+      "epoch": 0.22375201907545572,
+      "learning_rate": 0.002644419557703119,
+      "loss": 1.1125,
+      "step": 2909
+    },
+    {
+      "epoch": 0.22382893623567418,
+      "learning_rate": 0.0026441852054433046,
+      "loss": 1.1147,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2239058533958926,
+      "learning_rate": 0.0026439507863733236,
+      "loss": 1.4192,
+      "step": 2911
+    },
+    {
+      "epoch": 0.22398277055611107,
+      "learning_rate": 0.002643716300506863,
+      "loss": 1.5509,
+      "step": 2912
+    },
+    {
+      "epoch": 0.2240596877163295,
+      "learning_rate": 0.0026434817478576166,
+      "loss": 0.9808,
+      "step": 2913
+    },
+    {
+      "epoch": 0.22413660487654796,
+      "learning_rate": 0.002643247128439279,
+      "loss": 1.0333,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2242135220367664,
+      "learning_rate": 0.00264301244226555,
+      "loss": 1.192,
+      "step": 2915
+    },
+    {
+      "epoch": 0.22429043919698485,
+      "learning_rate": 0.002642777689350133,
+      "loss": 1.7771,
+      "step": 2916
+    },
+    {
+      "epoch": 0.22436735635720328,
+      "learning_rate": 0.002642542869706736,
+      "loss": 1.197,
+      "step": 2917
+    },
+    {
+      "epoch": 0.22444427351742174,
+      "learning_rate": 0.00264230798334907,
+      "loss": 1.1411,
+      "step": 2918
+    },
+    {
+      "epoch": 0.22452119067764018,
+      "learning_rate": 0.002642073030290851,
+      "loss": 1.2446,
+      "step": 2919
+    },
+    {
+      "epoch": 0.22459810783785863,
+      "learning_rate": 0.002641838010545797,
+      "loss": 1.2408,
+      "step": 2920
+    },
+    {
+      "epoch": 0.22467502499807707,
+      "learning_rate": 0.0026416029241276315,
+      "loss": 1.1546,
+      "step": 2921
+    },
+    {
+      "epoch": 0.22475194215829553,
+      "learning_rate": 0.002641367771050082,
+      "loss": 1.1168,
+      "step": 2922
+    },
+    {
+      "epoch": 0.22482885931851396,
+      "learning_rate": 0.0026411325513268783,
+      "loss": 1.2393,
+      "step": 2923
+    },
+    {
+      "epoch": 0.22490577647873242,
+      "learning_rate": 0.0026408972649717556,
+      "loss": 0.9176,
+      "step": 2924
+    },
+    {
+      "epoch": 0.22498269363895085,
+      "learning_rate": 0.0026406619119984528,
+      "loss": 1.2219,
+      "step": 2925
+    },
+    {
+      "epoch": 0.2250596107991693,
+      "learning_rate": 0.0026404264924207127,
+      "loss": 1.2197,
+      "step": 2926
+    },
+    {
+      "epoch": 0.22513652795938774,
+      "learning_rate": 0.0026401910062522804,
+      "loss": 0.7539,
+      "step": 2927
+    },
+    {
+      "epoch": 0.22521344511960617,
+      "learning_rate": 0.0026399554535069076,
+      "loss": 1.005,
+      "step": 2928
+    },
+    {
+      "epoch": 0.22529036227982463,
+      "learning_rate": 0.002639719834198347,
+      "loss": 0.9408,
+      "step": 2929
+    },
+    {
+      "epoch": 0.22536727944004306,
+      "learning_rate": 0.0026394841483403583,
+      "loss": 1.1059,
+      "step": 2930
+    },
+    {
+      "epoch": 0.22544419660026152,
+      "learning_rate": 0.0026392483959467024,
+      "loss": 1.0097,
+      "step": 2931
+    },
+    {
+      "epoch": 0.22552111376047995,
+      "learning_rate": 0.002639012577031145,
+      "loss": 1.1187,
+      "step": 2932
+    },
+    {
+      "epoch": 0.2255980309206984,
+      "learning_rate": 0.0026387766916074564,
+      "loss": 0.6283,
+      "step": 2933
+    },
+    {
+      "epoch": 0.22567494808091684,
+      "learning_rate": 0.0026385407396894104,
+      "loss": 1.0521,
+      "step": 2934
+    },
+    {
+      "epoch": 0.2257518652411353,
+      "learning_rate": 0.0026383047212907838,
+      "loss": 0.8396,
+      "step": 2935
+    },
+    {
+      "epoch": 0.22582878240135373,
+      "learning_rate": 0.002638068636425358,
+      "loss": 1.1625,
+      "step": 2936
+    },
+    {
+      "epoch": 0.2259056995615722,
+      "learning_rate": 0.0026378324851069183,
+      "loss": 1.3229,
+      "step": 2937
+    },
+    {
+      "epoch": 0.22598261672179062,
+      "learning_rate": 0.0026375962673492546,
+      "loss": 1.1008,
+      "step": 2938
+    },
+    {
+      "epoch": 0.22605953388200908,
+      "learning_rate": 0.002637359983166159,
+      "loss": 0.9938,
+      "step": 2939
+    },
+    {
+      "epoch": 0.22613645104222752,
+      "learning_rate": 0.0026371236325714283,
+      "loss": 0.9123,
+      "step": 2940
+    },
+    {
+      "epoch": 0.22621336820244597,
+      "learning_rate": 0.002636887215578864,
+      "loss": 1.1797,
+      "step": 2941
+    },
+    {
+      "epoch": 0.2262902853626644,
+      "learning_rate": 0.00263665073220227,
+      "loss": 1.1467,
+      "step": 2942
+    },
+    {
+      "epoch": 0.22636720252288287,
+      "learning_rate": 0.0026364141824554554,
+      "loss": 1.1814,
+      "step": 2943
+    },
+    {
+      "epoch": 0.2264441196831013,
+      "learning_rate": 0.002636177566352233,
+      "loss": 1.1443,
+      "step": 2944
+    },
+    {
+      "epoch": 0.22652103684331976,
+      "learning_rate": 0.002635940883906417,
+      "loss": 1.1127,
+      "step": 2945
+    },
+    {
+      "epoch": 0.2265979540035382,
+      "learning_rate": 0.0026357041351318298,
+      "loss": 1.0529,
+      "step": 2946
+    },
+    {
+      "epoch": 0.22667487116375665,
+      "learning_rate": 0.002635467320042294,
+      "loss": 1.28,
+      "step": 2947
+    },
+    {
+      "epoch": 0.22675178832397508,
+      "learning_rate": 0.002635230438651639,
+      "loss": 1.2115,
+      "step": 2948
+    },
+    {
+      "epoch": 0.2268287054841935,
+      "learning_rate": 0.002634993490973695,
+      "loss": 1.1188,
+      "step": 2949
+    },
+    {
+      "epoch": 0.22690562264441197,
+      "learning_rate": 0.002634756477022298,
+      "loss": 1.1421,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2269825398046304,
+      "learning_rate": 0.0026345193968112884,
+      "loss": 1.4197,
+      "step": 2951
+    },
+    {
+      "epoch": 0.22705945696484886,
+      "learning_rate": 0.0026342822503545083,
+      "loss": 0.92,
+      "step": 2952
+    },
+    {
+      "epoch": 0.2271363741250673,
+      "learning_rate": 0.002634045037665805,
+      "loss": 0.9803,
+      "step": 2953
+    },
+    {
+      "epoch": 0.22721329128528575,
+      "learning_rate": 0.002633807758759031,
+      "loss": 1.1468,
+      "step": 2954
+    },
+    {
+      "epoch": 0.22729020844550418,
+      "learning_rate": 0.00263357041364804,
+      "loss": 1.2968,
+      "step": 2955
+    },
+    {
+      "epoch": 0.22736712560572264,
+      "learning_rate": 0.0026333330023466913,
+      "loss": 1.3053,
+      "step": 2956
+    },
+    {
+      "epoch": 0.22744404276594107,
+      "learning_rate": 0.0026330955248688474,
+      "loss": 1.1437,
+      "step": 2957
+    },
+    {
+      "epoch": 0.22752095992615953,
+      "learning_rate": 0.0026328579812283743,
+      "loss": 0.997,
+      "step": 2958
+    },
+    {
+      "epoch": 0.22759787708637796,
+      "learning_rate": 0.0026326203714391438,
+      "loss": 0.9116,
+      "step": 2959
+    },
+    {
+      "epoch": 0.22767479424659642,
+      "learning_rate": 0.0026323826955150288,
+      "loss": 0.6847,
+      "step": 2960
+    },
+    {
+      "epoch": 0.22775171140681486,
+      "learning_rate": 0.002632144953469909,
+      "loss": 1.314,
+      "step": 2961
+    },
+    {
+      "epoch": 0.22782862856703331,
+      "learning_rate": 0.0026319071453176648,
+      "loss": 0.8372,
+      "step": 2962
+    },
+    {
+      "epoch": 0.22790554572725175,
+      "learning_rate": 0.002631669271072183,
+      "loss": 1.0382,
+      "step": 2963
+    },
+    {
+      "epoch": 0.2279824628874702,
+      "learning_rate": 0.0026314313307473527,
+      "loss": 1.3457,
+      "step": 2964
+    },
+    {
+      "epoch": 0.22805938004768864,
+      "learning_rate": 0.002631193324357068,
+      "loss": 0.8498,
+      "step": 2965
+    },
+    {
+      "epoch": 0.2281362972079071,
+      "learning_rate": 0.0026309552519152265,
+      "loss": 1.146,
+      "step": 2966
+    },
+    {
+      "epoch": 0.22821321436812553,
+      "learning_rate": 0.002630717113435729,
+      "loss": 1.3942,
+      "step": 2967
+    },
+    {
+      "epoch": 0.228290131528344,
+      "learning_rate": 0.002630478908932481,
+      "loss": 1.0987,
+      "step": 2968
+    },
+    {
+      "epoch": 0.22836704868856242,
+      "learning_rate": 0.002630240638419391,
+      "loss": 1.2825,
+      "step": 2969
+    },
+    {
+      "epoch": 0.22844396584878085,
+      "learning_rate": 0.0026300023019103723,
+      "loss": 1.5577,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2285208830089993,
+      "learning_rate": 0.0026297638994193413,
+      "loss": 1.0042,
+      "step": 2971
+    },
+    {
+      "epoch": 0.22859780016921774,
+      "learning_rate": 0.002629525430960219,
+      "loss": 0.9451,
+      "step": 2972
+    },
+    {
+      "epoch": 0.2286747173294362,
+      "learning_rate": 0.00262928689654693,
+      "loss": 1.2184,
+      "step": 2973
+    },
+    {
+      "epoch": 0.22875163448965463,
+      "learning_rate": 0.002629048296193402,
+      "loss": 1.1151,
+      "step": 2974
+    },
+    {
+      "epoch": 0.2288285516498731,
+      "learning_rate": 0.002628809629913567,
+      "loss": 0.9534,
+      "step": 2975
+    },
+    {
+      "epoch": 0.22890546881009152,
+      "learning_rate": 0.002628570897721362,
+      "loss": 1.1324,
+      "step": 2976
+    },
+    {
+      "epoch": 0.22898238597030998,
+      "learning_rate": 0.0026283320996307253,
+      "loss": 0.9374,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2290593031305284,
+      "learning_rate": 0.0026280932356556016,
+      "loss": 1.3866,
+      "step": 2978
+    },
+    {
+      "epoch": 0.22913622029074687,
+      "learning_rate": 0.0026278543058099384,
+      "loss": 0.8686,
+      "step": 2979
+    },
+    {
+      "epoch": 0.2292131374509653,
+      "learning_rate": 0.0026276153101076867,
+      "loss": 1.2933,
+      "step": 2980
+    },
+    {
+      "epoch": 0.22929005461118376,
+      "learning_rate": 0.002627376248562802,
+      "loss": 1.4172,
+      "step": 2981
+    },
+    {
+      "epoch": 0.2293669717714022,
+      "learning_rate": 0.002627137121189243,
+      "loss": 1.0657,
+      "step": 2982
+    },
+    {
+      "epoch": 0.22944388893162065,
+      "learning_rate": 0.0026268979280009734,
+      "loss": 1.0249,
+      "step": 2983
+    },
+    {
+      "epoch": 0.22952080609183909,
+      "learning_rate": 0.0026266586690119587,
+      "loss": 1.1686,
+      "step": 2984
+    },
+    {
+      "epoch": 0.22959772325205755,
+      "learning_rate": 0.0026264193442361704,
+      "loss": 0.9695,
+      "step": 2985
+    },
+    {
+      "epoch": 0.22967464041227598,
+      "learning_rate": 0.0026261799536875827,
+      "loss": 1.4658,
+      "step": 2986
+    },
+    {
+      "epoch": 0.22975155757249444,
+      "learning_rate": 0.0026259404973801737,
+      "loss": 0.9816,
+      "step": 2987
+    },
+    {
+      "epoch": 0.22982847473271287,
+      "learning_rate": 0.0026257009753279255,
+      "loss": 1.3082,
+      "step": 2988
+    },
+    {
+      "epoch": 0.22990539189293133,
+      "learning_rate": 0.0026254613875448247,
+      "loss": 1.3418,
+      "step": 2989
+    },
+    {
+      "epoch": 0.22998230905314976,
+      "learning_rate": 0.0026252217340448606,
+      "loss": 0.9574,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2300592262133682,
+      "learning_rate": 0.002624982014842026,
+      "loss": 1.0907,
+      "step": 2991
+    },
+    {
+      "epoch": 0.23013614337358665,
+      "learning_rate": 0.0026247422299503194,
+      "loss": 1.2523,
+      "step": 2992
+    },
+    {
+      "epoch": 0.23021306053380508,
+      "learning_rate": 0.002624502379383742,
+      "loss": 1.2739,
+      "step": 2993
+    },
+    {
+      "epoch": 0.23028997769402354,
+      "learning_rate": 0.002624262463156299,
+      "loss": 1.037,
+      "step": 2994
+    },
+    {
+      "epoch": 0.23036689485424197,
+      "learning_rate": 0.002624022481281999,
+      "loss": 1.1786,
+      "step": 2995
+    },
+    {
+      "epoch": 0.23044381201446043,
+      "learning_rate": 0.0026237824337748544,
+      "loss": 0.9719,
+      "step": 2996
+    },
+    {
+      "epoch": 0.23052072917467886,
+      "learning_rate": 0.0026235423206488824,
+      "loss": 1.4303,
+      "step": 2997
+    },
+    {
+      "epoch": 0.23059764633489732,
+      "learning_rate": 0.0026233021419181037,
+      "loss": 0.7923,
+      "step": 2998
+    },
+    {
+      "epoch": 0.23067456349511575,
+      "learning_rate": 0.002623061897596542,
+      "loss": 1.0055,
+      "step": 2999
+    },
+    {
+      "epoch": 0.2307514806553342,
+      "learning_rate": 0.002622821587698226,
+      "loss": 1.4625,
+      "step": 3000
+    },
+    {
+      "epoch": 0.23082839781555264,
+      "learning_rate": 0.002622581212237187,
+      "loss": 1.0458,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2309053149757711,
+      "learning_rate": 0.002622340771227461,
+      "loss": 1.2414,
+      "step": 3002
+    },
+    {
+      "epoch": 0.23098223213598953,
+      "learning_rate": 0.002622100264683088,
+      "loss": 1.251,
+      "step": 3003
+    },
+    {
+      "epoch": 0.231059149296208,
+      "learning_rate": 0.0026218596926181107,
+      "loss": 0.9071,
+      "step": 3004
+    },
+    {
+      "epoch": 0.23113606645642643,
+      "learning_rate": 0.002621619055046578,
+      "loss": 0.8977,
+      "step": 3005
+    },
+    {
+      "epoch": 0.23121298361664488,
+      "learning_rate": 0.0026213783519825383,
+      "loss": 1.1652,
+      "step": 3006
+    },
+    {
+      "epoch": 0.23128990077686332,
+      "learning_rate": 0.0026211375834400484,
+      "loss": 1.7237,
+      "step": 3007
+    },
+    {
+      "epoch": 0.23136681793708178,
+      "learning_rate": 0.0026208967494331667,
+      "loss": 0.8388,
+      "step": 3008
+    },
+    {
+      "epoch": 0.2314437350973002,
+      "learning_rate": 0.002620655849975956,
+      "loss": 1.2687,
+      "step": 3009
+    },
+    {
+      "epoch": 0.23152065225751864,
+      "learning_rate": 0.002620414885082482,
+      "loss": 1.6059,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2315975694177371,
+      "learning_rate": 0.002620173854766815,
+      "loss": 1.1278,
+      "step": 3011
+    },
+    {
+      "epoch": 0.23167448657795553,
+      "learning_rate": 0.002619932759043029,
+      "loss": 1.1644,
+      "step": 3012
+    },
+    {
+      "epoch": 0.231751403738174,
+      "learning_rate": 0.002619691597925203,
+      "loss": 1.0764,
+      "step": 3013
+    },
+    {
+      "epoch": 0.23182832089839242,
+      "learning_rate": 0.0026194503714274163,
+      "loss": 1.094,
+      "step": 3014
+    },
+    {
+      "epoch": 0.23190523805861088,
+      "learning_rate": 0.0026192090795637566,
+      "loss": 1.5712,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2319821552188293,
+      "learning_rate": 0.002618967722348312,
+      "loss": 1.2202,
+      "step": 3016
+    },
+    {
+      "epoch": 0.23205907237904777,
+      "learning_rate": 0.0026187262997951766,
+      "loss": 0.9639,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2321359895392662,
+      "learning_rate": 0.002618484811918446,
+      "loss": 1.2411,
+      "step": 3018
+    },
+    {
+      "epoch": 0.23221290669948466,
+      "learning_rate": 0.002618243258732222,
+      "loss": 1.0966,
+      "step": 3019
+    },
+    {
+      "epoch": 0.2322898238597031,
+      "learning_rate": 0.002618001640250608,
+      "loss": 0.8387,
+      "step": 3020
+    },
+    {
+      "epoch": 0.23236674101992155,
+      "learning_rate": 0.0026177599564877136,
+      "loss": 1.5191,
+      "step": 3021
+    },
+    {
+      "epoch": 0.23244365818013998,
+      "learning_rate": 0.0026175182074576506,
+      "loss": 1.2134,
+      "step": 3022
+    },
+    {
+      "epoch": 0.23252057534035844,
+      "learning_rate": 0.0026172763931745346,
+      "loss": 0.8338,
+      "step": 3023
+    },
+    {
+      "epoch": 0.23259749250057687,
+      "learning_rate": 0.0026170345136524854,
+      "loss": 1.2159,
+      "step": 3024
+    },
+    {
+      "epoch": 0.23267440966079533,
+      "learning_rate": 0.002616792568905627,
+      "loss": 1.2989,
+      "step": 3025
+    },
+    {
+      "epoch": 0.23275132682101377,
+      "learning_rate": 0.0026165505589480866,
+      "loss": 0.9664,
+      "step": 3026
+    },
+    {
+      "epoch": 0.23282824398123222,
+      "learning_rate": 0.0026163084837939957,
+      "loss": 1.0549,
+      "step": 3027
+    },
+    {
+      "epoch": 0.23290516114145066,
+      "learning_rate": 0.002616066343457489,
+      "loss": 1.5709,
+      "step": 3028
+    },
+    {
+      "epoch": 0.23298207830166912,
+      "learning_rate": 0.0026158241379527045,
+      "loss": 1.1979,
+      "step": 3029
+    },
+    {
+      "epoch": 0.23305899546188755,
+      "learning_rate": 0.0026155818672937868,
+      "loss": 1.2932,
+      "step": 3030
+    },
+    {
+      "epoch": 0.23313591262210598,
+      "learning_rate": 0.0026153395314948806,
+      "loss": 0.9649,
+      "step": 3031
+    },
+    {
+      "epoch": 0.23321282978232444,
+      "learning_rate": 0.002615097130570137,
+      "loss": 0.8093,
+      "step": 3032
+    },
+    {
+      "epoch": 0.23328974694254287,
+      "learning_rate": 0.00261485466453371,
+      "loss": 1.4998,
+      "step": 3033
+    },
+    {
+      "epoch": 0.23336666410276133,
+      "learning_rate": 0.0026146121333997566,
+      "loss": 1.4707,
+      "step": 3034
+    },
+    {
+      "epoch": 0.23344358126297976,
+      "learning_rate": 0.0026143695371824397,
+      "loss": 1.1959,
+      "step": 3035
+    },
+    {
+      "epoch": 0.23352049842319822,
+      "learning_rate": 0.002614126875895924,
+      "loss": 1.1982,
+      "step": 3036
+    },
+    {
+      "epoch": 0.23359741558341665,
+      "learning_rate": 0.002613884149554379,
+      "loss": 1.1764,
+      "step": 3037
+    },
+    {
+      "epoch": 0.2336743327436351,
+      "learning_rate": 0.0026136413581719776,
+      "loss": 1.2159,
+      "step": 3038
+    },
+    {
+      "epoch": 0.23375124990385354,
+      "learning_rate": 0.0026133985017628967,
+      "loss": 1.1605,
+      "step": 3039
+    },
+    {
+      "epoch": 0.233828167064072,
+      "learning_rate": 0.002613155580341317,
+      "loss": 1.3456,
+      "step": 3040
+    },
+    {
+      "epoch": 0.23390508422429043,
+      "learning_rate": 0.0026129125939214222,
+      "loss": 1.174,
+      "step": 3041
+    },
+    {
+      "epoch": 0.2339820013845089,
+      "learning_rate": 0.0026126695425174023,
+      "loss": 0.9774,
+      "step": 3042
+    },
+    {
+      "epoch": 0.23405891854472732,
+      "learning_rate": 0.0026124264261434475,
+      "loss": 1.4991,
+      "step": 3043
+    },
+    {
+      "epoch": 0.23413583570494578,
+      "learning_rate": 0.0026121832448137545,
+      "loss": 1.1045,
+      "step": 3044
+    },
+    {
+      "epoch": 0.23421275286516421,
+      "learning_rate": 0.002611939998542523,
+      "loss": 0.762,
+      "step": 3045
+    },
+    {
+      "epoch": 0.23428967002538267,
+      "learning_rate": 0.0026116966873439553,
+      "loss": 1.2953,
+      "step": 3046
+    },
+    {
+      "epoch": 0.2343665871856011,
+      "learning_rate": 0.00261145331123226,
+      "loss": 1.1405,
+      "step": 3047
+    },
+    {
+      "epoch": 0.23444350434581956,
+      "learning_rate": 0.0026112098702216475,
+      "loss": 1.0199,
+      "step": 3048
+    },
+    {
+      "epoch": 0.234520421506038,
+      "learning_rate": 0.002610966364326333,
+      "loss": 0.9345,
+      "step": 3049
+    },
+    {
+      "epoch": 0.23459733866625646,
+      "learning_rate": 0.002610722793560534,
+      "loss": 1.4446,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2346742558264749,
+      "learning_rate": 0.002610479157938474,
+      "loss": 0.9241,
+      "step": 3051
+    },
+    {
+      "epoch": 0.23475117298669332,
+      "learning_rate": 0.0026102354574743785,
+      "loss": 1.2471,
+      "step": 3052
+    },
+    {
+      "epoch": 0.23482809014691178,
+      "learning_rate": 0.002609991692182478,
+      "loss": 1.2094,
+      "step": 3053
+    },
+    {
+      "epoch": 0.2349050073071302,
+      "learning_rate": 0.002609747862077005,
+      "loss": 1.0292,
+      "step": 3054
+    },
+    {
+      "epoch": 0.23498192446734867,
+      "learning_rate": 0.0026095039671721982,
+      "loss": 1.069,
+      "step": 3055
+    },
+    {
+      "epoch": 0.2350588416275671,
+      "learning_rate": 0.002609260007482298,
+      "loss": 0.7545,
+      "step": 3056
+    },
+    {
+      "epoch": 0.23513575878778556,
+      "learning_rate": 0.0026090159830215506,
+      "loss": 1.3558,
+      "step": 3057
+    },
+    {
+      "epoch": 0.235212675948004,
+      "learning_rate": 0.0026087718938042036,
+      "loss": 1.3914,
+      "step": 3058
+    },
+    {
+      "epoch": 0.23528959310822245,
+      "learning_rate": 0.0026085277398445105,
+      "loss": 1.0881,
+      "step": 3059
+    },
+    {
+      "epoch": 0.23536651026844088,
+      "learning_rate": 0.0026082835211567275,
+      "loss": 1.2548,
+      "step": 3060
+    },
+    {
+      "epoch": 0.23544342742865934,
+      "learning_rate": 0.0026080392377551146,
+      "loss": 1.3836,
+      "step": 3061
+    },
+    {
+      "epoch": 0.23552034458887777,
+      "learning_rate": 0.0026077948896539357,
+      "loss": 1.0861,
+      "step": 3062
+    },
+    {
+      "epoch": 0.23559726174909623,
+      "learning_rate": 0.002607550476867459,
+      "loss": 1.1892,
+      "step": 3063
+    },
+    {
+      "epoch": 0.23567417890931466,
+      "learning_rate": 0.0026073059994099557,
+      "loss": 1.0239,
+      "step": 3064
+    },
+    {
+      "epoch": 0.23575109606953312,
+      "learning_rate": 0.0026070614572957004,
+      "loss": 0.9831,
+      "step": 3065
+    },
+    {
+      "epoch": 0.23582801322975155,
+      "learning_rate": 0.002606816850538973,
+      "loss": 1.5357,
+      "step": 3066
+    },
+    {
+      "epoch": 0.23590493038997,
+      "learning_rate": 0.0026065721791540567,
+      "loss": 1.1955,
+      "step": 3067
+    },
+    {
+      "epoch": 0.23598184755018844,
+      "learning_rate": 0.002606327443155237,
+      "loss": 1.1478,
+      "step": 3068
+    },
+    {
+      "epoch": 0.2360587647104069,
+      "learning_rate": 0.0026060826425568055,
+      "loss": 1.3183,
+      "step": 3069
+    },
+    {
+      "epoch": 0.23613568187062534,
+      "learning_rate": 0.002605837777373056,
+      "loss": 1.1357,
+      "step": 3070
+    },
+    {
+      "epoch": 0.2362125990308438,
+      "learning_rate": 0.0026055928476182853,
+      "loss": 0.9016,
+      "step": 3071
+    },
+    {
+      "epoch": 0.23628951619106223,
+      "learning_rate": 0.002605347853306797,
+      "loss": 1.3511,
+      "step": 3072
+    },
+    {
+      "epoch": 0.23636643335128066,
+      "learning_rate": 0.0026051027944528947,
+      "loss": 1.056,
+      "step": 3073
+    },
+    {
+      "epoch": 0.23644335051149912,
+      "learning_rate": 0.0026048576710708893,
+      "loss": 1.1846,
+      "step": 3074
+    },
+    {
+      "epoch": 0.23652026767171755,
+      "learning_rate": 0.0026046124831750925,
+      "loss": 1.0967,
+      "step": 3075
+    },
+    {
+      "epoch": 0.236597184831936,
+      "learning_rate": 0.0026043672307798217,
+      "loss": 1.1679,
+      "step": 3076
+    },
+    {
+      "epoch": 0.23667410199215444,
+      "learning_rate": 0.0026041219138993976,
+      "loss": 1.1025,
+      "step": 3077
+    },
+    {
+      "epoch": 0.2367510191523729,
+      "learning_rate": 0.002603876532548144,
+      "loss": 1.0724,
+      "step": 3078
+    },
+    {
+      "epoch": 0.23682793631259133,
+      "learning_rate": 0.002603631086740389,
+      "loss": 1.1619,
+      "step": 3079
+    },
+    {
+      "epoch": 0.2369048534728098,
+      "learning_rate": 0.002603385576490465,
+      "loss": 1.3809,
+      "step": 3080
+    },
+    {
+      "epoch": 0.23698177063302822,
+      "learning_rate": 0.002603140001812707,
+      "loss": 1.0079,
+      "step": 3081
+    },
+    {
+      "epoch": 0.23705868779324668,
+      "learning_rate": 0.002602894362721455,
+      "loss": 1.2764,
+      "step": 3082
+    },
+    {
+      "epoch": 0.2371356049534651,
+      "learning_rate": 0.0026026486592310516,
+      "loss": 1.2921,
+      "step": 3083
+    },
+    {
+      "epoch": 0.23721252211368357,
+      "learning_rate": 0.0026024028913558435,
+      "loss": 1.4317,
+      "step": 3084
+    },
+    {
+      "epoch": 0.237289439273902,
+      "learning_rate": 0.002602157059110182,
+      "loss": 1.2007,
+      "step": 3085
+    },
+    {
+      "epoch": 0.23736635643412046,
+      "learning_rate": 0.0026019111625084214,
+      "loss": 0.9553,
+      "step": 3086
+    },
+    {
+      "epoch": 0.2374432735943389,
+      "learning_rate": 0.0026016652015649192,
+      "loss": 1.2116,
+      "step": 3087
+    },
+    {
+      "epoch": 0.23752019075455735,
+      "learning_rate": 0.002601419176294038,
+      "loss": 1.376,
+      "step": 3088
+    },
+    {
+      "epoch": 0.23759710791477578,
+      "learning_rate": 0.0026011730867101434,
+      "loss": 1.2945,
+      "step": 3089
+    },
+    {
+      "epoch": 0.23767402507499424,
+      "learning_rate": 0.002600926932827604,
+      "loss": 0.9543,
+      "step": 3090
+    },
+    {
+      "epoch": 0.23775094223521268,
+      "learning_rate": 0.0026006807146607943,
+      "loss": 1.1269,
+      "step": 3091
+    },
+    {
+      "epoch": 0.2378278593954311,
+      "learning_rate": 0.0026004344322240903,
+      "loss": 1.3008,
+      "step": 3092
+    },
+    {
+      "epoch": 0.23790477655564957,
+      "learning_rate": 0.002600188085531873,
+      "loss": 0.9291,
+      "step": 3093
+    },
+    {
+      "epoch": 0.237981693715868,
+      "learning_rate": 0.002599941674598527,
+      "loss": 0.887,
+      "step": 3094
+    },
+    {
+      "epoch": 0.23805861087608646,
+      "learning_rate": 0.00259969519943844,
+      "loss": 0.9833,
+      "step": 3095
+    },
+    {
+      "epoch": 0.2381355280363049,
+      "learning_rate": 0.0025994486600660043,
+      "loss": 1.3557,
+      "step": 3096
+    },
+    {
+      "epoch": 0.23821244519652335,
+      "learning_rate": 0.002599202056495615,
+      "loss": 1.0028,
+      "step": 3097
+    },
+    {
+      "epoch": 0.23828936235674178,
+      "learning_rate": 0.002598955388741673,
+      "loss": 1.2178,
+      "step": 3098
+    },
+    {
+      "epoch": 0.23836627951696024,
+      "learning_rate": 0.0025987086568185803,
+      "loss": 0.8754,
+      "step": 3099
+    },
+    {
+      "epoch": 0.23844319667717867,
+      "learning_rate": 0.0025984618607407437,
+      "loss": 1.0655,
+      "step": 3100
+    },
+    {
+      "epoch": 0.23852011383739713,
+      "learning_rate": 0.0025982150005225742,
+      "loss": 1.1286,
+      "step": 3101
+    },
+    {
+      "epoch": 0.23859703099761556,
+      "learning_rate": 0.002597968076178487,
+      "loss": 1.0951,
+      "step": 3102
+    },
+    {
+      "epoch": 0.23867394815783402,
+      "learning_rate": 0.002597721087722899,
+      "loss": 1.1224,
+      "step": 3103
+    },
+    {
+      "epoch": 0.23875086531805245,
+      "learning_rate": 0.002597474035170233,
+      "loss": 1.2548,
+      "step": 3104
+    },
+    {
+      "epoch": 0.2388277824782709,
+      "learning_rate": 0.002597226918534914,
+      "loss": 1.2096,
+      "step": 3105
+    },
+    {
+      "epoch": 0.23890469963848934,
+      "learning_rate": 0.002596979737831372,
+      "loss": 1.1102,
+      "step": 3106
+    },
+    {
+      "epoch": 0.2389816167987078,
+      "learning_rate": 0.0025967324930740395,
+      "loss": 1.0185,
+      "step": 3107
+    },
+    {
+      "epoch": 0.23905853395892623,
+      "learning_rate": 0.002596485184277354,
+      "loss": 1.2696,
+      "step": 3108
+    },
+    {
+      "epoch": 0.2391354511191447,
+      "learning_rate": 0.002596237811455756,
+      "loss": 1.011,
+      "step": 3109
+    },
+    {
+      "epoch": 0.23921236827936312,
+      "learning_rate": 0.002595990374623689,
+      "loss": 1.3151,
+      "step": 3110
+    },
+    {
+      "epoch": 0.23928928543958158,
+      "learning_rate": 0.0025957428737956026,
+      "loss": 1.1528,
+      "step": 3111
+    },
+    {
+      "epoch": 0.23936620259980002,
+      "learning_rate": 0.0025954953089859477,
+      "loss": 0.8947,
+      "step": 3112
+    },
+    {
+      "epoch": 0.23944311976001845,
+      "learning_rate": 0.00259524768020918,
+      "loss": 1.2746,
+      "step": 3113
+    },
+    {
+      "epoch": 0.2395200369202369,
+      "learning_rate": 0.0025949999874797587,
+      "loss": 1.4013,
+      "step": 3114
+    },
+    {
+      "epoch": 0.23959695408045534,
+      "learning_rate": 0.002594752230812147,
+      "loss": 1.1671,
+      "step": 3115
+    },
+    {
+      "epoch": 0.2396738712406738,
+      "learning_rate": 0.0025945044102208114,
+      "loss": 0.9244,
+      "step": 3116
+    },
+    {
+      "epoch": 0.23975078840089223,
+      "learning_rate": 0.002594256525720223,
+      "loss": 1.1692,
+      "step": 3117
+    },
+    {
+      "epoch": 0.2398277055611107,
+      "learning_rate": 0.002594008577324855,
+      "loss": 1.4609,
+      "step": 3118
+    },
+    {
+      "epoch": 0.23990462272132912,
+      "learning_rate": 0.0025937605650491868,
+      "loss": 1.0407,
+      "step": 3119
+    },
+    {
+      "epoch": 0.23998153988154758,
+      "learning_rate": 0.0025935124889076987,
+      "loss": 0.9881,
+      "step": 3120
+    },
+    {
+      "epoch": 0.240058457041766,
+      "learning_rate": 0.0025932643489148775,
+      "loss": 1.289,
+      "step": 3121
+    },
+    {
+      "epoch": 0.24013537420198447,
+      "learning_rate": 0.002593016145085211,
+      "loss": 0.8986,
+      "step": 3122
+    },
+    {
+      "epoch": 0.2402122913622029,
+      "learning_rate": 0.0025927678774331927,
+      "loss": 1.3161,
+      "step": 3123
+    },
+    {
+      "epoch": 0.24028920852242136,
+      "learning_rate": 0.0025925195459733197,
+      "loss": 1.1897,
+      "step": 3124
+    },
+    {
+      "epoch": 0.2403661256826398,
+      "learning_rate": 0.0025922711507200915,
+      "loss": 1.3164,
+      "step": 3125
+    },
+    {
+      "epoch": 0.24044304284285825,
+      "learning_rate": 0.002592022691688012,
+      "loss": 1.0286,
+      "step": 3126
+    },
+    {
+      "epoch": 0.24051996000307668,
+      "learning_rate": 0.0025917741688915898,
+      "loss": 0.7483,
+      "step": 3127
+    },
+    {
+      "epoch": 0.24059687716329514,
+      "learning_rate": 0.002591525582345336,
+      "loss": 1.1146,
+      "step": 3128
+    },
+    {
+      "epoch": 0.24067379432351357,
+      "learning_rate": 0.002591276932063766,
+      "loss": 1.4364,
+      "step": 3129
+    },
+    {
+      "epoch": 0.24075071148373203,
+      "learning_rate": 0.002591028218061399,
+      "loss": 0.9923,
+      "step": 3130
+    },
+    {
+      "epoch": 0.24082762864395046,
+      "learning_rate": 0.0025907794403527565,
+      "loss": 0.8849,
+      "step": 3131
+    },
+    {
+      "epoch": 0.24090454580416892,
+      "learning_rate": 0.0025905305989523666,
+      "loss": 1.4039,
+      "step": 3132
+    },
+    {
+      "epoch": 0.24098146296438736,
+      "learning_rate": 0.0025902816938747574,
+      "loss": 1.1842,
+      "step": 3133
+    },
+    {
+      "epoch": 0.2410583801246058,
+      "learning_rate": 0.002590032725134465,
+      "loss": 1.0936,
+      "step": 3134
+    },
+    {
+      "epoch": 0.24113529728482425,
+      "learning_rate": 0.0025897836927460246,
+      "loss": 1.1739,
+      "step": 3135
+    },
+    {
+      "epoch": 0.24121221444504268,
+      "learning_rate": 0.00258953459672398,
+      "loss": 1.2149,
+      "step": 3136
+    },
+    {
+      "epoch": 0.24128913160526114,
+      "learning_rate": 0.002589285437082874,
+      "loss": 1.1202,
+      "step": 3137
+    },
+    {
+      "epoch": 0.24136604876547957,
+      "learning_rate": 0.0025890362138372555,
+      "loss": 1.5392,
+      "step": 3138
+    },
+    {
+      "epoch": 0.24144296592569803,
+      "learning_rate": 0.002588786927001678,
+      "loss": 1.0236,
+      "step": 3139
+    },
+    {
+      "epoch": 0.24151988308591646,
+      "learning_rate": 0.0025885375765906976,
+      "loss": 1.0635,
+      "step": 3140
+    },
+    {
+      "epoch": 0.24159680024613492,
+      "learning_rate": 0.0025882881626188734,
+      "loss": 0.8874,
+      "step": 3141
+    },
+    {
+      "epoch": 0.24167371740635335,
+      "learning_rate": 0.002588038685100769,
+      "loss": 0.9573,
+      "step": 3142
+    },
+    {
+      "epoch": 0.2417506345665718,
+      "learning_rate": 0.0025877891440509527,
+      "loss": 1.0899,
+      "step": 3143
+    },
+    {
+      "epoch": 0.24182755172679024,
+      "learning_rate": 0.0025875395394839933,
+      "loss": 1.8011,
+      "step": 3144
+    },
+    {
+      "epoch": 0.2419044688870087,
+      "learning_rate": 0.002587289871414467,
+      "loss": 1.4425,
+      "step": 3145
+    },
+    {
+      "epoch": 0.24198138604722713,
+      "learning_rate": 0.002587040139856953,
+      "loss": 1.1479,
+      "step": 3146
+    },
+    {
+      "epoch": 0.2420583032074456,
+      "learning_rate": 0.0025867903448260317,
+      "loss": 1.3082,
+      "step": 3147
+    },
+    {
+      "epoch": 0.24213522036766402,
+      "learning_rate": 0.0025865404863362894,
+      "loss": 1.0346,
+      "step": 3148
+    },
+    {
+      "epoch": 0.24221213752788248,
+      "learning_rate": 0.002586290564402316,
+      "loss": 1.2068,
+      "step": 3149
+    },
+    {
+      "epoch": 0.2422890546881009,
+      "learning_rate": 0.002586040579038704,
+      "loss": 1.2679,
+      "step": 3150
+    },
+    {
+      "epoch": 0.24236597184831937,
+      "learning_rate": 0.0025857905302600517,
+      "loss": 1.2732,
+      "step": 3151
+    },
+    {
+      "epoch": 0.2424428890085378,
+      "learning_rate": 0.0025855404180809578,
+      "loss": 1.1605,
+      "step": 3152
+    },
+    {
+      "epoch": 0.24251980616875624,
+      "learning_rate": 0.002585290242516028,
+      "loss": 1.1359,
+      "step": 3153
+    },
+    {
+      "epoch": 0.2425967233289747,
+      "learning_rate": 0.0025850400035798697,
+      "loss": 1.3522,
+      "step": 3154
+    },
+    {
+      "epoch": 0.24267364048919313,
+      "learning_rate": 0.0025847897012870945,
+      "loss": 1.2395,
+      "step": 3155
+    },
+    {
+      "epoch": 0.24275055764941159,
+      "learning_rate": 0.0025845393356523187,
+      "loss": 1.394,
+      "step": 3156
+    },
+    {
+      "epoch": 0.24282747480963002,
+      "learning_rate": 0.0025842889066901606,
+      "loss": 1.3979,
+      "step": 3157
+    },
+    {
+      "epoch": 0.24290439196984848,
+      "learning_rate": 0.0025840384144152433,
+      "loss": 1.0034,
+      "step": 3158
+    },
+    {
+      "epoch": 0.2429813091300669,
+      "learning_rate": 0.002583787858842193,
+      "loss": 1.0498,
+      "step": 3159
+    },
+    {
+      "epoch": 0.24305822629028537,
+      "learning_rate": 0.00258353723998564,
+      "loss": 1.3022,
+      "step": 3160
+    },
+    {
+      "epoch": 0.2431351434505038,
+      "learning_rate": 0.0025832865578602187,
+      "loss": 1.4339,
+      "step": 3161
+    },
+    {
+      "epoch": 0.24321206061072226,
+      "learning_rate": 0.0025830358124805664,
+      "loss": 1.281,
+      "step": 3162
+    },
+    {
+      "epoch": 0.2432889777709407,
+      "learning_rate": 0.002582785003861324,
+      "loss": 1.1118,
+      "step": 3163
+    },
+    {
+      "epoch": 0.24336589493115915,
+      "learning_rate": 0.002582534132017137,
+      "loss": 0.7518,
+      "step": 3164
+    },
+    {
+      "epoch": 0.24344281209137758,
+      "learning_rate": 0.0025822831969626535,
+      "loss": 1.0684,
+      "step": 3165
+    },
+    {
+      "epoch": 0.24351972925159604,
+      "learning_rate": 0.002582032198712526,
+      "loss": 1.2551,
+      "step": 3166
+    },
+    {
+      "epoch": 0.24359664641181447,
+      "learning_rate": 0.0025817811372814115,
+      "loss": 0.9243,
+      "step": 3167
+    },
+    {
+      "epoch": 0.24367356357203293,
+      "learning_rate": 0.0025815300126839683,
+      "loss": 1.2287,
+      "step": 3168
+    },
+    {
+      "epoch": 0.24375048073225136,
+      "learning_rate": 0.002581278824934861,
+      "loss": 1.2283,
+      "step": 3169
+    },
+    {
+      "epoch": 0.24382739789246982,
+      "learning_rate": 0.002581027574048756,
+      "loss": 1.0643,
+      "step": 3170
+    },
+    {
+      "epoch": 0.24390431505268825,
+      "learning_rate": 0.0025807762600403247,
+      "loss": 1.001,
+      "step": 3171
+    },
+    {
+      "epoch": 0.2439812322129067,
+      "learning_rate": 0.0025805248829242405,
+      "loss": 1.1246,
+      "step": 3172
+    },
+    {
+      "epoch": 0.24405814937312514,
+      "learning_rate": 0.002580273442715183,
+      "loss": 1.231,
+      "step": 3173
+    },
+    {
+      "epoch": 0.24413506653334358,
+      "learning_rate": 0.0025800219394278333,
+      "loss": 1.3881,
+      "step": 3174
+    },
+    {
+      "epoch": 0.24421198369356203,
+      "learning_rate": 0.0025797703730768766,
+      "loss": 1.21,
+      "step": 3175
+    },
+    {
+      "epoch": 0.24428890085378047,
+      "learning_rate": 0.0025795187436770026,
+      "loss": 1.2494,
+      "step": 3176
+    },
+    {
+      "epoch": 0.24436581801399893,
+      "learning_rate": 0.002579267051242904,
+      "loss": 1.3317,
+      "step": 3177
+    },
+    {
+      "epoch": 0.24444273517421736,
+      "learning_rate": 0.0025790152957892777,
+      "loss": 1.2651,
+      "step": 3178
+    },
+    {
+      "epoch": 0.24451965233443582,
+      "learning_rate": 0.0025787634773308237,
+      "loss": 1.108,
+      "step": 3179
+    },
+    {
+      "epoch": 0.24459656949465425,
+      "learning_rate": 0.0025785115958822462,
+      "loss": 0.9391,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2446734866548727,
+      "learning_rate": 0.002578259651458252,
+      "loss": 1.1423,
+      "step": 3181
+    },
+    {
+      "epoch": 0.24475040381509114,
+      "learning_rate": 0.002578007644073554,
+      "loss": 1.1358,
+      "step": 3182
+    },
+    {
+      "epoch": 0.2448273209753096,
+      "learning_rate": 0.0025777555737428654,
+      "loss": 1.2116,
+      "step": 3183
+    },
+    {
+      "epoch": 0.24490423813552803,
+      "learning_rate": 0.0025775034404809057,
+      "loss": 1.3328,
+      "step": 3184
+    },
+    {
+      "epoch": 0.2449811552957465,
+      "learning_rate": 0.0025772512443023973,
+      "loss": 0.993,
+      "step": 3185
+    },
+    {
+      "epoch": 0.24505807245596492,
+      "learning_rate": 0.0025769989852220663,
+      "loss": 0.756,
+      "step": 3186
+    },
+    {
+      "epoch": 0.24513498961618338,
+      "learning_rate": 0.0025767466632546417,
+      "loss": 1.2895,
+      "step": 3187
+    },
+    {
+      "epoch": 0.2452119067764018,
+      "learning_rate": 0.002576494278414858,
+      "loss": 1.0305,
+      "step": 3188
+    },
+    {
+      "epoch": 0.24528882393662027,
+      "learning_rate": 0.0025762418307174505,
+      "loss": 1.4926,
+      "step": 3189
+    },
+    {
+      "epoch": 0.2453657410968387,
+      "learning_rate": 0.0025759893201771616,
+      "loss": 1.1996,
+      "step": 3190
+    },
+    {
+      "epoch": 0.24544265825705716,
+      "learning_rate": 0.0025757367468087346,
+      "loss": 1.2316,
+      "step": 3191
+    },
+    {
+      "epoch": 0.2455195754172756,
+      "learning_rate": 0.002575484110626918,
+      "loss": 0.8187,
+      "step": 3192
+    },
+    {
+      "epoch": 0.24559649257749405,
+      "learning_rate": 0.0025752314116464633,
+      "loss": 1.4764,
+      "step": 3193
+    },
+    {
+      "epoch": 0.24567340973771248,
+      "learning_rate": 0.002574978649882126,
+      "loss": 1.0287,
+      "step": 3194
+    },
+    {
+      "epoch": 0.24575032689793092,
+      "learning_rate": 0.0025747258253486643,
+      "loss": 1.2878,
+      "step": 3195
+    },
+    {
+      "epoch": 0.24582724405814937,
+      "learning_rate": 0.0025744729380608425,
+      "loss": 1.1815,
+      "step": 3196
+    },
+    {
+      "epoch": 0.2459041612183678,
+      "learning_rate": 0.002574219988033426,
+      "loss": 0.8401,
+      "step": 3197
+    },
+    {
+      "epoch": 0.24598107837858627,
+      "learning_rate": 0.0025739669752811844,
+      "loss": 1.1133,
+      "step": 3198
+    },
+    {
+      "epoch": 0.2460579955388047,
+      "learning_rate": 0.0025737138998188917,
+      "loss": 1.265,
+      "step": 3199
+    },
+    {
+      "epoch": 0.24613491269902316,
+      "learning_rate": 0.002573460761661326,
+      "loss": 1.5385,
+      "step": 3200
+    },
+    {
+      "epoch": 0.2462118298592416,
+      "learning_rate": 0.0025732075608232677,
+      "loss": 0.9086,
+      "step": 3201
+    },
+    {
+      "epoch": 0.24628874701946005,
+      "learning_rate": 0.002572954297319501,
+      "loss": 0.8488,
+      "step": 3202
+    },
+    {
+      "epoch": 0.24636566417967848,
+      "learning_rate": 0.002572700971164815,
+      "loss": 0.7496,
+      "step": 3203
+    },
+    {
+      "epoch": 0.24644258133989694,
+      "learning_rate": 0.0025724475823740013,
+      "loss": 1.1925,
+      "step": 3204
+    },
+    {
+      "epoch": 0.24651949850011537,
+      "learning_rate": 0.0025721941309618554,
+      "loss": 1.2115,
+      "step": 3205
+    },
+    {
+      "epoch": 0.24659641566033383,
+      "learning_rate": 0.0025719406169431767,
+      "loss": 1.2288,
+      "step": 3206
+    },
+    {
+      "epoch": 0.24667333282055226,
+      "learning_rate": 0.0025716870403327684,
+      "loss": 1.0751,
+      "step": 3207
+    },
+    {
+      "epoch": 0.24675024998077072,
+      "learning_rate": 0.0025714334011454373,
+      "loss": 1.1344,
+      "step": 3208
+    },
+    {
+      "epoch": 0.24682716714098915,
+      "learning_rate": 0.0025711796993959928,
+      "loss": 1.2588,
+      "step": 3209
+    },
+    {
+      "epoch": 0.2469040843012076,
+      "learning_rate": 0.0025709259350992497,
+      "loss": 1.3606,
+      "step": 3210
+    },
+    {
+      "epoch": 0.24698100146142604,
+      "learning_rate": 0.0025706721082700244,
+      "loss": 1.2679,
+      "step": 3211
+    },
+    {
+      "epoch": 0.2470579186216445,
+      "learning_rate": 0.0025704182189231397,
+      "loss": 1.1779,
+      "step": 3212
+    },
+    {
+      "epoch": 0.24713483578186293,
+      "learning_rate": 0.002570164267073419,
+      "loss": 1.2455,
+      "step": 3213
+    },
+    {
+      "epoch": 0.2472117529420814,
+      "learning_rate": 0.002569910252735692,
+      "loss": 1.1148,
+      "step": 3214
+    },
+    {
+      "epoch": 0.24728867010229982,
+      "learning_rate": 0.00256965617592479,
+      "loss": 1.2124,
+      "step": 3215
+    },
+    {
+      "epoch": 0.24736558726251826,
+      "learning_rate": 0.002569402036655549,
+      "loss": 1.5176,
+      "step": 3216
+    },
+    {
+      "epoch": 0.24744250442273671,
+      "learning_rate": 0.002569147834942809,
+      "loss": 1.4819,
+      "step": 3217
+    },
+    {
+      "epoch": 0.24751942158295515,
+      "learning_rate": 0.0025688935708014123,
+      "loss": 1.2158,
+      "step": 3218
+    },
+    {
+      "epoch": 0.2475963387431736,
+      "learning_rate": 0.0025686392442462063,
+      "loss": 1.4652,
+      "step": 3219
+    },
+    {
+      "epoch": 0.24767325590339204,
+      "learning_rate": 0.0025683848552920407,
+      "loss": 1.0987,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2477501730636105,
+      "learning_rate": 0.0025681304039537702,
+      "loss": 1.4371,
+      "step": 3221
+    },
+    {
+      "epoch": 0.24782709022382893,
+      "learning_rate": 0.0025678758902462524,
+      "loss": 1.0194,
+      "step": 3222
+    },
+    {
+      "epoch": 0.2479040073840474,
+      "learning_rate": 0.002567621314184348,
+      "loss": 1.3744,
+      "step": 3223
+    },
+    {
+      "epoch": 0.24798092454426582,
+      "learning_rate": 0.0025673666757829233,
+      "loss": 0.9836,
+      "step": 3224
+    },
+    {
+      "epoch": 0.24805784170448428,
+      "learning_rate": 0.002567111975056845,
+      "loss": 1.145,
+      "step": 3225
+    },
+    {
+      "epoch": 0.2481347588647027,
+      "learning_rate": 0.0025668572120209872,
+      "loss": 1.5748,
+      "step": 3226
+    },
+    {
+      "epoch": 0.24821167602492117,
+      "learning_rate": 0.0025666023866902244,
+      "loss": 1.2518,
+      "step": 3227
+    },
+    {
+      "epoch": 0.2482885931851396,
+      "learning_rate": 0.0025663474990794367,
+      "loss": 1.1501,
+      "step": 3228
+    },
+    {
+      "epoch": 0.24836551034535806,
+      "learning_rate": 0.0025660925492035076,
+      "loss": 1.4456,
+      "step": 3229
+    },
+    {
+      "epoch": 0.2484424275055765,
+      "learning_rate": 0.0025658375370773232,
+      "loss": 1.1321,
+      "step": 3230
+    },
+    {
+      "epoch": 0.24851934466579495,
+      "learning_rate": 0.002565582462715774,
+      "loss": 1.1396,
+      "step": 3231
+    },
+    {
+      "epoch": 0.24859626182601338,
+      "learning_rate": 0.0025653273261337546,
+      "loss": 1.3533,
+      "step": 3232
+    },
+    {
+      "epoch": 0.24867317898623184,
+      "learning_rate": 0.002565072127346162,
+      "loss": 0.9253,
+      "step": 3233
+    },
+    {
+      "epoch": 0.24875009614645027,
+      "learning_rate": 0.002564816866367898,
+      "loss": 1.325,
+      "step": 3234
+    },
+    {
+      "epoch": 0.2488270133066687,
+      "learning_rate": 0.002564561543213868,
+      "loss": 0.815,
+      "step": 3235
+    },
+    {
+      "epoch": 0.24890393046688716,
+      "learning_rate": 0.0025643061578989796,
+      "loss": 1.3108,
+      "step": 3236
+    },
+    {
+      "epoch": 0.2489808476271056,
+      "learning_rate": 0.0025640507104381454,
+      "loss": 1.0179,
+      "step": 3237
+    },
+    {
+      "epoch": 0.24905776478732405,
+      "learning_rate": 0.0025637952008462817,
+      "loss": 1.049,
+      "step": 3238
+    },
+    {
+      "epoch": 0.24913468194754249,
+      "learning_rate": 0.002563539629138307,
+      "loss": 1.2534,
+      "step": 3239
+    },
+    {
+      "epoch": 0.24921159910776094,
+      "learning_rate": 0.0025632839953291455,
+      "loss": 0.7798,
+      "step": 3240
+    },
+    {
+      "epoch": 0.24928851626797938,
+      "learning_rate": 0.002563028299433723,
+      "loss": 1.1917,
+      "step": 3241
+    },
+    {
+      "epoch": 0.24936543342819784,
+      "learning_rate": 0.0025627725414669703,
+      "loss": 1.0835,
+      "step": 3242
+    },
+    {
+      "epoch": 0.24944235058841627,
+      "learning_rate": 0.0025625167214438213,
+      "loss": 1.2254,
+      "step": 3243
+    },
+    {
+      "epoch": 0.24951926774863473,
+      "learning_rate": 0.0025622608393792135,
+      "loss": 1.1073,
+      "step": 3244
+    },
+    {
+      "epoch": 0.24959618490885316,
+      "learning_rate": 0.0025620048952880886,
+      "loss": 1.2777,
+      "step": 3245
+    },
+    {
+      "epoch": 0.24967310206907162,
+      "learning_rate": 0.0025617488891853902,
+      "loss": 1.0222,
+      "step": 3246
+    },
+    {
+      "epoch": 0.24975001922929005,
+      "learning_rate": 0.0025614928210860688,
+      "loss": 1.0946,
+      "step": 3247
+    },
+    {
+      "epoch": 0.2498269363895085,
+      "learning_rate": 0.0025612366910050745,
+      "loss": 1.0655,
+      "step": 3248
+    },
+    {
+      "epoch": 0.24990385354972694,
+      "learning_rate": 0.0025609804989573646,
+      "loss": 1.0671,
+      "step": 3249
+    },
+    {
+      "epoch": 0.2499807707099454,
+      "learning_rate": 0.0025607242449578967,
+      "loss": 1.127,
+      "step": 3250
+    },
+    {
+      "epoch": 0.25005768787016386,
+      "learning_rate": 0.002560467929021635,
+      "loss": 1.0283,
+      "step": 3251
+    },
+    {
+      "epoch": 0.25013460503038226,
+      "learning_rate": 0.0025602115511635464,
+      "loss": 1.198,
+      "step": 3252
+    },
+    {
+      "epoch": 0.2502115221906007,
+      "learning_rate": 0.0025599551113985996,
+      "loss": 1.2752,
+      "step": 3253
+    },
+    {
+      "epoch": 0.2502884393508192,
+      "learning_rate": 0.0025596986097417692,
+      "loss": 0.867,
+      "step": 3254
+    },
+    {
+      "epoch": 0.25036535651103764,
+      "learning_rate": 0.0025594420462080334,
+      "loss": 1.1996,
+      "step": 3255
+    },
+    {
+      "epoch": 0.25044227367125604,
+      "learning_rate": 0.0025591854208123713,
+      "loss": 1.1055,
+      "step": 3256
+    },
+    {
+      "epoch": 0.2505191908314745,
+      "learning_rate": 0.0025589287335697694,
+      "loss": 1.3893,
+      "step": 3257
+    },
+    {
+      "epoch": 0.25059610799169296,
+      "learning_rate": 0.0025586719844952147,
+      "loss": 1.2048,
+      "step": 3258
+    },
+    {
+      "epoch": 0.25067302515191137,
+      "learning_rate": 0.0025584151736037,
+      "loss": 1.031,
+      "step": 3259
+    },
+    {
+      "epoch": 0.2507499423121298,
+      "learning_rate": 0.00255815830091022,
+      "loss": 1.6608,
+      "step": 3260
+    },
+    {
+      "epoch": 0.2508268594723483,
+      "learning_rate": 0.0025579013664297737,
+      "loss": 1.3659,
+      "step": 3261
+    },
+    {
+      "epoch": 0.25090377663256674,
+      "learning_rate": 0.0025576443701773644,
+      "loss": 1.1875,
+      "step": 3262
+    },
+    {
+      "epoch": 0.25098069379278515,
+      "learning_rate": 0.002557387312167998,
+      "loss": 1.1799,
+      "step": 3263
+    },
+    {
+      "epoch": 0.2510576109530036,
+      "learning_rate": 0.0025571301924166845,
+      "loss": 1.3114,
+      "step": 3264
+    },
+    {
+      "epoch": 0.25113452811322207,
+      "learning_rate": 0.002556873010938437,
+      "loss": 1.3133,
+      "step": 3265
+    },
+    {
+      "epoch": 0.2512114452734405,
+      "learning_rate": 0.002556615767748274,
+      "loss": 1.0744,
+      "step": 3266
+    },
+    {
+      "epoch": 0.25128836243365893,
+      "learning_rate": 0.0025563584628612145,
+      "loss": 1.1149,
+      "step": 3267
+    },
+    {
+      "epoch": 0.2513652795938774,
+      "learning_rate": 0.0025561010962922835,
+      "loss": 0.9961,
+      "step": 3268
+    },
+    {
+      "epoch": 0.25144219675409585,
+      "learning_rate": 0.0025558436680565085,
+      "loss": 1.5148,
+      "step": 3269
+    },
+    {
+      "epoch": 0.2515191139143143,
+      "learning_rate": 0.0025555861781689213,
+      "loss": 1.1389,
+      "step": 3270
+    },
+    {
+      "epoch": 0.2515960310745327,
+      "learning_rate": 0.002555328626644558,
+      "loss": 0.9799,
+      "step": 3271
+    },
+    {
+      "epoch": 0.25167294823475117,
+      "learning_rate": 0.0025550710134984553,
+      "loss": 1.1286,
+      "step": 3272
+    },
+    {
+      "epoch": 0.25174986539496963,
+      "learning_rate": 0.002554813338745657,
+      "loss": 1.282,
+      "step": 3273
+    },
+    {
+      "epoch": 0.2518267825551881,
+      "learning_rate": 0.0025545556024012084,
+      "loss": 1.2751,
+      "step": 3274
+    },
+    {
+      "epoch": 0.2519036997154065,
+      "learning_rate": 0.002554297804480159,
+      "loss": 1.2496,
+      "step": 3275
+    },
+    {
+      "epoch": 0.25198061687562495,
+      "learning_rate": 0.0025540399449975625,
+      "loss": 0.9874,
+      "step": 3276
+    },
+    {
+      "epoch": 0.2520575340358434,
+      "learning_rate": 0.002553782023968475,
+      "loss": 1.2749,
+      "step": 3277
+    },
+    {
+      "epoch": 0.2521344511960618,
+      "learning_rate": 0.002553524041407957,
+      "loss": 1.1992,
+      "step": 3278
+    },
+    {
+      "epoch": 0.2522113683562803,
+      "learning_rate": 0.002553265997331072,
+      "loss": 1.1654,
+      "step": 3279
+    },
+    {
+      "epoch": 0.25228828551649873,
+      "learning_rate": 0.002553007891752888,
+      "loss": 1.2483,
+      "step": 3280
+    },
+    {
+      "epoch": 0.2523652026767172,
+      "learning_rate": 0.002552749724688475,
+      "loss": 1.2375,
+      "step": 3281
+    },
+    {
+      "epoch": 0.2524421198369356,
+      "learning_rate": 0.0025524914961529096,
+      "loss": 1.2751,
+      "step": 3282
+    },
+    {
+      "epoch": 0.25251903699715406,
+      "learning_rate": 0.0025522332061612683,
+      "loss": 1.1489,
+      "step": 3283
+    },
+    {
+      "epoch": 0.2525959541573725,
+      "learning_rate": 0.002551974854728633,
+      "loss": 1.3221,
+      "step": 3284
+    },
+    {
+      "epoch": 0.252672871317591,
+      "learning_rate": 0.0025517164418700903,
+      "loss": 1.2766,
+      "step": 3285
+    },
+    {
+      "epoch": 0.2527497884778094,
+      "learning_rate": 0.002551457967600728,
+      "loss": 1.1042,
+      "step": 3286
+    },
+    {
+      "epoch": 0.25282670563802784,
+      "learning_rate": 0.0025511994319356393,
+      "loss": 1.1367,
+      "step": 3287
+    },
+    {
+      "epoch": 0.2529036227982463,
+      "learning_rate": 0.0025509408348899205,
+      "loss": 1.4032,
+      "step": 3288
+    },
+    {
+      "epoch": 0.25298053995846476,
+      "learning_rate": 0.002550682176478671,
+      "loss": 1.0402,
+      "step": 3289
+    },
+    {
+      "epoch": 0.25305745711868316,
+      "learning_rate": 0.0025504234567169937,
+      "loss": 1.1797,
+      "step": 3290
+    },
+    {
+      "epoch": 0.2531343742789016,
+      "learning_rate": 0.0025501646756199967,
+      "loss": 1.221,
+      "step": 3291
+    },
+    {
+      "epoch": 0.2532112914391201,
+      "learning_rate": 0.0025499058332027896,
+      "loss": 1.2086,
+      "step": 3292
+    },
+    {
+      "epoch": 0.25328820859933854,
+      "learning_rate": 0.0025496469294804865,
+      "loss": 1.2087,
+      "step": 3293
+    },
+    {
+      "epoch": 0.25336512575955694,
+      "learning_rate": 0.002549387964468206,
+      "loss": 0.7126,
+      "step": 3294
+    },
+    {
+      "epoch": 0.2534420429197754,
+      "learning_rate": 0.002549128938181068,
+      "loss": 1.2015,
+      "step": 3295
+    },
+    {
+      "epoch": 0.25351896007999386,
+      "learning_rate": 0.0025488698506341977,
+      "loss": 1.4902,
+      "step": 3296
+    },
+    {
+      "epoch": 0.25359587724021226,
+      "learning_rate": 0.0025486107018427243,
+      "loss": 1.5627,
+      "step": 3297
+    },
+    {
+      "epoch": 0.2536727944004307,
+      "learning_rate": 0.0025483514918217785,
+      "loss": 1.1413,
+      "step": 3298
+    },
+    {
+      "epoch": 0.2537497115606492,
+      "learning_rate": 0.002548092220586497,
+      "loss": 1.0084,
+      "step": 3299
+    },
+    {
+      "epoch": 0.25382662872086764,
+      "learning_rate": 0.0025478328881520184,
+      "loss": 1.2909,
+      "step": 3300
+    },
+    {
+      "epoch": 0.25390354588108605,
+      "learning_rate": 0.002547573494533486,
+      "loss": 1.3393,
+      "step": 3301
+    },
+    {
+      "epoch": 0.2539804630413045,
+      "learning_rate": 0.0025473140397460443,
+      "loss": 1.1288,
+      "step": 3302
+    },
+    {
+      "epoch": 0.25405738020152296,
+      "learning_rate": 0.002547054523804845,
+      "loss": 1.08,
+      "step": 3303
+    },
+    {
+      "epoch": 0.2541342973617414,
+      "learning_rate": 0.002546794946725041,
+      "loss": 0.8534,
+      "step": 3304
+    },
+    {
+      "epoch": 0.2542112145219598,
+      "learning_rate": 0.0025465353085217892,
+      "loss": 1.3742,
+      "step": 3305
+    },
+    {
+      "epoch": 0.2542881316821783,
+      "learning_rate": 0.0025462756092102493,
+      "loss": 0.9702,
+      "step": 3306
+    },
+    {
+      "epoch": 0.25436504884239675,
+      "learning_rate": 0.0025460158488055866,
+      "loss": 1.2243,
+      "step": 3307
+    },
+    {
+      "epoch": 0.2544419660026152,
+      "learning_rate": 0.0025457560273229687,
+      "loss": 1.3708,
+      "step": 3308
+    },
+    {
+      "epoch": 0.2545188831628336,
+      "learning_rate": 0.002545496144777566,
+      "loss": 1.0532,
+      "step": 3309
+    },
+    {
+      "epoch": 0.25459580032305207,
+      "learning_rate": 0.002545236201184554,
+      "loss": 1.21,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2546727174832705,
+      "learning_rate": 0.002544976196559111,
+      "loss": 1.0845,
+      "step": 3311
+    },
+    {
+      "epoch": 0.254749634643489,
+      "learning_rate": 0.002544716130916419,
+      "loss": 0.9006,
+      "step": 3312
+    },
+    {
+      "epoch": 0.2548265518037074,
+      "learning_rate": 0.002544456004271663,
+      "loss": 1.0491,
+      "step": 3313
+    },
+    {
+      "epoch": 0.25490346896392585,
+      "learning_rate": 0.0025441958166400325,
+      "loss": 1.2703,
+      "step": 3314
+    },
+    {
+      "epoch": 0.2549803861241443,
+      "learning_rate": 0.00254393556803672,
+      "loss": 1.1311,
+      "step": 3315
+    },
+    {
+      "epoch": 0.25505730328436277,
+      "learning_rate": 0.0025436752584769213,
+      "loss": 1.0844,
+      "step": 3316
+    },
+    {
+      "epoch": 0.2551342204445812,
+      "learning_rate": 0.0025434148879758373,
+      "loss": 1.5026,
+      "step": 3317
+    },
+    {
+      "epoch": 0.25521113760479963,
+      "learning_rate": 0.00254315445654867,
+      "loss": 1.1845,
+      "step": 3318
+    },
+    {
+      "epoch": 0.2552880547650181,
+      "learning_rate": 0.0025428939642106276,
+      "loss": 1.1035,
+      "step": 3319
+    },
+    {
+      "epoch": 0.2553649719252365,
+      "learning_rate": 0.002542633410976919,
+      "loss": 1.0475,
+      "step": 3320
+    },
+    {
+      "epoch": 0.25544188908545495,
+      "learning_rate": 0.002542372796862759,
+      "loss": 0.9373,
+      "step": 3321
+    },
+    {
+      "epoch": 0.2555188062456734,
+      "learning_rate": 0.0025421121218833654,
+      "loss": 0.7934,
+      "step": 3322
+    },
+    {
+      "epoch": 0.2555957234058919,
+      "learning_rate": 0.0025418513860539587,
+      "loss": 1.4146,
+      "step": 3323
+    },
+    {
+      "epoch": 0.2556726405661103,
+      "learning_rate": 0.0025415905893897642,
+      "loss": 1.0758,
+      "step": 3324
+    },
+    {
+      "epoch": 0.25574955772632874,
+      "learning_rate": 0.002541329731906009,
+      "loss": 1.0413,
+      "step": 3325
+    },
+    {
+      "epoch": 0.2558264748865472,
+      "learning_rate": 0.0025410688136179257,
+      "loss": 1.2682,
+      "step": 3326
+    },
+    {
+      "epoch": 0.25590339204676565,
+      "learning_rate": 0.00254080783454075,
+      "loss": 1.2823,
+      "step": 3327
+    },
+    {
+      "epoch": 0.25598030920698406,
+      "learning_rate": 0.0025405467946897197,
+      "loss": 1.1681,
+      "step": 3328
+    },
+    {
+      "epoch": 0.2560572263672025,
+      "learning_rate": 0.0025402856940800775,
+      "loss": 1.0219,
+      "step": 3329
+    },
+    {
+      "epoch": 0.256134143527421,
+      "learning_rate": 0.0025400245327270696,
+      "loss": 1.0044,
+      "step": 3330
+    },
+    {
+      "epoch": 0.25621106068763944,
+      "learning_rate": 0.002539763310645945,
+      "loss": 0.9554,
+      "step": 3331
+    },
+    {
+      "epoch": 0.25628797784785784,
+      "learning_rate": 0.0025395020278519576,
+      "loss": 1.429,
+      "step": 3332
+    },
+    {
+      "epoch": 0.2563648950080763,
+      "learning_rate": 0.0025392406843603634,
+      "loss": 0.9826,
+      "step": 3333
+    },
+    {
+      "epoch": 0.25644181216829476,
+      "learning_rate": 0.0025389792801864225,
+      "loss": 1.1936,
+      "step": 3334
+    },
+    {
+      "epoch": 0.2565187293285132,
+      "learning_rate": 0.0025387178153453986,
+      "loss": 0.9941,
+      "step": 3335
+    },
+    {
+      "epoch": 0.2565956464887316,
+      "learning_rate": 0.0025384562898525588,
+      "loss": 1.112,
+      "step": 3336
+    },
+    {
+      "epoch": 0.2566725636489501,
+      "learning_rate": 0.002538194703723174,
+      "loss": 1.2618,
+      "step": 3337
+    },
+    {
+      "epoch": 0.25674948080916854,
+      "learning_rate": 0.0025379330569725183,
+      "loss": 0.7113,
+      "step": 3338
+    },
+    {
+      "epoch": 0.25682639796938694,
+      "learning_rate": 0.0025376713496158697,
+      "loss": 1.2442,
+      "step": 3339
+    },
+    {
+      "epoch": 0.2569033151296054,
+      "learning_rate": 0.0025374095816685103,
+      "loss": 1.1023,
+      "step": 3340
+    },
+    {
+      "epoch": 0.25698023228982386,
+      "learning_rate": 0.0025371477531457233,
+      "loss": 1.3482,
+      "step": 3341
+    },
+    {
+      "epoch": 0.2570571494500423,
+      "learning_rate": 0.0025368858640627987,
+      "loss": 1.6154,
+      "step": 3342
+    },
+    {
+      "epoch": 0.2571340666102607,
+      "learning_rate": 0.002536623914435028,
+      "loss": 0.895,
+      "step": 3343
+    },
+    {
+      "epoch": 0.2572109837704792,
+      "learning_rate": 0.0025363619042777065,
+      "loss": 1.1876,
+      "step": 3344
+    },
+    {
+      "epoch": 0.25728790093069764,
+      "learning_rate": 0.002536099833606133,
+      "loss": 1.3241,
+      "step": 3345
+    },
+    {
+      "epoch": 0.2573648180909161,
+      "learning_rate": 0.0025358377024356105,
+      "loss": 0.7273,
+      "step": 3346
+    },
+    {
+      "epoch": 0.2574417352511345,
+      "learning_rate": 0.0025355755107814453,
+      "loss": 1.1544,
+      "step": 3347
+    },
+    {
+      "epoch": 0.25751865241135297,
+      "learning_rate": 0.0025353132586589462,
+      "loss": 1.1422,
+      "step": 3348
+    },
+    {
+      "epoch": 0.2575955695715714,
+      "learning_rate": 0.0025350509460834274,
+      "loss": 1.3658,
+      "step": 3349
+    },
+    {
+      "epoch": 0.2576724867317899,
+      "learning_rate": 0.002534788573070205,
+      "loss": 1.0071,
+      "step": 3350
+    },
+    {
+      "epoch": 0.2577494038920083,
+      "learning_rate": 0.0025345261396345995,
+      "loss": 1.0649,
+      "step": 3351
+    },
+    {
+      "epoch": 0.25782632105222675,
+      "learning_rate": 0.002534263645791935,
+      "loss": 1.415,
+      "step": 3352
+    },
+    {
+      "epoch": 0.2579032382124452,
+      "learning_rate": 0.0025340010915575373,
+      "loss": 1.492,
+      "step": 3353
+    },
+    {
+      "epoch": 0.25798015537266367,
+      "learning_rate": 0.002533738476946739,
+      "loss": 1.0133,
+      "step": 3354
+    },
+    {
+      "epoch": 0.25805707253288207,
+      "learning_rate": 0.002533475801974874,
+      "loss": 1.13,
+      "step": 3355
+    },
+    {
+      "epoch": 0.25813398969310053,
+      "learning_rate": 0.002533213066657279,
+      "loss": 1.1198,
+      "step": 3356
+    },
+    {
+      "epoch": 0.258210906853319,
+      "learning_rate": 0.0025329502710092966,
+      "loss": 0.7984,
+      "step": 3357
+    },
+    {
+      "epoch": 0.2582878240135374,
+      "learning_rate": 0.0025326874150462713,
+      "loss": 1.0155,
+      "step": 3358
+    },
+    {
+      "epoch": 0.25836474117375585,
+      "learning_rate": 0.002532424498783552,
+      "loss": 1.0591,
+      "step": 3359
+    },
+    {
+      "epoch": 0.2584416583339743,
+      "learning_rate": 0.0025321615222364895,
+      "loss": 0.9495,
+      "step": 3360
+    },
+    {
+      "epoch": 0.25851857549419277,
+      "learning_rate": 0.0025318984854204407,
+      "loss": 1.349,
+      "step": 3361
+    },
+    {
+      "epoch": 0.2585954926544112,
+      "learning_rate": 0.0025316353883507634,
+      "loss": 1.036,
+      "step": 3362
+    },
+    {
+      "epoch": 0.25867240981462963,
+      "learning_rate": 0.002531372231042821,
+      "loss": 1.448,
+      "step": 3363
+    },
+    {
+      "epoch": 0.2587493269748481,
+      "learning_rate": 0.002531109013511979,
+      "loss": 1.267,
+      "step": 3364
+    },
+    {
+      "epoch": 0.25882624413506655,
+      "learning_rate": 0.0025308457357736073,
+      "loss": 1.0893,
+      "step": 3365
+    },
+    {
+      "epoch": 0.25890316129528496,
+      "learning_rate": 0.0025305823978430788,
+      "loss": 1.2601,
+      "step": 3366
+    },
+    {
+      "epoch": 0.2589800784555034,
+      "learning_rate": 0.00253031899973577,
+      "loss": 1.5132,
+      "step": 3367
+    },
+    {
+      "epoch": 0.2590569956157219,
+      "learning_rate": 0.0025300555414670607,
+      "loss": 1.187,
+      "step": 3368
+    },
+    {
+      "epoch": 0.25913391277594033,
+      "learning_rate": 0.0025297920230523347,
+      "loss": 1.0362,
+      "step": 3369
+    },
+    {
+      "epoch": 0.25921082993615874,
+      "learning_rate": 0.00252952844450698,
+      "loss": 1.0495,
+      "step": 3370
+    },
+    {
+      "epoch": 0.2592877470963772,
+      "learning_rate": 0.002529264805846386,
+      "loss": 1.2691,
+      "step": 3371
+    },
+    {
+      "epoch": 0.25936466425659566,
+      "learning_rate": 0.0025290011070859472,
+      "loss": 1.1531,
+      "step": 3372
+    },
+    {
+      "epoch": 0.2594415814168141,
+      "learning_rate": 0.0025287373482410615,
+      "loss": 1.0173,
+      "step": 3373
+    },
+    {
+      "epoch": 0.2595184985770325,
+      "learning_rate": 0.00252847352932713,
+      "loss": 0.9931,
+      "step": 3374
+    },
+    {
+      "epoch": 0.259595415737251,
+      "learning_rate": 0.002528209650359557,
+      "loss": 1.2584,
+      "step": 3375
+    },
+    {
+      "epoch": 0.25967233289746944,
+      "learning_rate": 0.0025279457113537513,
+      "loss": 1.049,
+      "step": 3376
+    },
+    {
+      "epoch": 0.2597492500576879,
+      "learning_rate": 0.002527681712325124,
+      "loss": 1.3439,
+      "step": 3377
+    },
+    {
+      "epoch": 0.2598261672179063,
+      "learning_rate": 0.0025274176532890905,
+      "loss": 1.3921,
+      "step": 3378
+    },
+    {
+      "epoch": 0.25990308437812476,
+      "learning_rate": 0.0025271535342610697,
+      "loss": 0.5957,
+      "step": 3379
+    },
+    {
+      "epoch": 0.2599800015383432,
+      "learning_rate": 0.0025268893552564836,
+      "loss": 0.9592,
+      "step": 3380
+    },
+    {
+      "epoch": 0.2600569186985616,
+      "learning_rate": 0.0025266251162907576,
+      "loss": 1.0464,
+      "step": 3381
+    },
+    {
+      "epoch": 0.2601338358587801,
+      "learning_rate": 0.0025263608173793214,
+      "loss": 1.1172,
+      "step": 3382
+    },
+    {
+      "epoch": 0.26021075301899854,
+      "learning_rate": 0.002526096458537607,
+      "loss": 1.1096,
+      "step": 3383
+    },
+    {
+      "epoch": 0.260287670179217,
+      "learning_rate": 0.0025258320397810513,
+      "loss": 0.8709,
+      "step": 3384
+    },
+    {
+      "epoch": 0.2603645873394354,
+      "learning_rate": 0.0025255675611250933,
+      "loss": 0.8465,
+      "step": 3385
+    },
+    {
+      "epoch": 0.26044150449965386,
+      "learning_rate": 0.0025253030225851772,
+      "loss": 1.331,
+      "step": 3386
+    },
+    {
+      "epoch": 0.2605184216598723,
+      "learning_rate": 0.0025250384241767487,
+      "loss": 1.1716,
+      "step": 3387
+    },
+    {
+      "epoch": 0.2605953388200908,
+      "learning_rate": 0.0025247737659152584,
+      "loss": 1.1109,
+      "step": 3388
+    },
+    {
+      "epoch": 0.2606722559803092,
+      "learning_rate": 0.00252450904781616,
+      "loss": 1.1885,
+      "step": 3389
+    },
+    {
+      "epoch": 0.26074917314052765,
+      "learning_rate": 0.0025242442698949104,
+      "loss": 1.655,
+      "step": 3390
+    },
+    {
+      "epoch": 0.2608260903007461,
+      "learning_rate": 0.0025239794321669706,
+      "loss": 1.1128,
+      "step": 3391
+    },
+    {
+      "epoch": 0.26090300746096456,
+      "learning_rate": 0.0025237145346478046,
+      "loss": 0.9386,
+      "step": 3392
+    },
+    {
+      "epoch": 0.26097992462118297,
+      "learning_rate": 0.0025234495773528804,
+      "loss": 1.4328,
+      "step": 3393
+    },
+    {
+      "epoch": 0.2610568417814014,
+      "learning_rate": 0.002523184560297668,
+      "loss": 1.1064,
+      "step": 3394
+    },
+    {
+      "epoch": 0.2611337589416199,
+      "learning_rate": 0.002522919483497643,
+      "loss": 1.2672,
+      "step": 3395
+    },
+    {
+      "epoch": 0.26121067610183835,
+      "learning_rate": 0.0025226543469682836,
+      "loss": 1.6387,
+      "step": 3396
+    },
+    {
+      "epoch": 0.26128759326205675,
+      "learning_rate": 0.0025223891507250708,
+      "loss": 1.2837,
+      "step": 3397
+    },
+    {
+      "epoch": 0.2613645104222752,
+      "learning_rate": 0.00252212389478349,
+      "loss": 1.0324,
+      "step": 3398
+    },
+    {
+      "epoch": 0.26144142758249367,
+      "learning_rate": 0.00252185857915903,
+      "loss": 1.277,
+      "step": 3399
+    },
+    {
+      "epoch": 0.26151834474271207,
+      "learning_rate": 0.0025215932038671826,
+      "loss": 1.1001,
+      "step": 3400
+    },
+    {
+      "epoch": 0.26159526190293053,
+      "learning_rate": 0.002521327768923443,
+      "loss": 1.6005,
+      "step": 3401
+    },
+    {
+      "epoch": 0.261672179063149,
+      "learning_rate": 0.002521062274343311,
+      "loss": 1.1225,
+      "step": 3402
+    },
+    {
+      "epoch": 0.26174909622336745,
+      "learning_rate": 0.0025207967201422883,
+      "loss": 0.8252,
+      "step": 3403
+    },
+    {
+      "epoch": 0.26182601338358585,
+      "learning_rate": 0.0025205311063358814,
+      "loss": 1.2752,
+      "step": 3404
+    },
+    {
+      "epoch": 0.2619029305438043,
+      "learning_rate": 0.0025202654329395994,
+      "loss": 1.1486,
+      "step": 3405
+    },
+    {
+      "epoch": 0.2619798477040228,
+      "learning_rate": 0.002519999699968956,
+      "loss": 1.2136,
+      "step": 3406
+    },
+    {
+      "epoch": 0.26205676486424123,
+      "learning_rate": 0.002519733907439467,
+      "loss": 1.1214,
+      "step": 3407
+    },
+    {
+      "epoch": 0.26213368202445964,
+      "learning_rate": 0.0025194680553666526,
+      "loss": 1.0703,
+      "step": 3408
+    },
+    {
+      "epoch": 0.2622105991846781,
+      "learning_rate": 0.002519202143766035,
+      "loss": 1.2793,
+      "step": 3409
+    },
+    {
+      "epoch": 0.26228751634489655,
+      "learning_rate": 0.0025189361726531425,
+      "loss": 1.1329,
+      "step": 3410
+    },
+    {
+      "epoch": 0.262364433505115,
+      "learning_rate": 0.002518670142043505,
+      "loss": 1.107,
+      "step": 3411
+    },
+    {
+      "epoch": 0.2624413506653334,
+      "learning_rate": 0.0025184040519526567,
+      "loss": 0.8191,
+      "step": 3412
+    },
+    {
+      "epoch": 0.2625182678255519,
+      "learning_rate": 0.002518137902396134,
+      "loss": 0.9387,
+      "step": 3413
+    },
+    {
+      "epoch": 0.26259518498577034,
+      "learning_rate": 0.0025178716933894783,
+      "loss": 0.9596,
+      "step": 3414
+    },
+    {
+      "epoch": 0.2626721021459888,
+      "learning_rate": 0.0025176054249482337,
+      "loss": 0.8431,
+      "step": 3415
+    },
+    {
+      "epoch": 0.2627490193062072,
+      "learning_rate": 0.002517339097087948,
+      "loss": 1.1453,
+      "step": 3416
+    },
+    {
+      "epoch": 0.26282593646642566,
+      "learning_rate": 0.0025170727098241716,
+      "loss": 1.3631,
+      "step": 3417
+    },
+    {
+      "epoch": 0.2629028536266441,
+      "learning_rate": 0.00251680626317246,
+      "loss": 1.4974,
+      "step": 3418
+    },
+    {
+      "epoch": 0.2629797707868626,
+      "learning_rate": 0.002516539757148371,
+      "loss": 0.9919,
+      "step": 3419
+    },
+    {
+      "epoch": 0.263056687947081,
+      "learning_rate": 0.0025162731917674664,
+      "loss": 1.4821,
+      "step": 3420
+    },
+    {
+      "epoch": 0.26313360510729944,
+      "learning_rate": 0.0025160065670453103,
+      "loss": 1.1854,
+      "step": 3421
+    },
+    {
+      "epoch": 0.2632105222675179,
+      "learning_rate": 0.0025157398829974724,
+      "loss": 1.5294,
+      "step": 3422
+    },
+    {
+      "epoch": 0.2632874394277363,
+      "learning_rate": 0.002515473139639524,
+      "loss": 1.1279,
+      "step": 3423
+    },
+    {
+      "epoch": 0.26336435658795476,
+      "learning_rate": 0.002515206336987041,
+      "loss": 1.087,
+      "step": 3424
+    },
+    {
+      "epoch": 0.2634412737481732,
+      "learning_rate": 0.002514939475055602,
+      "loss": 1.1478,
+      "step": 3425
+    },
+    {
+      "epoch": 0.2635181909083917,
+      "learning_rate": 0.0025146725538607887,
+      "loss": 1.114,
+      "step": 3426
+    },
+    {
+      "epoch": 0.2635951080686101,
+      "learning_rate": 0.002514405573418188,
+      "loss": 1.2608,
+      "step": 3427
+    },
+    {
+      "epoch": 0.26367202522882854,
+      "learning_rate": 0.0025141385337433884,
+      "loss": 1.3321,
+      "step": 3428
+    },
+    {
+      "epoch": 0.263748942389047,
+      "learning_rate": 0.0025138714348519825,
+      "loss": 0.8488,
+      "step": 3429
+    },
+    {
+      "epoch": 0.26382585954926546,
+      "learning_rate": 0.002513604276759568,
+      "loss": 1.1953,
+      "step": 3430
+    },
+    {
+      "epoch": 0.26390277670948387,
+      "learning_rate": 0.002513337059481743,
+      "loss": 1.1408,
+      "step": 3431
+    },
+    {
+      "epoch": 0.2639796938697023,
+      "learning_rate": 0.002513069783034111,
+      "loss": 1.1631,
+      "step": 3432
+    },
+    {
+      "epoch": 0.2640566110299208,
+      "learning_rate": 0.002512802447432278,
+      "loss": 1.4827,
+      "step": 3433
+    },
+    {
+      "epoch": 0.26413352819013924,
+      "learning_rate": 0.002512535052691855,
+      "loss": 1.1841,
+      "step": 3434
+    },
+    {
+      "epoch": 0.26421044535035765,
+      "learning_rate": 0.0025122675988284555,
+      "loss": 0.8926,
+      "step": 3435
+    },
+    {
+      "epoch": 0.2642873625105761,
+      "learning_rate": 0.002512000085857695,
+      "loss": 1.0023,
+      "step": 3436
+    },
+    {
+      "epoch": 0.26436427967079457,
+      "learning_rate": 0.0025117325137951952,
+      "loss": 0.9331,
+      "step": 3437
+    },
+    {
+      "epoch": 0.264441196831013,
+      "learning_rate": 0.0025114648826565795,
+      "loss": 1.2008,
+      "step": 3438
+    },
+    {
+      "epoch": 0.26451811399123143,
+      "learning_rate": 0.0025111971924574755,
+      "loss": 1.0714,
+      "step": 3439
+    },
+    {
+      "epoch": 0.2645950311514499,
+      "learning_rate": 0.002510929443213513,
+      "loss": 1.1681,
+      "step": 3440
+    },
+    {
+      "epoch": 0.26467194831166835,
+      "learning_rate": 0.0025106616349403277,
+      "loss": 0.9974,
+      "step": 3441
+    },
+    {
+      "epoch": 0.26474886547188675,
+      "learning_rate": 0.002510393767653556,
+      "loss": 1.2625,
+      "step": 3442
+    },
+    {
+      "epoch": 0.2648257826321052,
+      "learning_rate": 0.0025101258413688385,
+      "loss": 0.9855,
+      "step": 3443
+    },
+    {
+      "epoch": 0.26490269979232367,
+      "learning_rate": 0.002509857856101821,
+      "loss": 1.0247,
+      "step": 3444
+    },
+    {
+      "epoch": 0.26497961695254213,
+      "learning_rate": 0.0025095898118681504,
+      "loss": 1.1187,
+      "step": 3445
+    },
+    {
+      "epoch": 0.26505653411276053,
+      "learning_rate": 0.0025093217086834788,
+      "loss": 1.1966,
+      "step": 3446
+    },
+    {
+      "epoch": 0.265133451272979,
+      "learning_rate": 0.0025090535465634605,
+      "loss": 1.6931,
+      "step": 3447
+    },
+    {
+      "epoch": 0.26521036843319745,
+      "learning_rate": 0.002508785325523754,
+      "loss": 1.1623,
+      "step": 3448
+    },
+    {
+      "epoch": 0.2652872855934159,
+      "learning_rate": 0.002508517045580021,
+      "loss": 1.3653,
+      "step": 3449
+    },
+    {
+      "epoch": 0.2653642027536343,
+      "learning_rate": 0.0025082487067479274,
+      "loss": 1.1628,
+      "step": 3450
+    },
+    {
+      "epoch": 0.2654411199138528,
+      "learning_rate": 0.0025079803090431404,
+      "loss": 1.2927,
+      "step": 3451
+    },
+    {
+      "epoch": 0.26551803707407123,
+      "learning_rate": 0.002507711852481332,
+      "loss": 0.9913,
+      "step": 3452
+    },
+    {
+      "epoch": 0.2655949542342897,
+      "learning_rate": 0.0025074433370781794,
+      "loss": 1.4476,
+      "step": 3453
+    },
+    {
+      "epoch": 0.2656718713945081,
+      "learning_rate": 0.0025071747628493605,
+      "loss": 0.8821,
+      "step": 3454
+    },
+    {
+      "epoch": 0.26574878855472656,
+      "learning_rate": 0.002506906129810557,
+      "loss": 1.3082,
+      "step": 3455
+    },
+    {
+      "epoch": 0.265825705714945,
+      "learning_rate": 0.0025066374379774554,
+      "loss": 1.1008,
+      "step": 3456
+    },
+    {
+      "epoch": 0.2659026228751635,
+      "learning_rate": 0.0025063686873657443,
+      "loss": 0.8634,
+      "step": 3457
+    },
+    {
+      "epoch": 0.2659795400353819,
+      "learning_rate": 0.0025060998779911177,
+      "loss": 1.2657,
+      "step": 3458
+    },
+    {
+      "epoch": 0.26605645719560034,
+      "learning_rate": 0.0025058310098692695,
+      "loss": 1.0036,
+      "step": 3459
+    },
+    {
+      "epoch": 0.2661333743558188,
+      "learning_rate": 0.002505562083015901,
+      "loss": 0.9075,
+      "step": 3460
+    },
+    {
+      "epoch": 0.2662102915160372,
+      "learning_rate": 0.0025052930974467143,
+      "loss": 1.2736,
+      "step": 3461
+    },
+    {
+      "epoch": 0.26628720867625566,
+      "learning_rate": 0.0025050240531774163,
+      "loss": 1.2193,
+      "step": 3462
+    },
+    {
+      "epoch": 0.2663641258364741,
+      "learning_rate": 0.0025047549502237158,
+      "loss": 1.2223,
+      "step": 3463
+    },
+    {
+      "epoch": 0.2664410429966926,
+      "learning_rate": 0.0025044857886013273,
+      "loss": 1.0444,
+      "step": 3464
+    },
+    {
+      "epoch": 0.266517960156911,
+      "learning_rate": 0.0025042165683259667,
+      "loss": 1.1946,
+      "step": 3465
+    },
+    {
+      "epoch": 0.26659487731712944,
+      "learning_rate": 0.002503947289413354,
+      "loss": 0.9865,
+      "step": 3466
+    },
+    {
+      "epoch": 0.2666717944773479,
+      "learning_rate": 0.0025036779518792123,
+      "loss": 1.3231,
+      "step": 3467
+    },
+    {
+      "epoch": 0.26674871163756636,
+      "learning_rate": 0.002503408555739269,
+      "loss": 1.0282,
+      "step": 3468
+    },
+    {
+      "epoch": 0.26682562879778476,
+      "learning_rate": 0.0025031391010092554,
+      "loss": 1.5534,
+      "step": 3469
+    },
+    {
+      "epoch": 0.2669025459580032,
+      "learning_rate": 0.0025028695877049034,
+      "loss": 1.2378,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2669794631182217,
+      "learning_rate": 0.0025026000158419512,
+      "loss": 1.0397,
+      "step": 3471
+    },
+    {
+      "epoch": 0.26705638027844014,
+      "learning_rate": 0.0025023303854361393,
+      "loss": 1.2919,
+      "step": 3472
+    },
+    {
+      "epoch": 0.26713329743865855,
+      "learning_rate": 0.0025020606965032114,
+      "loss": 1.1562,
+      "step": 3473
+    },
+    {
+      "epoch": 0.267210214598877,
+      "learning_rate": 0.002501790949058915,
+      "loss": 1.002,
+      "step": 3474
+    },
+    {
+      "epoch": 0.26728713175909546,
+      "learning_rate": 0.002501521143119001,
+      "loss": 1.2503,
+      "step": 3475
+    },
+    {
+      "epoch": 0.2673640489193139,
+      "learning_rate": 0.0025012512786992246,
+      "loss": 0.9608,
+      "step": 3476
+    },
+    {
+      "epoch": 0.2674409660795323,
+      "learning_rate": 0.0025009813558153414,
+      "loss": 0.7097,
+      "step": 3477
+    },
+    {
+      "epoch": 0.2675178832397508,
+      "learning_rate": 0.0025007113744831143,
+      "loss": 0.8568,
+      "step": 3478
+    },
+    {
+      "epoch": 0.26759480039996925,
+      "learning_rate": 0.0025004413347183073,
+      "loss": 1.1077,
+      "step": 3479
+    },
+    {
+      "epoch": 0.2676717175601877,
+      "learning_rate": 0.002500171236536687,
+      "loss": 0.9338,
+      "step": 3480
+    },
+    {
+      "epoch": 0.2677486347204061,
+      "learning_rate": 0.0024999010799540275,
+      "loss": 1.3056,
+      "step": 3481
+    },
+    {
+      "epoch": 0.26782555188062457,
+      "learning_rate": 0.0024996308649861006,
+      "loss": 1.1272,
+      "step": 3482
+    },
+    {
+      "epoch": 0.267902469040843,
+      "learning_rate": 0.002499360591648686,
+      "loss": 0.8716,
+      "step": 3483
+    },
+    {
+      "epoch": 0.26797938620106143,
+      "learning_rate": 0.0024990902599575653,
+      "loss": 0.7946,
+      "step": 3484
+    },
+    {
+      "epoch": 0.2680563033612799,
+      "learning_rate": 0.002498819869928523,
+      "loss": 1.2492,
+      "step": 3485
+    },
+    {
+      "epoch": 0.26813322052149835,
+      "learning_rate": 0.0024985494215773473,
+      "loss": 1.1077,
+      "step": 3486
+    },
+    {
+      "epoch": 0.2682101376817168,
+      "learning_rate": 0.00249827891491983,
+      "loss": 1.0933,
+      "step": 3487
+    },
+    {
+      "epoch": 0.2682870548419352,
+      "learning_rate": 0.002498008349971767,
+      "loss": 1.0039,
+      "step": 3488
+    },
+    {
+      "epoch": 0.2683639720021537,
+      "learning_rate": 0.0024977377267489565,
+      "loss": 0.8665,
+      "step": 3489
+    },
+    {
+      "epoch": 0.26844088916237213,
+      "learning_rate": 0.0024974670452671995,
+      "loss": 1.0237,
+      "step": 3490
+    },
+    {
+      "epoch": 0.2685178063225906,
+      "learning_rate": 0.0024971963055423025,
+      "loss": 1.3132,
+      "step": 3491
+    },
+    {
+      "epoch": 0.268594723482809,
+      "learning_rate": 0.0024969255075900742,
+      "loss": 0.9821,
+      "step": 3492
+    },
+    {
+      "epoch": 0.26867164064302745,
+      "learning_rate": 0.0024966546514263266,
+      "loss": 1.2235,
+      "step": 3493
+    },
+    {
+      "epoch": 0.2687485578032459,
+      "learning_rate": 0.002496383737066875,
+      "loss": 0.8197,
+      "step": 3494
+    },
+    {
+      "epoch": 0.2688254749634644,
+      "learning_rate": 0.0024961127645275385,
+      "loss": 1.2345,
+      "step": 3495
+    },
+    {
+      "epoch": 0.2689023921236828,
+      "learning_rate": 0.0024958417338241403,
+      "loss": 1.2893,
+      "step": 3496
+    },
+    {
+      "epoch": 0.26897930928390124,
+      "learning_rate": 0.0024955706449725046,
+      "loss": 0.9128,
+      "step": 3497
+    },
+    {
+      "epoch": 0.2690562264441197,
+      "learning_rate": 0.0024952994979884615,
+      "loss": 1.0012,
+      "step": 3498
+    },
+    {
+      "epoch": 0.26913314360433815,
+      "learning_rate": 0.0024950282928878436,
+      "loss": 0.927,
+      "step": 3499
+    },
+    {
+      "epoch": 0.26921006076455656,
+      "learning_rate": 0.0024947570296864866,
+      "loss": 1.5774,
+      "step": 3500
+    },
+    {
+      "epoch": 0.269286977924775,
+      "learning_rate": 0.0024944857084002302,
+      "loss": 1.029,
+      "step": 3501
+    },
+    {
+      "epoch": 0.2693638950849935,
+      "learning_rate": 0.0024942143290449166,
+      "loss": 0.9192,
+      "step": 3502
+    },
+    {
+      "epoch": 0.2694408122452119,
+      "learning_rate": 0.002493942891636392,
+      "loss": 0.9637,
+      "step": 3503
+    },
+    {
+      "epoch": 0.26951772940543034,
+      "learning_rate": 0.002493671396190506,
+      "loss": 1.0466,
+      "step": 3504
+    },
+    {
+      "epoch": 0.2695946465656488,
+      "learning_rate": 0.002493399842723112,
+      "loss": 1.235,
+      "step": 3505
+    },
+    {
+      "epoch": 0.26967156372586726,
+      "learning_rate": 0.002493128231250066,
+      "loss": 1.0379,
+      "step": 3506
+    },
+    {
+      "epoch": 0.26974848088608566,
+      "learning_rate": 0.0024928565617872273,
+      "loss": 0.975,
+      "step": 3507
+    },
+    {
+      "epoch": 0.2698253980463041,
+      "learning_rate": 0.0024925848343504593,
+      "loss": 0.7817,
+      "step": 3508
+    },
+    {
+      "epoch": 0.2699023152065226,
+      "learning_rate": 0.002492313048955628,
+      "loss": 1.0762,
+      "step": 3509
+    },
+    {
+      "epoch": 0.26997923236674104,
+      "learning_rate": 0.002492041205618604,
+      "loss": 1.2626,
+      "step": 3510
+    },
+    {
+      "epoch": 0.27005614952695944,
+      "learning_rate": 0.00249176930435526,
+      "loss": 1.6238,
+      "step": 3511
+    },
+    {
+      "epoch": 0.2701330666871779,
+      "learning_rate": 0.002491497345181473,
+      "loss": 1.3942,
+      "step": 3512
+    },
+    {
+      "epoch": 0.27020998384739636,
+      "learning_rate": 0.0024912253281131227,
+      "loss": 0.9294,
+      "step": 3513
+    },
+    {
+      "epoch": 0.2702869010076148,
+      "learning_rate": 0.0024909532531660924,
+      "loss": 1.2417,
+      "step": 3514
+    },
+    {
+      "epoch": 0.2703638181678332,
+      "learning_rate": 0.0024906811203562686,
+      "loss": 0.9573,
+      "step": 3515
+    },
+    {
+      "epoch": 0.2704407353280517,
+      "learning_rate": 0.0024904089296995417,
+      "loss": 1.312,
+      "step": 3516
+    },
+    {
+      "epoch": 0.27051765248827014,
+      "learning_rate": 0.0024901366812118057,
+      "loss": 1.0174,
+      "step": 3517
+    },
+    {
+      "epoch": 0.2705945696484886,
+      "learning_rate": 0.0024898643749089564,
+      "loss": 1.3006,
+      "step": 3518
+    },
+    {
+      "epoch": 0.270671486808707,
+      "learning_rate": 0.0024895920108068954,
+      "loss": 1.2633,
+      "step": 3519
+    },
+    {
+      "epoch": 0.27074840396892547,
+      "learning_rate": 0.0024893195889215246,
+      "loss": 1.3204,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2708253211291439,
+      "learning_rate": 0.0024890471092687523,
+      "loss": 0.9997,
+      "step": 3521
+    },
+    {
+      "epoch": 0.27090223828936233,
+      "learning_rate": 0.0024887745718644894,
+      "loss": 1.1393,
+      "step": 3522
+    },
+    {
+      "epoch": 0.2709791554495808,
+      "learning_rate": 0.002488501976724648,
+      "loss": 1.2127,
+      "step": 3523
+    },
+    {
+      "epoch": 0.27105607260979925,
+      "learning_rate": 0.0024882293238651454,
+      "loss": 1.2812,
+      "step": 3524
+    },
+    {
+      "epoch": 0.2711329897700177,
+      "learning_rate": 0.0024879566133019036,
+      "loss": 1.0829,
+      "step": 3525
+    },
+    {
+      "epoch": 0.2712099069302361,
+      "learning_rate": 0.002487683845050845,
+      "loss": 1.2449,
+      "step": 3526
+    },
+    {
+      "epoch": 0.27128682409045457,
+      "learning_rate": 0.0024874110191278978,
+      "loss": 1.3035,
+      "step": 3527
+    },
+    {
+      "epoch": 0.27136374125067303,
+      "learning_rate": 0.0024871381355489916,
+      "loss": 0.972,
+      "step": 3528
+    },
+    {
+      "epoch": 0.2714406584108915,
+      "learning_rate": 0.002486865194330061,
+      "loss": 0.9873,
+      "step": 3529
+    },
+    {
+      "epoch": 0.2715175755711099,
+      "learning_rate": 0.0024865921954870442,
+      "loss": 1.1617,
+      "step": 3530
+    },
+    {
+      "epoch": 0.27159449273132835,
+      "learning_rate": 0.0024863191390358795,
+      "loss": 0.7969,
+      "step": 3531
+    },
+    {
+      "epoch": 0.2716714098915468,
+      "learning_rate": 0.0024860460249925133,
+      "loss": 1.3353,
+      "step": 3532
+    },
+    {
+      "epoch": 0.27174832705176527,
+      "learning_rate": 0.0024857728533728917,
+      "loss": 0.9596,
+      "step": 3533
+    },
+    {
+      "epoch": 0.2718252442119837,
+      "learning_rate": 0.002485499624192966,
+      "loss": 1.1102,
+      "step": 3534
+    },
+    {
+      "epoch": 0.27190216137220213,
+      "learning_rate": 0.0024852263374686902,
+      "loss": 1.0854,
+      "step": 3535
+    },
+    {
+      "epoch": 0.2719790785324206,
+      "learning_rate": 0.0024849529932160215,
+      "loss": 1.0259,
+      "step": 3536
+    },
+    {
+      "epoch": 0.27205599569263905,
+      "learning_rate": 0.002484679591450921,
+      "loss": 1.3463,
+      "step": 3537
+    },
+    {
+      "epoch": 0.27213291285285746,
+      "learning_rate": 0.002484406132189353,
+      "loss": 1.1481,
+      "step": 3538
+    },
+    {
+      "epoch": 0.2722098300130759,
+      "learning_rate": 0.002484132615447285,
+      "loss": 0.9511,
+      "step": 3539
+    },
+    {
+      "epoch": 0.2722867471732944,
+      "learning_rate": 0.002483859041240688,
+      "loss": 1.3642,
+      "step": 3540
+    },
+    {
+      "epoch": 0.27236366433351283,
+      "learning_rate": 0.0024835854095855365,
+      "loss": 1.2416,
+      "step": 3541
+    },
+    {
+      "epoch": 0.27244058149373124,
+      "learning_rate": 0.002483311720497808,
+      "loss": 0.8573,
+      "step": 3542
+    },
+    {
+      "epoch": 0.2725174986539497,
+      "learning_rate": 0.0024830379739934827,
+      "loss": 1.0931,
+      "step": 3543
+    },
+    {
+      "epoch": 0.27259441581416816,
+      "learning_rate": 0.002482764170088546,
+      "loss": 1.2971,
+      "step": 3544
+    },
+    {
+      "epoch": 0.27267133297438656,
+      "learning_rate": 0.0024824903087989855,
+      "loss": 1.4864,
+      "step": 3545
+    },
+    {
+      "epoch": 0.272748250134605,
+      "learning_rate": 0.0024822163901407914,
+      "loss": 1.2757,
+      "step": 3546
+    },
+    {
+      "epoch": 0.2728251672948235,
+      "learning_rate": 0.0024819424141299586,
+      "loss": 1.1284,
+      "step": 3547
+    },
+    {
+      "epoch": 0.27290208445504194,
+      "learning_rate": 0.002481668380782485,
+      "loss": 1.2827,
+      "step": 3548
+    },
+    {
+      "epoch": 0.27297900161526034,
+      "learning_rate": 0.0024813942901143722,
+      "loss": 1.1603,
+      "step": 3549
+    },
+    {
+      "epoch": 0.2730559187754788,
+      "learning_rate": 0.002481120142141623,
+      "loss": 1.0989,
+      "step": 3550
+    },
+    {
+      "epoch": 0.27313283593569726,
+      "learning_rate": 0.0024808459368802467,
+      "loss": 1.1768,
+      "step": 3551
+    },
+    {
+      "epoch": 0.2732097530959157,
+      "learning_rate": 0.002480571674346254,
+      "loss": 1.3563,
+      "step": 3552
+    },
+    {
+      "epoch": 0.2732866702561341,
+      "learning_rate": 0.00248029735455566,
+      "loss": 1.5579,
+      "step": 3553
+    },
+    {
+      "epoch": 0.2733635874163526,
+      "learning_rate": 0.0024800229775244803,
+      "loss": 1.052,
+      "step": 3554
+    },
+    {
+      "epoch": 0.27344050457657104,
+      "learning_rate": 0.0024797485432687385,
+      "loss": 0.9302,
+      "step": 3555
+    },
+    {
+      "epoch": 0.2735174217367895,
+      "learning_rate": 0.002479474051804458,
+      "loss": 1.2097,
+      "step": 3556
+    },
+    {
+      "epoch": 0.2735943388970079,
+      "learning_rate": 0.002479199503147667,
+      "loss": 0.9372,
+      "step": 3557
+    },
+    {
+      "epoch": 0.27367125605722636,
+      "learning_rate": 0.0024789248973143964,
+      "loss": 0.8427,
+      "step": 3558
+    },
+    {
+      "epoch": 0.2737481732174448,
+      "learning_rate": 0.0024786502343206807,
+      "loss": 1.0251,
+      "step": 3559
+    },
+    {
+      "epoch": 0.2738250903776633,
+      "learning_rate": 0.0024783755141825578,
+      "loss": 1.4592,
+      "step": 3560
+    },
+    {
+      "epoch": 0.2739020075378817,
+      "learning_rate": 0.0024781007369160694,
+      "loss": 1.468,
+      "step": 3561
+    },
+    {
+      "epoch": 0.27397892469810015,
+      "learning_rate": 0.0024778259025372592,
+      "loss": 0.9707,
+      "step": 3562
+    },
+    {
+      "epoch": 0.2740558418583186,
+      "learning_rate": 0.002477551011062176,
+      "loss": 1.1006,
+      "step": 3563
+    },
+    {
+      "epoch": 0.274132759018537,
+      "learning_rate": 0.0024772760625068704,
+      "loss": 1.2722,
+      "step": 3564
+    },
+    {
+      "epoch": 0.27420967617875547,
+      "learning_rate": 0.0024770010568873966,
+      "loss": 1.1913,
+      "step": 3565
+    },
+    {
+      "epoch": 0.2742865933389739,
+      "learning_rate": 0.002476725994219813,
+      "loss": 1.4452,
+      "step": 3566
+    },
+    {
+      "epoch": 0.2743635104991924,
+      "learning_rate": 0.002476450874520181,
+      "loss": 1.3885,
+      "step": 3567
+    },
+    {
+      "epoch": 0.2744404276594108,
+      "learning_rate": 0.0024761756978045643,
+      "loss": 0.9229,
+      "step": 3568
+    },
+    {
+      "epoch": 0.27451734481962925,
+      "learning_rate": 0.0024759004640890316,
+      "loss": 1.1716,
+      "step": 3569
+    },
+    {
+      "epoch": 0.2745942619798477,
+      "learning_rate": 0.0024756251733896537,
+      "loss": 1.1123,
+      "step": 3570
+    },
+    {
+      "epoch": 0.27467117914006617,
+      "learning_rate": 0.002475349825722505,
+      "loss": 1.6297,
+      "step": 3571
+    },
+    {
+      "epoch": 0.27474809630028457,
+      "learning_rate": 0.002475074421103664,
+      "loss": 0.9711,
+      "step": 3572
+    },
+    {
+      "epoch": 0.27482501346050303,
+      "learning_rate": 0.0024747989595492108,
+      "loss": 1.18,
+      "step": 3573
+    },
+    {
+      "epoch": 0.2749019306207215,
+      "learning_rate": 0.0024745234410752303,
+      "loss": 1.1942,
+      "step": 3574
+    },
+    {
+      "epoch": 0.27497884778093995,
+      "learning_rate": 0.0024742478656978105,
+      "loss": 1.2194,
+      "step": 3575
+    },
+    {
+      "epoch": 0.27505576494115835,
+      "learning_rate": 0.0024739722334330434,
+      "loss": 0.9593,
+      "step": 3576
+    },
+    {
+      "epoch": 0.2751326821013768,
+      "learning_rate": 0.0024736965442970216,
+      "loss": 1.352,
+      "step": 3577
+    },
+    {
+      "epoch": 0.2752095992615953,
+      "learning_rate": 0.0024734207983058442,
+      "loss": 1.0481,
+      "step": 3578
+    },
+    {
+      "epoch": 0.27528651642181373,
+      "learning_rate": 0.0024731449954756117,
+      "loss": 1.1783,
+      "step": 3579
+    },
+    {
+      "epoch": 0.27536343358203214,
+      "learning_rate": 0.002472869135822429,
+      "loss": 1.1786,
+      "step": 3580
+    },
+    {
+      "epoch": 0.2754403507422506,
+      "learning_rate": 0.0024725932193624036,
+      "loss": 0.9528,
+      "step": 3581
+    },
+    {
+      "epoch": 0.27551726790246905,
+      "learning_rate": 0.002472317246111646,
+      "loss": 1.1879,
+      "step": 3582
+    },
+    {
+      "epoch": 0.27559418506268746,
+      "learning_rate": 0.0024720412160862716,
+      "loss": 0.9679,
+      "step": 3583
+    },
+    {
+      "epoch": 0.2756711022229059,
+      "learning_rate": 0.0024717651293023977,
+      "loss": 1.1578,
+      "step": 3584
+    },
+    {
+      "epoch": 0.2757480193831244,
+      "learning_rate": 0.0024714889857761443,
+      "loss": 1.125,
+      "step": 3585
+    },
+    {
+      "epoch": 0.27582493654334284,
+      "learning_rate": 0.002471212785523637,
+      "loss": 1.2239,
+      "step": 3586
+    },
+    {
+      "epoch": 0.27590185370356124,
+      "learning_rate": 0.002470936528561003,
+      "loss": 1.25,
+      "step": 3587
+    },
+    {
+      "epoch": 0.2759787708637797,
+      "learning_rate": 0.0024706602149043733,
+      "loss": 1.1979,
+      "step": 3588
+    },
+    {
+      "epoch": 0.27605568802399816,
+      "learning_rate": 0.002470383844569882,
+      "loss": 1.1033,
+      "step": 3589
+    },
+    {
+      "epoch": 0.2761326051842166,
+      "learning_rate": 0.0024701074175736666,
+      "loss": 1.0717,
+      "step": 3590
+    },
+    {
+      "epoch": 0.276209522344435,
+      "learning_rate": 0.0024698309339318686,
+      "loss": 1.2374,
+      "step": 3591
+    },
+    {
+      "epoch": 0.2762864395046535,
+      "learning_rate": 0.0024695543936606307,
+      "loss": 1.2066,
+      "step": 3592
+    },
+    {
+      "epoch": 0.27636335666487194,
+      "learning_rate": 0.002469277796776102,
+      "loss": 1.4984,
+      "step": 3593
+    },
+    {
+      "epoch": 0.2764402738250904,
+      "learning_rate": 0.0024690011432944325,
+      "loss": 1.2349,
+      "step": 3594
+    },
+    {
+      "epoch": 0.2765171909853088,
+      "learning_rate": 0.0024687244332317762,
+      "loss": 1.0037,
+      "step": 3595
+    },
+    {
+      "epoch": 0.27659410814552726,
+      "learning_rate": 0.0024684476666042908,
+      "loss": 1.0457,
+      "step": 3596
+    },
+    {
+      "epoch": 0.2766710253057457,
+      "learning_rate": 0.002468170843428137,
+      "loss": 0.9947,
+      "step": 3597
+    },
+    {
+      "epoch": 0.2767479424659642,
+      "learning_rate": 0.002467893963719478,
+      "loss": 1.3606,
+      "step": 3598
+    },
+    {
+      "epoch": 0.2768248596261826,
+      "learning_rate": 0.0024676170274944825,
+      "loss": 1.1684,
+      "step": 3599
+    },
+    {
+      "epoch": 0.27690177678640104,
+      "learning_rate": 0.00246734003476932,
+      "loss": 1.1067,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2769786939466195,
+      "learning_rate": 0.0024670629855601646,
+      "loss": 1.1872,
+      "step": 3601
+    },
+    {
+      "epoch": 0.27705561110683796,
+      "learning_rate": 0.0024667858798831936,
+      "loss": 0.9808,
+      "step": 3602
+    },
+    {
+      "epoch": 0.27713252826705637,
+      "learning_rate": 0.002466508717754588,
+      "loss": 1.1883,
+      "step": 3603
+    },
+    {
+      "epoch": 0.2772094454272748,
+      "learning_rate": 0.0024662314991905306,
+      "loss": 1.243,
+      "step": 3604
+    },
+    {
+      "epoch": 0.2772863625874933,
+      "learning_rate": 0.0024659542242072095,
+      "loss": 1.1596,
+      "step": 3605
+    },
+    {
+      "epoch": 0.2773632797477117,
+      "learning_rate": 0.0024656768928208142,
+      "loss": 1.1401,
+      "step": 3606
+    },
+    {
+      "epoch": 0.27744019690793015,
+      "learning_rate": 0.0024653995050475386,
+      "loss": 1.2892,
+      "step": 3607
+    },
+    {
+      "epoch": 0.2775171140681486,
+      "learning_rate": 0.00246512206090358,
+      "loss": 0.7485,
+      "step": 3608
+    },
+    {
+      "epoch": 0.27759403122836707,
+      "learning_rate": 0.002464844560405138,
+      "loss": 0.9778,
+      "step": 3609
+    },
+    {
+      "epoch": 0.27767094838858547,
+      "learning_rate": 0.0024645670035684172,
+      "loss": 1.1943,
+      "step": 3610
+    },
+    {
+      "epoch": 0.27774786554880393,
+      "learning_rate": 0.0024642893904096236,
+      "loss": 0.7985,
+      "step": 3611
+    },
+    {
+      "epoch": 0.2778247827090224,
+      "learning_rate": 0.0024640117209449677,
+      "loss": 0.995,
+      "step": 3612
+    },
+    {
+      "epoch": 0.27790169986924085,
+      "learning_rate": 0.002463733995190662,
+      "loss": 1.1795,
+      "step": 3613
+    },
+    {
+      "epoch": 0.27797861702945925,
+      "learning_rate": 0.002463456213162925,
+      "loss": 1.114,
+      "step": 3614
+    },
+    {
+      "epoch": 0.2780555341896777,
+      "learning_rate": 0.002463178374877975,
+      "loss": 1.2579,
+      "step": 3615
+    },
+    {
+      "epoch": 0.27813245134989617,
+      "learning_rate": 0.002462900480352036,
+      "loss": 1.1171,
+      "step": 3616
+    },
+    {
+      "epoch": 0.27820936851011463,
+      "learning_rate": 0.0024626225296013344,
+      "loss": 0.9647,
+      "step": 3617
+    },
+    {
+      "epoch": 0.27828628567033303,
+      "learning_rate": 0.0024623445226421006,
+      "loss": 1.0202,
+      "step": 3618
+    },
+    {
+      "epoch": 0.2783632028305515,
+      "learning_rate": 0.002462066459490566,
+      "loss": 1.4352,
+      "step": 3619
+    },
+    {
+      "epoch": 0.27844011999076995,
+      "learning_rate": 0.0024617883401629693,
+      "loss": 1.032,
+      "step": 3620
+    },
+    {
+      "epoch": 0.2785170371509884,
+      "learning_rate": 0.0024615101646755485,
+      "loss": 1.2368,
+      "step": 3621
+    },
+    {
+      "epoch": 0.2785939543112068,
+      "learning_rate": 0.0024612319330445474,
+      "loss": 0.7482,
+      "step": 3622
+    },
+    {
+      "epoch": 0.2786708714714253,
+      "learning_rate": 0.002460953645286212,
+      "loss": 0.9346,
+      "step": 3623
+    },
+    {
+      "epoch": 0.27874778863164373,
+      "learning_rate": 0.0024606753014167915,
+      "loss": 0.919,
+      "step": 3624
+    },
+    {
+      "epoch": 0.27882470579186214,
+      "learning_rate": 0.002460396901452539,
+      "loss": 1.1329,
+      "step": 3625
+    },
+    {
+      "epoch": 0.2789016229520806,
+      "learning_rate": 0.0024601184454097107,
+      "loss": 1.5333,
+      "step": 3626
+    },
+    {
+      "epoch": 0.27897854011229906,
+      "learning_rate": 0.0024598399333045653,
+      "loss": 1.2765,
+      "step": 3627
+    },
+    {
+      "epoch": 0.2790554572725175,
+      "learning_rate": 0.002459561365153366,
+      "loss": 1.0334,
+      "step": 3628
+    },
+    {
+      "epoch": 0.2791323744327359,
+      "learning_rate": 0.002459282740972379,
+      "loss": 1.2858,
+      "step": 3629
+    },
+    {
+      "epoch": 0.2792092915929544,
+      "learning_rate": 0.002459004060777873,
+      "loss": 1.2152,
+      "step": 3630
+    },
+    {
+      "epoch": 0.27928620875317284,
+      "learning_rate": 0.0024587253245861207,
+      "loss": 1.3838,
+      "step": 3631
+    },
+    {
+      "epoch": 0.2793631259133913,
+      "learning_rate": 0.0024584465324133968,
+      "loss": 0.9945,
+      "step": 3632
+    },
+    {
+      "epoch": 0.2794400430736097,
+      "learning_rate": 0.002458167684275981,
+      "loss": 1.118,
+      "step": 3633
+    },
+    {
+      "epoch": 0.27951696023382816,
+      "learning_rate": 0.002457888780190156,
+      "loss": 1.2525,
+      "step": 3634
+    },
+    {
+      "epoch": 0.2795938773940466,
+      "learning_rate": 0.0024576098201722065,
+      "loss": 1.3401,
+      "step": 3635
+    },
+    {
+      "epoch": 0.2796707945542651,
+      "learning_rate": 0.0024573308042384217,
+      "loss": 1.0822,
+      "step": 3636
+    },
+    {
+      "epoch": 0.2797477117144835,
+      "learning_rate": 0.002457051732405094,
+      "loss": 1.2132,
+      "step": 3637
+    },
+    {
+      "epoch": 0.27982462887470194,
+      "learning_rate": 0.002456772604688518,
+      "loss": 1.1069,
+      "step": 3638
+    },
+    {
+      "epoch": 0.2799015460349204,
+      "learning_rate": 0.0024564934211049917,
+      "loss": 0.8862,
+      "step": 3639
+    },
+    {
+      "epoch": 0.27997846319513886,
+      "learning_rate": 0.0024562141816708185,
+      "loss": 1.2245,
+      "step": 3640
+    },
+    {
+      "epoch": 0.28005538035535726,
+      "learning_rate": 0.002455934886402303,
+      "loss": 1.2619,
+      "step": 3641
+    },
+    {
+      "epoch": 0.2801322975155757,
+      "learning_rate": 0.0024556555353157524,
+      "loss": 1.1116,
+      "step": 3642
+    },
+    {
+      "epoch": 0.2802092146757942,
+      "learning_rate": 0.0024553761284274796,
+      "loss": 1.1542,
+      "step": 3643
+    },
+    {
+      "epoch": 0.28028613183601264,
+      "learning_rate": 0.002455096665753799,
+      "loss": 1.2643,
+      "step": 3644
+    },
+    {
+      "epoch": 0.28036304899623105,
+      "learning_rate": 0.0024548171473110286,
+      "loss": 1.1555,
+      "step": 3645
+    },
+    {
+      "epoch": 0.2804399661564495,
+      "learning_rate": 0.0024545375731154897,
+      "loss": 0.8542,
+      "step": 3646
+    },
+    {
+      "epoch": 0.28051688331666796,
+      "learning_rate": 0.002454257943183507,
+      "loss": 2.0818,
+      "step": 3647
+    },
+    {
+      "epoch": 0.28059380047688637,
+      "learning_rate": 0.0024539782575314093,
+      "loss": 1.2558,
+      "step": 3648
+    },
+    {
+      "epoch": 0.2806707176371048,
+      "learning_rate": 0.002453698516175526,
+      "loss": 0.8071,
+      "step": 3649
+    },
+    {
+      "epoch": 0.2807476347973233,
+      "learning_rate": 0.0024534187191321928,
+      "loss": 1.1587,
+      "step": 3650
+    },
+    {
+      "epoch": 0.28082455195754175,
+      "learning_rate": 0.002453138866417747,
+      "loss": 1.2945,
+      "step": 3651
+    },
+    {
+      "epoch": 0.28090146911776015,
+      "learning_rate": 0.00245285895804853,
+      "loss": 1.2694,
+      "step": 3652
+    },
+    {
+      "epoch": 0.2809783862779786,
+      "learning_rate": 0.0024525789940408843,
+      "loss": 0.9074,
+      "step": 3653
+    },
+    {
+      "epoch": 0.28105530343819707,
+      "learning_rate": 0.0024522989744111593,
+      "loss": 1.0224,
+      "step": 3654
+    },
+    {
+      "epoch": 0.2811322205984155,
+      "learning_rate": 0.002452018899175704,
+      "loss": 1.1235,
+      "step": 3655
+    },
+    {
+      "epoch": 0.28120913775863393,
+      "learning_rate": 0.0024517387683508736,
+      "loss": 1.1201,
+      "step": 3656
+    },
+    {
+      "epoch": 0.2812860549188524,
+      "learning_rate": 0.002451458581953024,
+      "loss": 1.1645,
+      "step": 3657
+    },
+    {
+      "epoch": 0.28136297207907085,
+      "learning_rate": 0.002451178339998517,
+      "loss": 1.1267,
+      "step": 3658
+    },
+    {
+      "epoch": 0.2814398892392893,
+      "learning_rate": 0.002450898042503715,
+      "loss": 1.298,
+      "step": 3659
+    },
+    {
+      "epoch": 0.2815168063995077,
+      "learning_rate": 0.002450617689484985,
+      "loss": 1.2631,
+      "step": 3660
+    },
+    {
+      "epoch": 0.2815937235597262,
+      "learning_rate": 0.0024503372809586976,
+      "loss": 1.2871,
+      "step": 3661
+    },
+    {
+      "epoch": 0.28167064071994463,
+      "learning_rate": 0.002450056816941226,
+      "loss": 1.3321,
+      "step": 3662
+    },
+    {
+      "epoch": 0.2817475578801631,
+      "learning_rate": 0.002449776297448947,
+      "loss": 0.9775,
+      "step": 3663
+    },
+    {
+      "epoch": 0.2818244750403815,
+      "learning_rate": 0.00244949572249824,
+      "loss": 1.445,
+      "step": 3664
+    },
+    {
+      "epoch": 0.28190139220059995,
+      "learning_rate": 0.002449215092105488,
+      "loss": 0.9526,
+      "step": 3665
+    },
+    {
+      "epoch": 0.2819783093608184,
+      "learning_rate": 0.002448934406287078,
+      "loss": 1.1173,
+      "step": 3666
+    },
+    {
+      "epoch": 0.2820552265210368,
+      "learning_rate": 0.0024486536650593985,
+      "loss": 1.3006,
+      "step": 3667
+    },
+    {
+      "epoch": 0.2821321436812553,
+      "learning_rate": 0.0024483728684388433,
+      "loss": 0.9413,
+      "step": 3668
+    },
+    {
+      "epoch": 0.28220906084147374,
+      "learning_rate": 0.002448092016441808,
+      "loss": 1.1158,
+      "step": 3669
+    },
+    {
+      "epoch": 0.2822859780016922,
+      "learning_rate": 0.002447811109084691,
+      "loss": 1.1066,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2823628951619106,
+      "learning_rate": 0.002447530146383896,
+      "loss": 1.2595,
+      "step": 3671
+    },
+    {
+      "epoch": 0.28243981232212906,
+      "learning_rate": 0.0024472491283558284,
+      "loss": 0.7838,
+      "step": 3672
+    },
+    {
+      "epoch": 0.2825167294823475,
+      "learning_rate": 0.0024469680550168967,
+      "loss": 1.1258,
+      "step": 3673
+    },
+    {
+      "epoch": 0.282593646642566,
+      "learning_rate": 0.002446686926383514,
+      "loss": 0.934,
+      "step": 3674
+    },
+    {
+      "epoch": 0.2826705638027844,
+      "learning_rate": 0.002446405742472095,
+      "loss": 1.1691,
+      "step": 3675
+    },
+    {
+      "epoch": 0.28274748096300284,
+      "learning_rate": 0.0024461245032990574,
+      "loss": 1.3394,
+      "step": 3676
+    },
+    {
+      "epoch": 0.2828243981232213,
+      "learning_rate": 0.002445843208880825,
+      "loss": 1.3558,
+      "step": 3677
+    },
+    {
+      "epoch": 0.28290131528343976,
+      "learning_rate": 0.0024455618592338214,
+      "loss": 1.2262,
+      "step": 3678
+    },
+    {
+      "epoch": 0.28297823244365816,
+      "learning_rate": 0.0024452804543744757,
+      "loss": 1.4341,
+      "step": 3679
+    },
+    {
+      "epoch": 0.2830551496038766,
+      "learning_rate": 0.002444998994319219,
+      "loss": 1.3195,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2831320667640951,
+      "learning_rate": 0.002444717479084486,
+      "loss": 1.0892,
+      "step": 3681
+    },
+    {
+      "epoch": 0.28320898392431354,
+      "learning_rate": 0.002444435908686715,
+      "loss": 1.1523,
+      "step": 3682
+    },
+    {
+      "epoch": 0.28328590108453194,
+      "learning_rate": 0.0024441542831423473,
+      "loss": 0.8263,
+      "step": 3683
+    },
+    {
+      "epoch": 0.2833628182447504,
+      "learning_rate": 0.002443872602467826,
+      "loss": 1.1053,
+      "step": 3684
+    },
+    {
+      "epoch": 0.28343973540496886,
+      "learning_rate": 0.0024435908666796007,
+      "loss": 0.9128,
+      "step": 3685
+    },
+    {
+      "epoch": 0.28351665256518727,
+      "learning_rate": 0.0024433090757941217,
+      "loss": 1.2603,
+      "step": 3686
+    },
+    {
+      "epoch": 0.2835935697254057,
+      "learning_rate": 0.002443027229827842,
+      "loss": 1.3255,
+      "step": 3687
+    },
+    {
+      "epoch": 0.2836704868856242,
+      "learning_rate": 0.002442745328797219,
+      "loss": 0.8551,
+      "step": 3688
+    },
+    {
+      "epoch": 0.28374740404584264,
+      "learning_rate": 0.002442463372718715,
+      "loss": 1.2226,
+      "step": 3689
+    },
+    {
+      "epoch": 0.28382432120606105,
+      "learning_rate": 0.002442181361608792,
+      "loss": 1.0914,
+      "step": 3690
+    },
+    {
+      "epoch": 0.2839012383662795,
+      "learning_rate": 0.002441899295483917,
+      "loss": 1.2384,
+      "step": 3691
+    },
+    {
+      "epoch": 0.28397815552649797,
+      "learning_rate": 0.002441617174360561,
+      "loss": 1.3121,
+      "step": 3692
+    },
+    {
+      "epoch": 0.2840550726867164,
+      "learning_rate": 0.002441334998255196,
+      "loss": 1.0933,
+      "step": 3693
+    },
+    {
+      "epoch": 0.28413198984693483,
+      "learning_rate": 0.0024410527671843004,
+      "loss": 0.8185,
+      "step": 3694
+    },
+    {
+      "epoch": 0.2842089070071533,
+      "learning_rate": 0.0024407704811643524,
+      "loss": 0.9503,
+      "step": 3695
+    },
+    {
+      "epoch": 0.28428582416737175,
+      "learning_rate": 0.0024404881402118354,
+      "loss": 1.2215,
+      "step": 3696
+    },
+    {
+      "epoch": 0.2843627413275902,
+      "learning_rate": 0.0024402057443432363,
+      "loss": 0.7783,
+      "step": 3697
+    },
+    {
+      "epoch": 0.2844396584878086,
+      "learning_rate": 0.002439923293575044,
+      "loss": 0.9716,
+      "step": 3698
+    },
+    {
+      "epoch": 0.28451657564802707,
+      "learning_rate": 0.0024396407879237504,
+      "loss": 1.3816,
+      "step": 3699
+    },
+    {
+      "epoch": 0.28459349280824553,
+      "learning_rate": 0.002439358227405852,
+      "loss": 0.9018,
+      "step": 3700
+    },
+    {
+      "epoch": 0.284670409968464,
+      "learning_rate": 0.002439075612037848,
+      "loss": 0.8355,
+      "step": 3701
+    },
+    {
+      "epoch": 0.2847473271286824,
+      "learning_rate": 0.0024387929418362405,
+      "loss": 1.3953,
+      "step": 3702
+    },
+    {
+      "epoch": 0.28482424428890085,
+      "learning_rate": 0.0024385102168175345,
+      "loss": 1.1761,
+      "step": 3703
+    },
+    {
+      "epoch": 0.2849011614491193,
+      "learning_rate": 0.0024382274369982387,
+      "loss": 1.2913,
+      "step": 3704
+    },
+    {
+      "epoch": 0.28497807860933777,
+      "learning_rate": 0.0024379446023948654,
+      "loss": 0.9577,
+      "step": 3705
+    },
+    {
+      "epoch": 0.2850549957695562,
+      "learning_rate": 0.0024376617130239292,
+      "loss": 1.095,
+      "step": 3706
+    },
+    {
+      "epoch": 0.28513191292977463,
+      "learning_rate": 0.002437378768901948,
+      "loss": 1.3237,
+      "step": 3707
+    },
+    {
+      "epoch": 0.2852088300899931,
+      "learning_rate": 0.0024370957700454443,
+      "loss": 1.1775,
+      "step": 3708
+    },
+    {
+      "epoch": 0.2852857472502115,
+      "learning_rate": 0.002436812716470941,
+      "loss": 1.2467,
+      "step": 3709
+    },
+    {
+      "epoch": 0.28536266441042996,
+      "learning_rate": 0.0024365296081949677,
+      "loss": 0.9706,
+      "step": 3710
+    },
+    {
+      "epoch": 0.2854395815706484,
+      "learning_rate": 0.0024362464452340545,
+      "loss": 1.258,
+      "step": 3711
+    },
+    {
+      "epoch": 0.2855164987308669,
+      "learning_rate": 0.002435963227604735,
+      "loss": 1.1666,
+      "step": 3712
+    },
+    {
+      "epoch": 0.2855934158910853,
+      "learning_rate": 0.0024356799553235478,
+      "loss": 0.7968,
+      "step": 3713
+    },
+    {
+      "epoch": 0.28567033305130374,
+      "learning_rate": 0.0024353966284070324,
+      "loss": 1.0997,
+      "step": 3714
+    },
+    {
+      "epoch": 0.2857472502115222,
+      "learning_rate": 0.002435113246871733,
+      "loss": 1.0189,
+      "step": 3715
+    },
+    {
+      "epoch": 0.28582416737174066,
+      "learning_rate": 0.0024348298107341967,
+      "loss": 1.1352,
+      "step": 3716
+    },
+    {
+      "epoch": 0.28590108453195906,
+      "learning_rate": 0.002434546320010974,
+      "loss": 0.9592,
+      "step": 3717
+    },
+    {
+      "epoch": 0.2859780016921775,
+      "learning_rate": 0.002434262774718618,
+      "loss": 1.073,
+      "step": 3718
+    },
+    {
+      "epoch": 0.286054918852396,
+      "learning_rate": 0.002433979174873684,
+      "loss": 1.1065,
+      "step": 3719
+    },
+    {
+      "epoch": 0.28613183601261444,
+      "learning_rate": 0.0024336955204927323,
+      "loss": 1.1247,
+      "step": 3720
+    },
+    {
+      "epoch": 0.28620875317283284,
+      "learning_rate": 0.0024334118115923267,
+      "loss": 1.1813,
+      "step": 3721
+    },
+    {
+      "epoch": 0.2862856703330513,
+      "learning_rate": 0.0024331280481890327,
+      "loss": 1.3859,
+      "step": 3722
+    },
+    {
+      "epoch": 0.28636258749326976,
+      "learning_rate": 0.0024328442302994193,
+      "loss": 1.3367,
+      "step": 3723
+    },
+    {
+      "epoch": 0.2864395046534882,
+      "learning_rate": 0.0024325603579400594,
+      "loss": 1.3159,
+      "step": 3724
+    },
+    {
+      "epoch": 0.2865164218137066,
+      "learning_rate": 0.002432276431127528,
+      "loss": 1.1745,
+      "step": 3725
+    },
+    {
+      "epoch": 0.2865933389739251,
+      "learning_rate": 0.002431992449878404,
+      "loss": 0.9767,
+      "step": 3726
+    },
+    {
+      "epoch": 0.28667025613414354,
+      "learning_rate": 0.00243170841420927,
+      "loss": 1.24,
+      "step": 3727
+    },
+    {
+      "epoch": 0.28674717329436195,
+      "learning_rate": 0.0024314243241367107,
+      "loss": 0.8242,
+      "step": 3728
+    },
+    {
+      "epoch": 0.2868240904545804,
+      "learning_rate": 0.002431140179677314,
+      "loss": 0.8692,
+      "step": 3729
+    },
+    {
+      "epoch": 0.28690100761479886,
+      "learning_rate": 0.002430855980847672,
+      "loss": 1.1353,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2869779247750173,
+      "learning_rate": 0.002430571727664379,
+      "loss": 1.2912,
+      "step": 3731
+    },
+    {
+      "epoch": 0.2870548419352357,
+      "learning_rate": 0.0024302874201440334,
+      "loss": 1.0579,
+      "step": 3732
+    },
+    {
+      "epoch": 0.2871317590954542,
+      "learning_rate": 0.0024300030583032355,
+      "loss": 1.1849,
+      "step": 3733
+    },
+    {
+      "epoch": 0.28720867625567265,
+      "learning_rate": 0.0024297186421585897,
+      "loss": 1.1034,
+      "step": 3734
+    },
+    {
+      "epoch": 0.2872855934158911,
+      "learning_rate": 0.002429434171726704,
+      "loss": 0.9201,
+      "step": 3735
+    },
+    {
+      "epoch": 0.2873625105761095,
+      "learning_rate": 0.002429149647024188,
+      "loss": 1.1395,
+      "step": 3736
+    },
+    {
+      "epoch": 0.28743942773632797,
+      "learning_rate": 0.0024288650680676558,
+      "loss": 1.0924,
+      "step": 3737
+    },
+    {
+      "epoch": 0.2875163448965464,
+      "learning_rate": 0.0024285804348737243,
+      "loss": 1.451,
+      "step": 3738
+    },
+    {
+      "epoch": 0.2875932620567649,
+      "learning_rate": 0.002428295747459013,
+      "loss": 0.8121,
+      "step": 3739
+    },
+    {
+      "epoch": 0.2876701792169833,
+      "learning_rate": 0.0024280110058401464,
+      "loss": 0.8656,
+      "step": 3740
+    },
+    {
+      "epoch": 0.28774709637720175,
+      "learning_rate": 0.0024277262100337493,
+      "loss": 0.8274,
+      "step": 3741
+    },
+    {
+      "epoch": 0.2878240135374202,
+      "learning_rate": 0.0024274413600564523,
+      "loss": 1.1982,
+      "step": 3742
+    },
+    {
+      "epoch": 0.28790093069763867,
+      "learning_rate": 0.0024271564559248874,
+      "loss": 1.1877,
+      "step": 3743
+    },
+    {
+      "epoch": 0.28797784785785707,
+      "learning_rate": 0.0024268714976556915,
+      "loss": 0.7434,
+      "step": 3744
+    },
+    {
+      "epoch": 0.28805476501807553,
+      "learning_rate": 0.0024265864852655024,
+      "loss": 1.0574,
+      "step": 3745
+    },
+    {
+      "epoch": 0.288131682178294,
+      "learning_rate": 0.002426301418770963,
+      "loss": 1.2975,
+      "step": 3746
+    },
+    {
+      "epoch": 0.2882085993385124,
+      "learning_rate": 0.002426016298188718,
+      "loss": 1.3393,
+      "step": 3747
+    },
+    {
+      "epoch": 0.28828551649873085,
+      "learning_rate": 0.0024257311235354163,
+      "loss": 1.0958,
+      "step": 3748
+    },
+    {
+      "epoch": 0.2883624336589493,
+      "learning_rate": 0.00242544589482771,
+      "loss": 0.8638,
+      "step": 3749
+    },
+    {
+      "epoch": 0.2884393508191678,
+      "learning_rate": 0.0024251606120822525,
+      "loss": 1.2106,
+      "step": 3750
+    },
+    {
+      "epoch": 0.2885162679793862,
+      "learning_rate": 0.0024248752753157038,
+      "loss": 1.0024,
+      "step": 3751
+    },
+    {
+      "epoch": 0.28859318513960464,
+      "learning_rate": 0.0024245898845447235,
+      "loss": 0.8348,
+      "step": 3752
+    },
+    {
+      "epoch": 0.2886701022998231,
+      "learning_rate": 0.002424304439785976,
+      "loss": 1.1999,
+      "step": 3753
+    },
+    {
+      "epoch": 0.28874701946004155,
+      "learning_rate": 0.0024240189410561293,
+      "loss": 1.11,
+      "step": 3754
+    },
+    {
+      "epoch": 0.28882393662025996,
+      "learning_rate": 0.0024237333883718533,
+      "loss": 1.1857,
+      "step": 3755
+    },
+    {
+      "epoch": 0.2889008537804784,
+      "learning_rate": 0.002423447781749822,
+      "loss": 1.0494,
+      "step": 3756
+    },
+    {
+      "epoch": 0.2889777709406969,
+      "learning_rate": 0.002423162121206712,
+      "loss": 1.0682,
+      "step": 3757
+    },
+    {
+      "epoch": 0.28905468810091534,
+      "learning_rate": 0.0024228764067592045,
+      "loss": 1.0615,
+      "step": 3758
+    },
+    {
+      "epoch": 0.28913160526113374,
+      "learning_rate": 0.002422590638423982,
+      "loss": 1.4432,
+      "step": 3759
+    },
+    {
+      "epoch": 0.2892085224213522,
+      "learning_rate": 0.00242230481621773,
+      "loss": 1.26,
+      "step": 3760
+    },
+    {
+      "epoch": 0.28928543958157066,
+      "learning_rate": 0.0024220189401571385,
+      "loss": 1.0253,
+      "step": 3761
+    },
+    {
+      "epoch": 0.2893623567417891,
+      "learning_rate": 0.0024217330102589,
+      "loss": 1.2405,
+      "step": 3762
+    },
+    {
+      "epoch": 0.2894392739020075,
+      "learning_rate": 0.0024214470265397116,
+      "loss": 1.1035,
+      "step": 3763
+    },
+    {
+      "epoch": 0.289516191062226,
+      "learning_rate": 0.00242116098901627,
+      "loss": 1.0638,
+      "step": 3764
+    },
+    {
+      "epoch": 0.28959310822244444,
+      "learning_rate": 0.0024208748977052785,
+      "loss": 1.1296,
+      "step": 3765
+    },
+    {
+      "epoch": 0.2896700253826629,
+      "learning_rate": 0.002420588752623442,
+      "loss": 1.0415,
+      "step": 3766
+    },
+    {
+      "epoch": 0.2897469425428813,
+      "learning_rate": 0.0024203025537874686,
+      "loss": 1.5781,
+      "step": 3767
+    },
+    {
+      "epoch": 0.28982385970309976,
+      "learning_rate": 0.00242001630121407,
+      "loss": 1.0284,
+      "step": 3768
+    },
+    {
+      "epoch": 0.2899007768633182,
+      "learning_rate": 0.002419729994919961,
+      "loss": 1.1952,
+      "step": 3769
+    },
+    {
+      "epoch": 0.2899776940235366,
+      "learning_rate": 0.0024194436349218593,
+      "loss": 0.9078,
+      "step": 3770
+    },
+    {
+      "epoch": 0.2900546111837551,
+      "learning_rate": 0.0024191572212364848,
+      "loss": 0.93,
+      "step": 3771
+    },
+    {
+      "epoch": 0.29013152834397354,
+      "learning_rate": 0.0024188707538805626,
+      "loss": 0.9572,
+      "step": 3772
+    },
+    {
+      "epoch": 0.290208445504192,
+      "learning_rate": 0.0024185842328708195,
+      "loss": 0.8494,
+      "step": 3773
+    },
+    {
+      "epoch": 0.2902853626644104,
+      "learning_rate": 0.0024182976582239855,
+      "loss": 1.2251,
+      "step": 3774
+    },
+    {
+      "epoch": 0.29036227982462887,
+      "learning_rate": 0.0024180110299567947,
+      "loss": 1.0756,
+      "step": 3775
+    },
+    {
+      "epoch": 0.2904391969848473,
+      "learning_rate": 0.0024177243480859828,
+      "loss": 1.2445,
+      "step": 3776
+    },
+    {
+      "epoch": 0.2905161141450658,
+      "learning_rate": 0.00241743761262829,
+      "loss": 1.0082,
+      "step": 3777
+    },
+    {
+      "epoch": 0.2905930313052842,
+      "learning_rate": 0.002417150823600459,
+      "loss": 1.6257,
+      "step": 3778
+    },
+    {
+      "epoch": 0.29066994846550265,
+      "learning_rate": 0.002416863981019235,
+      "loss": 1.2663,
+      "step": 3779
+    },
+    {
+      "epoch": 0.2907468656257211,
+      "learning_rate": 0.0024165770849013677,
+      "loss": 1.21,
+      "step": 3780
+    },
+    {
+      "epoch": 0.29082378278593957,
+      "learning_rate": 0.002416290135263609,
+      "loss": 1.0922,
+      "step": 3781
+    },
+    {
+      "epoch": 0.29090069994615797,
+      "learning_rate": 0.0024160031321227154,
+      "loss": 1.2834,
+      "step": 3782
+    },
+    {
+      "epoch": 0.29097761710637643,
+      "learning_rate": 0.0024157160754954437,
+      "loss": 1.0925,
+      "step": 3783
+    },
+    {
+      "epoch": 0.2910545342665949,
+      "learning_rate": 0.0024154289653985556,
+      "loss": 1.0391,
+      "step": 3784
+    },
+    {
+      "epoch": 0.29113145142681335,
+      "learning_rate": 0.0024151418018488165,
+      "loss": 1.0189,
+      "step": 3785
+    },
+    {
+      "epoch": 0.29120836858703175,
+      "learning_rate": 0.002414854584862994,
+      "loss": 1.3025,
+      "step": 3786
+    },
+    {
+      "epoch": 0.2912852857472502,
+      "learning_rate": 0.002414567314457859,
+      "loss": 1.1334,
+      "step": 3787
+    },
+    {
+      "epoch": 0.29136220290746867,
+      "learning_rate": 0.002414279990650185,
+      "loss": 1.2363,
+      "step": 3788
+    },
+    {
+      "epoch": 0.2914391200676871,
+      "learning_rate": 0.0024139926134567496,
+      "loss": 1.0794,
+      "step": 3789
+    },
+    {
+      "epoch": 0.29151603722790553,
+      "learning_rate": 0.002413705182894333,
+      "loss": 1.3255,
+      "step": 3790
+    },
+    {
+      "epoch": 0.291592954388124,
+      "learning_rate": 0.0024134176989797185,
+      "loss": 0.7982,
+      "step": 3791
+    },
+    {
+      "epoch": 0.29166987154834245,
+      "learning_rate": 0.0024131301617296924,
+      "loss": 1.1689,
+      "step": 3792
+    },
+    {
+      "epoch": 0.29174678870856086,
+      "learning_rate": 0.0024128425711610452,
+      "loss": 1.076,
+      "step": 3793
+    },
+    {
+      "epoch": 0.2918237058687793,
+      "learning_rate": 0.002412554927290568,
+      "loss": 0.9094,
+      "step": 3794
+    },
+    {
+      "epoch": 0.2919006230289978,
+      "learning_rate": 0.0024122672301350578,
+      "loss": 1.0881,
+      "step": 3795
+    },
+    {
+      "epoch": 0.29197754018921623,
+      "learning_rate": 0.002411979479711314,
+      "loss": 1.0724,
+      "step": 3796
+    },
+    {
+      "epoch": 0.29205445734943464,
+      "learning_rate": 0.0024116916760361373,
+      "loss": 1.0396,
+      "step": 3797
+    },
+    {
+      "epoch": 0.2921313745096531,
+      "learning_rate": 0.0024114038191263335,
+      "loss": 1.3407,
+      "step": 3798
+    },
+    {
+      "epoch": 0.29220829166987156,
+      "learning_rate": 0.0024111159089987106,
+      "loss": 1.3073,
+      "step": 3799
+    },
+    {
+      "epoch": 0.29228520883009,
+      "learning_rate": 0.002410827945670081,
+      "loss": 1.0673,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2923621259903084,
+      "learning_rate": 0.0024105399291572582,
+      "loss": 1.3231,
+      "step": 3801
+    },
+    {
+      "epoch": 0.2924390431505269,
+      "learning_rate": 0.00241025185947706,
+      "loss": 1.0472,
+      "step": 3802
+    },
+    {
+      "epoch": 0.29251596031074534,
+      "learning_rate": 0.0024099637366463067,
+      "loss": 1.1477,
+      "step": 3803
+    },
+    {
+      "epoch": 0.2925928774709638,
+      "learning_rate": 0.0024096755606818226,
+      "loss": 1.341,
+      "step": 3804
+    },
+    {
+      "epoch": 0.2926697946311822,
+      "learning_rate": 0.0024093873316004344,
+      "loss": 1.3286,
+      "step": 3805
+    },
+    {
+      "epoch": 0.29274671179140066,
+      "learning_rate": 0.0024090990494189724,
+      "loss": 1.1217,
+      "step": 3806
+    },
+    {
+      "epoch": 0.2928236289516191,
+      "learning_rate": 0.0024088107141542695,
+      "loss": 1.1633,
+      "step": 3807
+    },
+    {
+      "epoch": 0.2929005461118376,
+      "learning_rate": 0.0024085223258231616,
+      "loss": 1.3188,
+      "step": 3808
+    },
+    {
+      "epoch": 0.292977463272056,
+      "learning_rate": 0.0024082338844424885,
+      "loss": 1.0036,
+      "step": 3809
+    },
+    {
+      "epoch": 0.29305438043227444,
+      "learning_rate": 0.002407945390029092,
+      "loss": 1.3416,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2931312975924929,
+      "learning_rate": 0.0024076568425998182,
+      "loss": 0.8524,
+      "step": 3811
+    },
+    {
+      "epoch": 0.2932082147527113,
+      "learning_rate": 0.002407368242171516,
+      "loss": 1.2356,
+      "step": 3812
+    },
+    {
+      "epoch": 0.29328513191292976,
+      "learning_rate": 0.002407079588761036,
+      "loss": 1.1089,
+      "step": 3813
+    },
+    {
+      "epoch": 0.2933620490731482,
+      "learning_rate": 0.002406790882385233,
+      "loss": 1.4834,
+      "step": 3814
+    },
+    {
+      "epoch": 0.2934389662333667,
+      "learning_rate": 0.0024065021230609655,
+      "loss": 0.8913,
+      "step": 3815
+    },
+    {
+      "epoch": 0.2935158833935851,
+      "learning_rate": 0.002406213310805095,
+      "loss": 1.3506,
+      "step": 3816
+    },
+    {
+      "epoch": 0.29359280055380355,
+      "learning_rate": 0.002405924445634484,
+      "loss": 0.9484,
+      "step": 3817
+    },
+    {
+      "epoch": 0.293669717714022,
+      "learning_rate": 0.0024056355275660006,
+      "loss": 1.6326,
+      "step": 3818
+    },
+    {
+      "epoch": 0.29374663487424046,
+      "learning_rate": 0.0024053465566165155,
+      "loss": 1.4045,
+      "step": 3819
+    },
+    {
+      "epoch": 0.29382355203445887,
+      "learning_rate": 0.002405057532802901,
+      "loss": 1.3565,
+      "step": 3820
+    },
+    {
+      "epoch": 0.2939004691946773,
+      "learning_rate": 0.002404768456142034,
+      "loss": 1.2735,
+      "step": 3821
+    },
+    {
+      "epoch": 0.2939773863548958,
+      "learning_rate": 0.002404479326650794,
+      "loss": 1.3249,
+      "step": 3822
+    },
+    {
+      "epoch": 0.29405430351511425,
+      "learning_rate": 0.0024041901443460634,
+      "loss": 1.3285,
+      "step": 3823
+    },
+    {
+      "epoch": 0.29413122067533265,
+      "learning_rate": 0.002403900909244728,
+      "loss": 1.3567,
+      "step": 3824
+    },
+    {
+      "epoch": 0.2942081378355511,
+      "learning_rate": 0.0024036116213636766,
+      "loss": 0.9761,
+      "step": 3825
+    },
+    {
+      "epoch": 0.29428505499576957,
+      "learning_rate": 0.002403322280719801,
+      "loss": 0.9587,
+      "step": 3826
+    },
+    {
+      "epoch": 0.294361972155988,
+      "learning_rate": 0.0024030328873299953,
+      "loss": 1.0449,
+      "step": 3827
+    },
+    {
+      "epoch": 0.29443888931620643,
+      "learning_rate": 0.0024027434412111593,
+      "loss": 1.0609,
+      "step": 3828
+    },
+    {
+      "epoch": 0.2945158064764249,
+      "learning_rate": 0.002402453942380192,
+      "loss": 1.0602,
+      "step": 3829
+    },
+    {
+      "epoch": 0.29459272363664335,
+      "learning_rate": 0.002402164390853999,
+      "loss": 0.949,
+      "step": 3830
+    },
+    {
+      "epoch": 0.29466964079686175,
+      "learning_rate": 0.0024018747866494865,
+      "loss": 1.1173,
+      "step": 3831
+    },
+    {
+      "epoch": 0.2947465579570802,
+      "learning_rate": 0.0024015851297835658,
+      "loss": 0.6908,
+      "step": 3832
+    },
+    {
+      "epoch": 0.2948234751172987,
+      "learning_rate": 0.0024012954202731495,
+      "loss": 0.8463,
+      "step": 3833
+    },
+    {
+      "epoch": 0.29490039227751713,
+      "learning_rate": 0.0024010056581351543,
+      "loss": 1.4411,
+      "step": 3834
+    },
+    {
+      "epoch": 0.29497730943773554,
+      "learning_rate": 0.0024007158433865003,
+      "loss": 1.1703,
+      "step": 3835
+    },
+    {
+      "epoch": 0.295054226597954,
+      "learning_rate": 0.0024004259760441087,
+      "loss": 1.1127,
+      "step": 3836
+    },
+    {
+      "epoch": 0.29513114375817245,
+      "learning_rate": 0.0024001360561249066,
+      "loss": 1.3076,
+      "step": 3837
+    },
+    {
+      "epoch": 0.2952080609183909,
+      "learning_rate": 0.002399846083645822,
+      "loss": 0.7568,
+      "step": 3838
+    },
+    {
+      "epoch": 0.2952849780786093,
+      "learning_rate": 0.002399556058623786,
+      "loss": 1.1442,
+      "step": 3839
+    },
+    {
+      "epoch": 0.2953618952388278,
+      "learning_rate": 0.0023992659810757352,
+      "loss": 0.9839,
+      "step": 3840
+    },
+    {
+      "epoch": 0.29543881239904624,
+      "learning_rate": 0.0023989758510186057,
+      "loss": 1.0016,
+      "step": 3841
+    },
+    {
+      "epoch": 0.2955157295592647,
+      "learning_rate": 0.00239868566846934,
+      "loss": 1.1556,
+      "step": 3842
+    },
+    {
+      "epoch": 0.2955926467194831,
+      "learning_rate": 0.0023983954334448813,
+      "loss": 1.1459,
+      "step": 3843
+    },
+    {
+      "epoch": 0.29566956387970156,
+      "learning_rate": 0.0023981051459621766,
+      "loss": 1.6542,
+      "step": 3844
+    },
+    {
+      "epoch": 0.29574648103992,
+      "learning_rate": 0.0023978148060381775,
+      "loss": 1.0692,
+      "step": 3845
+    },
+    {
+      "epoch": 0.2958233982001385,
+      "learning_rate": 0.0023975244136898353,
+      "loss": 0.9232,
+      "step": 3846
+    },
+    {
+      "epoch": 0.2959003153603569,
+      "learning_rate": 0.0023972339689341073,
+      "loss": 1.0787,
+      "step": 3847
+    },
+    {
+      "epoch": 0.29597723252057534,
+      "learning_rate": 0.0023969434717879532,
+      "loss": 1.2244,
+      "step": 3848
+    },
+    {
+      "epoch": 0.2960541496807938,
+      "learning_rate": 0.0023966529222683346,
+      "loss": 1.1639,
+      "step": 3849
+    },
+    {
+      "epoch": 0.2961310668410122,
+      "learning_rate": 0.0023963623203922175,
+      "loss": 1.1043,
+      "step": 3850
+    },
+    {
+      "epoch": 0.29620798400123066,
+      "learning_rate": 0.0023960716661765706,
+      "loss": 1.1608,
+      "step": 3851
+    },
+    {
+      "epoch": 0.2962849011614491,
+      "learning_rate": 0.002395780959638365,
+      "loss": 1.0279,
+      "step": 3852
+    },
+    {
+      "epoch": 0.2963618183216676,
+      "learning_rate": 0.0023954902007945755,
+      "loss": 1.1058,
+      "step": 3853
+    },
+    {
+      "epoch": 0.296438735481886,
+      "learning_rate": 0.00239519938966218,
+      "loss": 1.2928,
+      "step": 3854
+    },
+    {
+      "epoch": 0.29651565264210444,
+      "learning_rate": 0.0023949085262581597,
+      "loss": 1.0284,
+      "step": 3855
+    },
+    {
+      "epoch": 0.2965925698023229,
+      "learning_rate": 0.0023946176105994967,
+      "loss": 1.5161,
+      "step": 3856
+    },
+    {
+      "epoch": 0.29666948696254136,
+      "learning_rate": 0.00239432664270318,
+      "loss": 1.0126,
+      "step": 3857
+    },
+    {
+      "epoch": 0.29674640412275977,
+      "learning_rate": 0.002394035622586199,
+      "loss": 1.4353,
+      "step": 3858
+    },
+    {
+      "epoch": 0.2968233212829782,
+      "learning_rate": 0.0023937445502655455,
+      "loss": 1.2178,
+      "step": 3859
+    },
+    {
+      "epoch": 0.2969002384431967,
+      "learning_rate": 0.0023934534257582163,
+      "loss": 1.2485,
+      "step": 3860
+    },
+    {
+      "epoch": 0.29697715560341514,
+      "learning_rate": 0.0023931622490812104,
+      "loss": 1.1826,
+      "step": 3861
+    },
+    {
+      "epoch": 0.29705407276363355,
+      "learning_rate": 0.00239287102025153,
+      "loss": 1.4365,
+      "step": 3862
+    },
+    {
+      "epoch": 0.297130989923852,
+      "learning_rate": 0.00239257973928618,
+      "loss": 1.123,
+      "step": 3863
+    },
+    {
+      "epoch": 0.29720790708407047,
+      "learning_rate": 0.00239228840620217,
+      "loss": 1.1501,
+      "step": 3864
+    },
+    {
+      "epoch": 0.2972848242442889,
+      "learning_rate": 0.0023919970210165087,
+      "loss": 1.2852,
+      "step": 3865
+    },
+    {
+      "epoch": 0.29736174140450733,
+      "learning_rate": 0.002391705583746212,
+      "loss": 1.384,
+      "step": 3866
+    },
+    {
+      "epoch": 0.2974386585647258,
+      "learning_rate": 0.0023914140944082973,
+      "loss": 1.5268,
+      "step": 3867
+    },
+    {
+      "epoch": 0.29751557572494425,
+      "learning_rate": 0.0023911225530197843,
+      "loss": 1.1828,
+      "step": 3868
+    },
+    {
+      "epoch": 0.2975924928851627,
+      "learning_rate": 0.0023908309595976965,
+      "loss": 1.4875,
+      "step": 3869
+    },
+    {
+      "epoch": 0.2976694100453811,
+      "learning_rate": 0.0023905393141590613,
+      "loss": 1.1925,
+      "step": 3870
+    },
+    {
+      "epoch": 0.29774632720559957,
+      "learning_rate": 0.0023902476167209077,
+      "loss": 0.9692,
+      "step": 3871
+    },
+    {
+      "epoch": 0.29782324436581803,
+      "learning_rate": 0.002389955867300267,
+      "loss": 0.8897,
+      "step": 3872
+    },
+    {
+      "epoch": 0.29790016152603643,
+      "learning_rate": 0.002389664065914176,
+      "loss": 0.7786,
+      "step": 3873
+    },
+    {
+      "epoch": 0.2979770786862549,
+      "learning_rate": 0.002389372212579673,
+      "loss": 1.1258,
+      "step": 3874
+    },
+    {
+      "epoch": 0.29805399584647335,
+      "learning_rate": 0.0023890803073138,
+      "loss": 1.0717,
+      "step": 3875
+    },
+    {
+      "epoch": 0.2981309130066918,
+      "learning_rate": 0.0023887883501336007,
+      "loss": 0.6882,
+      "step": 3876
+    },
+    {
+      "epoch": 0.2982078301669102,
+      "learning_rate": 0.0023884963410561234,
+      "loss": 1.1122,
+      "step": 3877
+    },
+    {
+      "epoch": 0.2982847473271287,
+      "learning_rate": 0.002388204280098419,
+      "loss": 1.3263,
+      "step": 3878
+    },
+    {
+      "epoch": 0.29836166448734713,
+      "learning_rate": 0.0023879121672775415,
+      "loss": 0.8651,
+      "step": 3879
+    },
+    {
+      "epoch": 0.2984385816475656,
+      "learning_rate": 0.0023876200026105463,
+      "loss": 1.3739,
+      "step": 3880
+    },
+    {
+      "epoch": 0.298515498807784,
+      "learning_rate": 0.002387327786114495,
+      "loss": 0.985,
+      "step": 3881
+    },
+    {
+      "epoch": 0.29859241596800246,
+      "learning_rate": 0.002387035517806449,
+      "loss": 1.0379,
+      "step": 3882
+    },
+    {
+      "epoch": 0.2986693331282209,
+      "learning_rate": 0.0023867431977034744,
+      "loss": 1.0018,
+      "step": 3883
+    },
+    {
+      "epoch": 0.2987462502884394,
+      "learning_rate": 0.0023864508258226408,
+      "loss": 0.9304,
+      "step": 3884
+    },
+    {
+      "epoch": 0.2988231674486578,
+      "learning_rate": 0.0023861584021810196,
+      "loss": 0.8609,
+      "step": 3885
+    },
+    {
+      "epoch": 0.29890008460887624,
+      "learning_rate": 0.002385865926795686,
+      "loss": 1.1605,
+      "step": 3886
+    },
+    {
+      "epoch": 0.2989770017690947,
+      "learning_rate": 0.002385573399683717,
+      "loss": 1.0836,
+      "step": 3887
+    },
+    {
+      "epoch": 0.29905391892931316,
+      "learning_rate": 0.0023852808208621947,
+      "loss": 1.4671,
+      "step": 3888
+    },
+    {
+      "epoch": 0.29913083608953156,
+      "learning_rate": 0.0023849881903482027,
+      "loss": 1.2837,
+      "step": 3889
+    },
+    {
+      "epoch": 0.29920775324975,
+      "learning_rate": 0.0023846955081588276,
+      "loss": 1.079,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2992846704099685,
+      "learning_rate": 0.00238440277431116,
+      "loss": 1.202,
+      "step": 3891
+    },
+    {
+      "epoch": 0.2993615875701869,
+      "learning_rate": 0.002384109988822293,
+      "loss": 1.0768,
+      "step": 3892
+    },
+    {
+      "epoch": 0.29943850473040534,
+      "learning_rate": 0.002383817151709322,
+      "loss": 1.8944,
+      "step": 3893
+    },
+    {
+      "epoch": 0.2995154218906238,
+      "learning_rate": 0.002383524262989346,
+      "loss": 1.1621,
+      "step": 3894
+    },
+    {
+      "epoch": 0.29959233905084226,
+      "learning_rate": 0.002383231322679468,
+      "loss": 0.9345,
+      "step": 3895
+    },
+    {
+      "epoch": 0.29966925621106066,
+      "learning_rate": 0.0023829383307967925,
+      "loss": 1.178,
+      "step": 3896
+    },
+    {
+      "epoch": 0.2997461733712791,
+      "learning_rate": 0.002382645287358428,
+      "loss": 1.193,
+      "step": 3897
+    },
+    {
+      "epoch": 0.2998230905314976,
+      "learning_rate": 0.002382352192381485,
+      "loss": 1.0396,
+      "step": 3898
+    },
+    {
+      "epoch": 0.29990000769171604,
+      "learning_rate": 0.002382059045883078,
+      "loss": 1.4754,
+      "step": 3899
+    },
+    {
+      "epoch": 0.29997692485193445,
+      "learning_rate": 0.0023817658478803246,
+      "loss": 1.4991,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3000538420121529,
+      "learning_rate": 0.0023814725983903437,
+      "loss": 0.7582,
+      "step": 3901
+    },
+    {
+      "epoch": 0.30013075917237136,
+      "learning_rate": 0.002381179297430259,
+      "loss": 1.2912,
+      "step": 3902
+    },
+    {
+      "epoch": 0.3002076763325898,
+      "learning_rate": 0.002380885945017197,
+      "loss": 1.1994,
+      "step": 3903
+    },
+    {
+      "epoch": 0.3002845934928082,
+      "learning_rate": 0.002380592541168287,
+      "loss": 1.0211,
+      "step": 3904
+    },
+    {
+      "epoch": 0.3003615106530267,
+      "learning_rate": 0.00238029908590066,
+      "loss": 1.4471,
+      "step": 3905
+    },
+    {
+      "epoch": 0.30043842781324515,
+      "learning_rate": 0.0023800055792314524,
+      "loss": 1.5325,
+      "step": 3906
+    },
+    {
+      "epoch": 0.3005153449734636,
+      "learning_rate": 0.002379712021177802,
+      "loss": 1.1501,
+      "step": 3907
+    },
+    {
+      "epoch": 0.300592262133682,
+      "learning_rate": 0.00237941841175685,
+      "loss": 0.637,
+      "step": 3908
+    },
+    {
+      "epoch": 0.30066917929390047,
+      "learning_rate": 0.00237912475098574,
+      "loss": 1.2793,
+      "step": 3909
+    },
+    {
+      "epoch": 0.3007460964541189,
+      "learning_rate": 0.0023788310388816202,
+      "loss": 1.1716,
+      "step": 3910
+    },
+    {
+      "epoch": 0.30082301361433733,
+      "learning_rate": 0.00237853727546164,
+      "loss": 1.6392,
+      "step": 3911
+    },
+    {
+      "epoch": 0.3008999307745558,
+      "learning_rate": 0.002378243460742952,
+      "loss": 0.9761,
+      "step": 3912
+    },
+    {
+      "epoch": 0.30097684793477425,
+      "learning_rate": 0.0023779495947427137,
+      "loss": 1.1093,
+      "step": 3913
+    },
+    {
+      "epoch": 0.3010537650949927,
+      "learning_rate": 0.0023776556774780834,
+      "loss": 1.2699,
+      "step": 3914
+    },
+    {
+      "epoch": 0.3011306822552111,
+      "learning_rate": 0.002377361708966223,
+      "loss": 1.3389,
+      "step": 3915
+    },
+    {
+      "epoch": 0.30120759941542957,
+      "learning_rate": 0.002377067689224299,
+      "loss": 1.124,
+      "step": 3916
+    },
+    {
+      "epoch": 0.30128451657564803,
+      "learning_rate": 0.0023767736182694777,
+      "loss": 1.1271,
+      "step": 3917
+    },
+    {
+      "epoch": 0.3013614337358665,
+      "learning_rate": 0.002376479496118931,
+      "loss": 1.1094,
+      "step": 3918
+    },
+    {
+      "epoch": 0.3014383508960849,
+      "learning_rate": 0.0023761853227898338,
+      "loss": 1.0094,
+      "step": 3919
+    },
+    {
+      "epoch": 0.30151526805630335,
+      "learning_rate": 0.0023758910982993623,
+      "loss": 1.157,
+      "step": 3920
+    },
+    {
+      "epoch": 0.3015921852165218,
+      "learning_rate": 0.002375596822664696,
+      "loss": 1.0624,
+      "step": 3921
+    },
+    {
+      "epoch": 0.3016691023767403,
+      "learning_rate": 0.0023753024959030193,
+      "loss": 1.0249,
+      "step": 3922
+    },
+    {
+      "epoch": 0.3017460195369587,
+      "learning_rate": 0.002375008118031518,
+      "loss": 1.2358,
+      "step": 3923
+    },
+    {
+      "epoch": 0.30182293669717714,
+      "learning_rate": 0.0023747136890673805,
+      "loss": 1.0304,
+      "step": 3924
+    },
+    {
+      "epoch": 0.3018998538573956,
+      "learning_rate": 0.0023744192090277986,
+      "loss": 1.3563,
+      "step": 3925
+    },
+    {
+      "epoch": 0.30197677101761405,
+      "learning_rate": 0.0023741246779299687,
+      "loss": 1.2886,
+      "step": 3926
+    },
+    {
+      "epoch": 0.30205368817783246,
+      "learning_rate": 0.002373830095791087,
+      "loss": 1.0715,
+      "step": 3927
+    },
+    {
+      "epoch": 0.3021306053380509,
+      "learning_rate": 0.0023735354626283557,
+      "loss": 1.3066,
+      "step": 3928
+    },
+    {
+      "epoch": 0.3022075224982694,
+      "learning_rate": 0.0023732407784589783,
+      "loss": 1.2763,
+      "step": 3929
+    },
+    {
+      "epoch": 0.30228443965848784,
+      "learning_rate": 0.0023729460433001615,
+      "loss": 1.4041,
+      "step": 3930
+    },
+    {
+      "epoch": 0.30236135681870624,
+      "learning_rate": 0.0023726512571691162,
+      "loss": 1.1763,
+      "step": 3931
+    },
+    {
+      "epoch": 0.3024382739789247,
+      "learning_rate": 0.0023723564200830542,
+      "loss": 1.3388,
+      "step": 3932
+    },
+    {
+      "epoch": 0.30251519113914316,
+      "learning_rate": 0.0023720615320591914,
+      "loss": 1.1631,
+      "step": 3933
+    },
+    {
+      "epoch": 0.30259210829936156,
+      "learning_rate": 0.0023717665931147474,
+      "loss": 0.9775,
+      "step": 3934
+    },
+    {
+      "epoch": 0.30266902545958,
+      "learning_rate": 0.002371471603266943,
+      "loss": 1.1244,
+      "step": 3935
+    },
+    {
+      "epoch": 0.3027459426197985,
+      "learning_rate": 0.0023711765625330037,
+      "loss": 1.2908,
+      "step": 3936
+    },
+    {
+      "epoch": 0.30282285978001694,
+      "learning_rate": 0.002370881470930157,
+      "loss": 0.8563,
+      "step": 3937
+    },
+    {
+      "epoch": 0.30289977694023534,
+      "learning_rate": 0.0023705863284756335,
+      "loss": 1.1576,
+      "step": 3938
+    },
+    {
+      "epoch": 0.3029766941004538,
+      "learning_rate": 0.0023702911351866666,
+      "loss": 1.0741,
+      "step": 3939
+    },
+    {
+      "epoch": 0.30305361126067226,
+      "learning_rate": 0.0023699958910804943,
+      "loss": 0.9602,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3031305284208907,
+      "learning_rate": 0.0023697005961743544,
+      "loss": 1.0266,
+      "step": 3941
+    },
+    {
+      "epoch": 0.3032074455811091,
+      "learning_rate": 0.0023694052504854906,
+      "loss": 1.3567,
+      "step": 3942
+    },
+    {
+      "epoch": 0.3032843627413276,
+      "learning_rate": 0.0023691098540311478,
+      "loss": 1.1743,
+      "step": 3943
+    },
+    {
+      "epoch": 0.30336127990154604,
+      "learning_rate": 0.0023688144068285755,
+      "loss": 1.0754,
+      "step": 3944
+    },
+    {
+      "epoch": 0.3034381970617645,
+      "learning_rate": 0.002368518908895024,
+      "loss": 1.1054,
+      "step": 3945
+    },
+    {
+      "epoch": 0.3035151142219829,
+      "learning_rate": 0.0023682233602477483,
+      "loss": 1.0102,
+      "step": 3946
+    },
+    {
+      "epoch": 0.30359203138220137,
+      "learning_rate": 0.002367927760904006,
+      "loss": 1.1887,
+      "step": 3947
+    },
+    {
+      "epoch": 0.3036689485424198,
+      "learning_rate": 0.002367632110881057,
+      "loss": 0.8444,
+      "step": 3948
+    },
+    {
+      "epoch": 0.3037458657026383,
+      "learning_rate": 0.002367336410196165,
+      "loss": 1.2563,
+      "step": 3949
+    },
+    {
+      "epoch": 0.3038227828628567,
+      "learning_rate": 0.002367040658866596,
+      "loss": 1.0764,
+      "step": 3950
+    },
+    {
+      "epoch": 0.30389970002307515,
+      "learning_rate": 0.00236674485690962,
+      "loss": 1.0052,
+      "step": 3951
+    },
+    {
+      "epoch": 0.3039766171832936,
+      "learning_rate": 0.0023664490043425078,
+      "loss": 0.9344,
+      "step": 3952
+    },
+    {
+      "epoch": 0.304053534343512,
+      "learning_rate": 0.002366153101182535,
+      "loss": 0.9685,
+      "step": 3953
+    },
+    {
+      "epoch": 0.30413045150373047,
+      "learning_rate": 0.0023658571474469807,
+      "loss": 0.9161,
+      "step": 3954
+    },
+    {
+      "epoch": 0.30420736866394893,
+      "learning_rate": 0.0023655611431531247,
+      "loss": 1.4992,
+      "step": 3955
+    },
+    {
+      "epoch": 0.3042842858241674,
+      "learning_rate": 0.0023652650883182512,
+      "loss": 1.3994,
+      "step": 3956
+    },
+    {
+      "epoch": 0.3043612029843858,
+      "learning_rate": 0.002364968982959648,
+      "loss": 1.1665,
+      "step": 3957
+    },
+    {
+      "epoch": 0.30443812014460425,
+      "learning_rate": 0.002364672827094604,
+      "loss": 1.2797,
+      "step": 3958
+    },
+    {
+      "epoch": 0.3045150373048227,
+      "learning_rate": 0.0023643766207404127,
+      "loss": 1.1326,
+      "step": 3959
+    },
+    {
+      "epoch": 0.30459195446504117,
+      "learning_rate": 0.00236408036391437,
+      "loss": 1.1032,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3046688716252596,
+      "learning_rate": 0.002363784056633774,
+      "loss": 1.1207,
+      "step": 3961
+    },
+    {
+      "epoch": 0.30474578878547803,
+      "learning_rate": 0.0023634876989159265,
+      "loss": 0.9382,
+      "step": 3962
+    },
+    {
+      "epoch": 0.3048227059456965,
+      "learning_rate": 0.0023631912907781326,
+      "loss": 0.9859,
+      "step": 3963
+    },
+    {
+      "epoch": 0.30489962310591495,
+      "learning_rate": 0.0023628948322376997,
+      "loss": 1.4293,
+      "step": 3964
+    },
+    {
+      "epoch": 0.30497654026613336,
+      "learning_rate": 0.0023625983233119385,
+      "loss": 1.1807,
+      "step": 3965
+    },
+    {
+      "epoch": 0.3050534574263518,
+      "learning_rate": 0.0023623017640181615,
+      "loss": 1.0156,
+      "step": 3966
+    },
+    {
+      "epoch": 0.3051303745865703,
+      "learning_rate": 0.002362005154373686,
+      "loss": 0.7093,
+      "step": 3967
+    },
+    {
+      "epoch": 0.30520729174678873,
+      "learning_rate": 0.0023617084943958313,
+      "loss": 0.9847,
+      "step": 3968
+    },
+    {
+      "epoch": 0.30528420890700714,
+      "learning_rate": 0.00236141178410192,
+      "loss": 1.0178,
+      "step": 3969
+    },
+    {
+      "epoch": 0.3053611260672256,
+      "learning_rate": 0.0023611150235092766,
+      "loss": 0.8512,
+      "step": 3970
+    },
+    {
+      "epoch": 0.30543804322744406,
+      "learning_rate": 0.0023608182126352295,
+      "loss": 1.0829,
+      "step": 3971
+    },
+    {
+      "epoch": 0.30551496038766246,
+      "learning_rate": 0.00236052135149711,
+      "loss": 1.1115,
+      "step": 3972
+    },
+    {
+      "epoch": 0.3055918775478809,
+      "learning_rate": 0.0023602244401122515,
+      "loss": 1.2082,
+      "step": 3973
+    },
+    {
+      "epoch": 0.3056687947080994,
+      "learning_rate": 0.0023599274784979916,
+      "loss": 1.5141,
+      "step": 3974
+    },
+    {
+      "epoch": 0.30574571186831784,
+      "learning_rate": 0.00235963046667167,
+      "loss": 1.2173,
+      "step": 3975
+    },
+    {
+      "epoch": 0.30582262902853624,
+      "learning_rate": 0.00235933340465063,
+      "loss": 1.3931,
+      "step": 3976
+    },
+    {
+      "epoch": 0.3058995461887547,
+      "learning_rate": 0.0023590362924522166,
+      "loss": 1.0222,
+      "step": 3977
+    },
+    {
+      "epoch": 0.30597646334897316,
+      "learning_rate": 0.0023587391300937794,
+      "loss": 1.5301,
+      "step": 3978
+    },
+    {
+      "epoch": 0.3060533805091916,
+      "learning_rate": 0.0023584419175926683,
+      "loss": 1.0222,
+      "step": 3979
+    },
+    {
+      "epoch": 0.30613029766941,
+      "learning_rate": 0.00235814465496624,
+      "loss": 0.9644,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3062072148296285,
+      "learning_rate": 0.0023578473422318507,
+      "loss": 1.4717,
+      "step": 3981
+    },
+    {
+      "epoch": 0.30628413198984694,
+      "learning_rate": 0.0023575499794068608,
+      "loss": 1.1743,
+      "step": 3982
+    },
+    {
+      "epoch": 0.3063610491500654,
+      "learning_rate": 0.002357252566508634,
+      "loss": 0.9093,
+      "step": 3983
+    },
+    {
+      "epoch": 0.3064379663102838,
+      "learning_rate": 0.002356955103554536,
+      "loss": 1.092,
+      "step": 3984
+    },
+    {
+      "epoch": 0.30651488347050226,
+      "learning_rate": 0.0023566575905619373,
+      "loss": 1.0453,
+      "step": 3985
+    },
+    {
+      "epoch": 0.3065918006307207,
+      "learning_rate": 0.0023563600275482086,
+      "loss": 0.8073,
+      "step": 3986
+    },
+    {
+      "epoch": 0.3066687177909392,
+      "learning_rate": 0.0023560624145307256,
+      "loss": 1.0317,
+      "step": 3987
+    },
+    {
+      "epoch": 0.3067456349511576,
+      "learning_rate": 0.002355764751526866,
+      "loss": 0.9002,
+      "step": 3988
+    },
+    {
+      "epoch": 0.30682255211137605,
+      "learning_rate": 0.0023554670385540107,
+      "loss": 1.3194,
+      "step": 3989
+    },
+    {
+      "epoch": 0.3068994692715945,
+      "learning_rate": 0.0023551692756295433,
+      "loss": 0.9924,
+      "step": 3990
+    },
+    {
+      "epoch": 0.30697638643181296,
+      "learning_rate": 0.0023548714627708504,
+      "loss": 1.0192,
+      "step": 3991
+    },
+    {
+      "epoch": 0.30705330359203137,
+      "learning_rate": 0.0023545735999953225,
+      "loss": 1.0896,
+      "step": 3992
+    },
+    {
+      "epoch": 0.3071302207522498,
+      "learning_rate": 0.002354275687320351,
+      "loss": 1.0743,
+      "step": 3993
+    },
+    {
+      "epoch": 0.3072071379124683,
+      "learning_rate": 0.002353977724763332,
+      "loss": 1.3867,
+      "step": 3994
+    },
+    {
+      "epoch": 0.3072840550726867,
+      "learning_rate": 0.002353679712341664,
+      "loss": 1.0094,
+      "step": 3995
+    },
+    {
+      "epoch": 0.30736097223290515,
+      "learning_rate": 0.0023533816500727472,
+      "loss": 1.2725,
+      "step": 3996
+    },
+    {
+      "epoch": 0.3074378893931236,
+      "learning_rate": 0.002353083537973987,
+      "loss": 1.4228,
+      "step": 3997
+    },
+    {
+      "epoch": 0.30751480655334207,
+      "learning_rate": 0.0023527853760627894,
+      "loss": 1.0509,
+      "step": 3998
+    },
+    {
+      "epoch": 0.30759172371356047,
+      "learning_rate": 0.002352487164356566,
+      "loss": 0.9508,
+      "step": 3999
+    },
+    {
+      "epoch": 0.30766864087377893,
+      "learning_rate": 0.002352188902872728,
+      "loss": 1.188,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3077455580339974,
+      "learning_rate": 0.0023518905916286915,
+      "loss": 1.5387,
+      "step": 4001
+    },
+    {
+      "epoch": 0.30782247519421585,
+      "learning_rate": 0.002351592230641876,
+      "loss": 1.3311,
+      "step": 4002
+    },
+    {
+      "epoch": 0.30789939235443425,
+      "learning_rate": 0.0023512938199297027,
+      "loss": 1.3555,
+      "step": 4003
+    },
+    {
+      "epoch": 0.3079763095146527,
+      "learning_rate": 0.0023509953595095955,
+      "loss": 0.9321,
+      "step": 4004
+    },
+    {
+      "epoch": 0.3080532266748712,
+      "learning_rate": 0.0023506968493989828,
+      "loss": 1.0828,
+      "step": 4005
+    },
+    {
+      "epoch": 0.30813014383508963,
+      "learning_rate": 0.0023503982896152945,
+      "loss": 1.1461,
+      "step": 4006
+    },
+    {
+      "epoch": 0.30820706099530804,
+      "learning_rate": 0.0023500996801759638,
+      "loss": 1.1811,
+      "step": 4007
+    },
+    {
+      "epoch": 0.3082839781555265,
+      "learning_rate": 0.0023498010210984267,
+      "loss": 1.2982,
+      "step": 4008
+    },
+    {
+      "epoch": 0.30836089531574495,
+      "learning_rate": 0.002349502312400122,
+      "loss": 1.238,
+      "step": 4009
+    },
+    {
+      "epoch": 0.3084378124759634,
+      "learning_rate": 0.0023492035540984926,
+      "loss": 1.1188,
+      "step": 4010
+    },
+    {
+      "epoch": 0.3085147296361818,
+      "learning_rate": 0.0023489047462109825,
+      "loss": 1.2138,
+      "step": 4011
+    },
+    {
+      "epoch": 0.3085916467964003,
+      "learning_rate": 0.0023486058887550393,
+      "loss": 1.0183,
+      "step": 4012
+    },
+    {
+      "epoch": 0.30866856395661874,
+      "learning_rate": 0.0023483069817481143,
+      "loss": 1.0868,
+      "step": 4013
+    },
+    {
+      "epoch": 0.30874548111683714,
+      "learning_rate": 0.00234800802520766,
+      "loss": 0.971,
+      "step": 4014
+    },
+    {
+      "epoch": 0.3088223982770556,
+      "learning_rate": 0.0023477090191511337,
+      "loss": 1.1807,
+      "step": 4015
+    },
+    {
+      "epoch": 0.30889931543727406,
+      "learning_rate": 0.002347409963595994,
+      "loss": 1.1903,
+      "step": 4016
+    },
+    {
+      "epoch": 0.3089762325974925,
+      "learning_rate": 0.0023471108585597033,
+      "loss": 1.2067,
+      "step": 4017
+    },
+    {
+      "epoch": 0.3090531497577109,
+      "learning_rate": 0.0023468117040597275,
+      "loss": 1.2638,
+      "step": 4018
+    },
+    {
+      "epoch": 0.3091300669179294,
+      "learning_rate": 0.002346512500113533,
+      "loss": 1.1632,
+      "step": 4019
+    },
+    {
+      "epoch": 0.30920698407814784,
+      "learning_rate": 0.002346213246738592,
+      "loss": 1.1534,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3092839012383663,
+      "learning_rate": 0.002345913943952377,
+      "loss": 0.7224,
+      "step": 4021
+    },
+    {
+      "epoch": 0.3093608183985847,
+      "learning_rate": 0.002345614591772366,
+      "loss": 0.9565,
+      "step": 4022
+    },
+    {
+      "epoch": 0.30943773555880316,
+      "learning_rate": 0.002345315190216037,
+      "loss": 1.1661,
+      "step": 4023
+    },
+    {
+      "epoch": 0.3095146527190216,
+      "learning_rate": 0.0023450157393008733,
+      "loss": 1.1602,
+      "step": 4024
+    },
+    {
+      "epoch": 0.3095915698792401,
+      "learning_rate": 0.00234471623904436,
+      "loss": 0.9932,
+      "step": 4025
+    },
+    {
+      "epoch": 0.3096684870394585,
+      "learning_rate": 0.0023444166894639852,
+      "loss": 1.0469,
+      "step": 4026
+    },
+    {
+      "epoch": 0.30974540419967694,
+      "learning_rate": 0.00234411709057724,
+      "loss": 1.4895,
+      "step": 4027
+    },
+    {
+      "epoch": 0.3098223213598954,
+      "learning_rate": 0.0023438174424016173,
+      "loss": 1.4801,
+      "step": 4028
+    },
+    {
+      "epoch": 0.30989923852011386,
+      "learning_rate": 0.0023435177449546156,
+      "loss": 0.8212,
+      "step": 4029
+    },
+    {
+      "epoch": 0.30997615568033227,
+      "learning_rate": 0.0023432179982537335,
+      "loss": 1.0803,
+      "step": 4030
+    },
+    {
+      "epoch": 0.3100530728405507,
+      "learning_rate": 0.0023429182023164737,
+      "loss": 1.3106,
+      "step": 4031
+    },
+    {
+      "epoch": 0.3101299900007692,
+      "learning_rate": 0.0023426183571603415,
+      "loss": 0.9331,
+      "step": 4032
+    },
+    {
+      "epoch": 0.31020690716098764,
+      "learning_rate": 0.002342318462802845,
+      "loss": 1.1308,
+      "step": 4033
+    },
+    {
+      "epoch": 0.31028382432120605,
+      "learning_rate": 0.002342018519261496,
+      "loss": 1.585,
+      "step": 4034
+    },
+    {
+      "epoch": 0.3103607414814245,
+      "learning_rate": 0.0023417185265538084,
+      "loss": 1.0905,
+      "step": 4035
+    },
+    {
+      "epoch": 0.31043765864164297,
+      "learning_rate": 0.002341418484697298,
+      "loss": 1.3169,
+      "step": 4036
+    },
+    {
+      "epoch": 0.31051457580186137,
+      "learning_rate": 0.002341118393709486,
+      "loss": 1.0269,
+      "step": 4037
+    },
+    {
+      "epoch": 0.31059149296207983,
+      "learning_rate": 0.0023408182536078944,
+      "loss": 1.0275,
+      "step": 4038
+    },
+    {
+      "epoch": 0.3106684101222983,
+      "learning_rate": 0.0023405180644100487,
+      "loss": 1.3213,
+      "step": 4039
+    },
+    {
+      "epoch": 0.31074532728251675,
+      "learning_rate": 0.0023402178261334766,
+      "loss": 1.1029,
+      "step": 4040
+    },
+    {
+      "epoch": 0.31082224444273515,
+      "learning_rate": 0.0023399175387957107,
+      "loss": 1.0443,
+      "step": 4041
+    },
+    {
+      "epoch": 0.3108991616029536,
+      "learning_rate": 0.0023396172024142846,
+      "loss": 1.1436,
+      "step": 4042
+    },
+    {
+      "epoch": 0.31097607876317207,
+      "learning_rate": 0.002339316817006734,
+      "loss": 1.0799,
+      "step": 4043
+    },
+    {
+      "epoch": 0.31105299592339053,
+      "learning_rate": 0.0023390163825906004,
+      "loss": 1.241,
+      "step": 4044
+    },
+    {
+      "epoch": 0.31112991308360893,
+      "learning_rate": 0.002338715899183426,
+      "loss": 1.2144,
+      "step": 4045
+    },
+    {
+      "epoch": 0.3112068302438274,
+      "learning_rate": 0.002338415366802756,
+      "loss": 1.4211,
+      "step": 4046
+    },
+    {
+      "epoch": 0.31128374740404585,
+      "learning_rate": 0.0023381147854661384,
+      "loss": 1.2576,
+      "step": 4047
+    },
+    {
+      "epoch": 0.3113606645642643,
+      "learning_rate": 0.0023378141551911257,
+      "loss": 1.1191,
+      "step": 4048
+    },
+    {
+      "epoch": 0.3114375817244827,
+      "learning_rate": 0.0023375134759952712,
+      "loss": 1.3006,
+      "step": 4049
+    },
+    {
+      "epoch": 0.3115144988847012,
+      "learning_rate": 0.0023372127478961317,
+      "loss": 1.401,
+      "step": 4050
+    },
+    {
+      "epoch": 0.31159141604491963,
+      "learning_rate": 0.0023369119709112673,
+      "loss": 0.8778,
+      "step": 4051
+    },
+    {
+      "epoch": 0.3116683332051381,
+      "learning_rate": 0.0023366111450582407,
+      "loss": 1.3404,
+      "step": 4052
+    },
+    {
+      "epoch": 0.3117452503653565,
+      "learning_rate": 0.0023363102703546183,
+      "loss": 1.3865,
+      "step": 4053
+    },
+    {
+      "epoch": 0.31182216752557496,
+      "learning_rate": 0.0023360093468179667,
+      "loss": 1.2789,
+      "step": 4054
+    },
+    {
+      "epoch": 0.3118990846857934,
+      "learning_rate": 0.0023357083744658586,
+      "loss": 1.0586,
+      "step": 4055
+    },
+    {
+      "epoch": 0.3119760018460118,
+      "learning_rate": 0.0023354073533158673,
+      "loss": 1.1281,
+      "step": 4056
+    },
+    {
+      "epoch": 0.3120529190062303,
+      "learning_rate": 0.00233510628338557,
+      "loss": 1.0259,
+      "step": 4057
+    },
+    {
+      "epoch": 0.31212983616644874,
+      "learning_rate": 0.002334805164692546,
+      "loss": 1.2627,
+      "step": 4058
+    },
+    {
+      "epoch": 0.3122067533266672,
+      "learning_rate": 0.0023345039972543797,
+      "loss": 1.2103,
+      "step": 4059
+    },
+    {
+      "epoch": 0.3122836704868856,
+      "learning_rate": 0.0023342027810886543,
+      "loss": 1.1784,
+      "step": 4060
+    },
+    {
+      "epoch": 0.31236058764710406,
+      "learning_rate": 0.0023339015162129593,
+      "loss": 1.3992,
+      "step": 4061
+    },
+    {
+      "epoch": 0.3124375048073225,
+      "learning_rate": 0.0023336002026448856,
+      "loss": 1.173,
+      "step": 4062
+    },
+    {
+      "epoch": 0.312514421967541,
+      "learning_rate": 0.0023332988404020276,
+      "loss": 1.2913,
+      "step": 4063
+    },
+    {
+      "epoch": 0.3125913391277594,
+      "learning_rate": 0.002332997429501982,
+      "loss": 0.9053,
+      "step": 4064
+    },
+    {
+      "epoch": 0.31266825628797784,
+      "learning_rate": 0.002332695969962348,
+      "loss": 1.0656,
+      "step": 4065
+    },
+    {
+      "epoch": 0.3127451734481963,
+      "learning_rate": 0.0023323944618007285,
+      "loss": 1.3543,
+      "step": 4066
+    },
+    {
+      "epoch": 0.31282209060841476,
+      "learning_rate": 0.002332092905034729,
+      "loss": 1.1413,
+      "step": 4067
+    },
+    {
+      "epoch": 0.31289900776863316,
+      "learning_rate": 0.002331791299681958,
+      "loss": 0.9858,
+      "step": 4068
+    },
+    {
+      "epoch": 0.3129759249288516,
+      "learning_rate": 0.0023314896457600253,
+      "loss": 0.8808,
+      "step": 4069
+    },
+    {
+      "epoch": 0.3130528420890701,
+      "learning_rate": 0.0023311879432865463,
+      "loss": 1.2249,
+      "step": 4070
+    },
+    {
+      "epoch": 0.31312975924928854,
+      "learning_rate": 0.0023308861922791365,
+      "loss": 0.693,
+      "step": 4071
+    },
+    {
+      "epoch": 0.31320667640950695,
+      "learning_rate": 0.0023305843927554164,
+      "loss": 1.0655,
+      "step": 4072
+    },
+    {
+      "epoch": 0.3132835935697254,
+      "learning_rate": 0.0023302825447330078,
+      "loss": 0.8207,
+      "step": 4073
+    },
+    {
+      "epoch": 0.31336051072994386,
+      "learning_rate": 0.0023299806482295357,
+      "loss": 1.6583,
+      "step": 4074
+    },
+    {
+      "epoch": 0.31343742789016227,
+      "learning_rate": 0.0023296787032626295,
+      "loss": 1.1852,
+      "step": 4075
+    },
+    {
+      "epoch": 0.3135143450503807,
+      "learning_rate": 0.0023293767098499186,
+      "loss": 1.0463,
+      "step": 4076
+    },
+    {
+      "epoch": 0.3135912622105992,
+      "learning_rate": 0.0023290746680090374,
+      "loss": 1.3161,
+      "step": 4077
+    },
+    {
+      "epoch": 0.31366817937081765,
+      "learning_rate": 0.0023287725777576223,
+      "loss": 1.1284,
+      "step": 4078
+    },
+    {
+      "epoch": 0.31374509653103605,
+      "learning_rate": 0.002328470439113313,
+      "loss": 1.3051,
+      "step": 4079
+    },
+    {
+      "epoch": 0.3138220136912545,
+      "learning_rate": 0.0023281682520937506,
+      "loss": 1.2625,
+      "step": 4080
+    },
+    {
+      "epoch": 0.31389893085147297,
+      "learning_rate": 0.0023278660167165813,
+      "loss": 0.9103,
+      "step": 4081
+    },
+    {
+      "epoch": 0.3139758480116914,
+      "learning_rate": 0.0023275637329994523,
+      "loss": 1.3384,
+      "step": 4082
+    },
+    {
+      "epoch": 0.31405276517190983,
+      "learning_rate": 0.0023272614009600145,
+      "loss": 1.1082,
+      "step": 4083
+    },
+    {
+      "epoch": 0.3141296823321283,
+      "learning_rate": 0.0023269590206159213,
+      "loss": 1.1432,
+      "step": 4084
+    },
+    {
+      "epoch": 0.31420659949234675,
+      "learning_rate": 0.002326656591984829,
+      "loss": 1.2934,
+      "step": 4085
+    },
+    {
+      "epoch": 0.3142835166525652,
+      "learning_rate": 0.0023263541150843974,
+      "loss": 1.097,
+      "step": 4086
+    },
+    {
+      "epoch": 0.3143604338127836,
+      "learning_rate": 0.0023260515899322875,
+      "loss": 0.9245,
+      "step": 4087
+    },
+    {
+      "epoch": 0.31443735097300207,
+      "learning_rate": 0.0023257490165461644,
+      "loss": 1.3487,
+      "step": 4088
+    },
+    {
+      "epoch": 0.31451426813322053,
+      "learning_rate": 0.002325446394943695,
+      "loss": 1.3961,
+      "step": 4089
+    },
+    {
+      "epoch": 0.314591185293439,
+      "learning_rate": 0.002325143725142551,
+      "loss": 0.8543,
+      "step": 4090
+    },
+    {
+      "epoch": 0.3146681024536574,
+      "learning_rate": 0.002324841007160405,
+      "loss": 1.5301,
+      "step": 4091
+    },
+    {
+      "epoch": 0.31474501961387585,
+      "learning_rate": 0.0023245382410149325,
+      "loss": 1.275,
+      "step": 4092
+    },
+    {
+      "epoch": 0.3148219367740943,
+      "learning_rate": 0.002324235426723813,
+      "loss": 1.4179,
+      "step": 4093
+    },
+    {
+      "epoch": 0.3148988539343128,
+      "learning_rate": 0.002323932564304728,
+      "loss": 1.6431,
+      "step": 4094
+    },
+    {
+      "epoch": 0.3149757710945312,
+      "learning_rate": 0.0023236296537753618,
+      "loss": 1.0404,
+      "step": 4095
+    },
+    {
+      "epoch": 0.31505268825474964,
+      "learning_rate": 0.002323326695153402,
+      "loss": 0.6271,
+      "step": 4096
+    },
+    {
+      "epoch": 0.3151296054149681,
+      "learning_rate": 0.002323023688456538,
+      "loss": 1.2285,
+      "step": 4097
+    },
+    {
+      "epoch": 0.3152065225751865,
+      "learning_rate": 0.0023227206337024637,
+      "loss": 1.1556,
+      "step": 4098
+    },
+    {
+      "epoch": 0.31528343973540496,
+      "learning_rate": 0.0023224175309088735,
+      "loss": 0.8867,
+      "step": 4099
+    },
+    {
+      "epoch": 0.3153603568956234,
+      "learning_rate": 0.002322114380093467,
+      "loss": 1.1383,
+      "step": 4100
+    },
+    {
+      "epoch": 0.3154372740558419,
+      "learning_rate": 0.0023218111812739447,
+      "loss": 1.4028,
+      "step": 4101
+    },
+    {
+      "epoch": 0.3155141912160603,
+      "learning_rate": 0.0023215079344680116,
+      "loss": 1.4083,
+      "step": 4102
+    },
+    {
+      "epoch": 0.31559110837627874,
+      "learning_rate": 0.0023212046396933737,
+      "loss": 1.3346,
+      "step": 4103
+    },
+    {
+      "epoch": 0.3156680255364972,
+      "learning_rate": 0.002320901296967741,
+      "loss": 0.8614,
+      "step": 4104
+    },
+    {
+      "epoch": 0.31574494269671566,
+      "learning_rate": 0.002320597906308826,
+      "loss": 1.1192,
+      "step": 4105
+    },
+    {
+      "epoch": 0.31582185985693406,
+      "learning_rate": 0.0023202944677343444,
+      "loss": 1.0291,
+      "step": 4106
+    },
+    {
+      "epoch": 0.3158987770171525,
+      "learning_rate": 0.0023199909812620136,
+      "loss": 1.2566,
+      "step": 4107
+    },
+    {
+      "epoch": 0.315975694177371,
+      "learning_rate": 0.002319687446909555,
+      "loss": 1.0361,
+      "step": 4108
+    },
+    {
+      "epoch": 0.31605261133758944,
+      "learning_rate": 0.0023193838646946917,
+      "loss": 1.1147,
+      "step": 4109
+    },
+    {
+      "epoch": 0.31612952849780784,
+      "learning_rate": 0.0023190802346351514,
+      "loss": 1.281,
+      "step": 4110
+    },
+    {
+      "epoch": 0.3162064456580263,
+      "learning_rate": 0.002318776556748662,
+      "loss": 1.2085,
+      "step": 4111
+    },
+    {
+      "epoch": 0.31628336281824476,
+      "learning_rate": 0.0023184728310529563,
+      "loss": 1.219,
+      "step": 4112
+    },
+    {
+      "epoch": 0.3163602799784632,
+      "learning_rate": 0.002318169057565769,
+      "loss": 1.1128,
+      "step": 4113
+    },
+    {
+      "epoch": 0.3164371971386816,
+      "learning_rate": 0.002317865236304838,
+      "loss": 1.2199,
+      "step": 4114
+    },
+    {
+      "epoch": 0.3165141142989001,
+      "learning_rate": 0.002317561367287903,
+      "loss": 0.9051,
+      "step": 4115
+    },
+    {
+      "epoch": 0.31659103145911854,
+      "learning_rate": 0.0023172574505327076,
+      "loss": 0.8508,
+      "step": 4116
+    },
+    {
+      "epoch": 0.31666794861933695,
+      "learning_rate": 0.0023169534860569985,
+      "loss": 1.1534,
+      "step": 4117
+    },
+    {
+      "epoch": 0.3167448657795554,
+      "learning_rate": 0.002316649473878524,
+      "loss": 1.2333,
+      "step": 4118
+    },
+    {
+      "epoch": 0.31682178293977387,
+      "learning_rate": 0.0023163454140150354,
+      "loss": 1.0667,
+      "step": 4119
+    },
+    {
+      "epoch": 0.3168987000999923,
+      "learning_rate": 0.002316041306484287,
+      "loss": 1.2747,
+      "step": 4120
+    },
+    {
+      "epoch": 0.31697561726021073,
+      "learning_rate": 0.0023157371513040366,
+      "loss": 1.0607,
+      "step": 4121
+    },
+    {
+      "epoch": 0.3170525344204292,
+      "learning_rate": 0.0023154329484920437,
+      "loss": 1.0513,
+      "step": 4122
+    },
+    {
+      "epoch": 0.31712945158064765,
+      "learning_rate": 0.002315128698066071,
+      "loss": 1.1036,
+      "step": 4123
+    },
+    {
+      "epoch": 0.3172063687408661,
+      "learning_rate": 0.0023148244000438844,
+      "loss": 1.3047,
+      "step": 4124
+    },
+    {
+      "epoch": 0.3172832859010845,
+      "learning_rate": 0.0023145200544432516,
+      "loss": 1.1692,
+      "step": 4125
+    },
+    {
+      "epoch": 0.31736020306130297,
+      "learning_rate": 0.0023142156612819442,
+      "loss": 1.297,
+      "step": 4126
+    },
+    {
+      "epoch": 0.31743712022152143,
+      "learning_rate": 0.0023139112205777354,
+      "loss": 0.9115,
+      "step": 4127
+    },
+    {
+      "epoch": 0.3175140373817399,
+      "learning_rate": 0.0023136067323484026,
+      "loss": 1.3012,
+      "step": 4128
+    },
+    {
+      "epoch": 0.3175909545419583,
+      "learning_rate": 0.002313302196611725,
+      "loss": 1.0477,
+      "step": 4129
+    },
+    {
+      "epoch": 0.31766787170217675,
+      "learning_rate": 0.0023129976133854837,
+      "loss": 0.9798,
+      "step": 4130
+    },
+    {
+      "epoch": 0.3177447888623952,
+      "learning_rate": 0.002312692982687465,
+      "loss": 1.22,
+      "step": 4131
+    },
+    {
+      "epoch": 0.31782170602261367,
+      "learning_rate": 0.002312388304535456,
+      "loss": 1.0055,
+      "step": 4132
+    },
+    {
+      "epoch": 0.3178986231828321,
+      "learning_rate": 0.0023120835789472477,
+      "loss": 1.3632,
+      "step": 4133
+    },
+    {
+      "epoch": 0.31797554034305053,
+      "learning_rate": 0.0023117788059406324,
+      "loss": 1.1977,
+      "step": 4134
+    },
+    {
+      "epoch": 0.318052457503269,
+      "learning_rate": 0.0023114739855334063,
+      "loss": 1.1121,
+      "step": 4135
+    },
+    {
+      "epoch": 0.3181293746634874,
+      "learning_rate": 0.002311169117743369,
+      "loss": 1.1742,
+      "step": 4136
+    },
+    {
+      "epoch": 0.31820629182370586,
+      "learning_rate": 0.0023108642025883213,
+      "loss": 1.1818,
+      "step": 4137
+    },
+    {
+      "epoch": 0.3182832089839243,
+      "learning_rate": 0.002310559240086068,
+      "loss": 1.237,
+      "step": 4138
+    },
+    {
+      "epoch": 0.3183601261441428,
+      "learning_rate": 0.0023102542302544157,
+      "loss": 1.0121,
+      "step": 4139
+    },
+    {
+      "epoch": 0.3184370433043612,
+      "learning_rate": 0.002309949173111174,
+      "loss": 1.2504,
+      "step": 4140
+    },
+    {
+      "epoch": 0.31851396046457964,
+      "learning_rate": 0.0023096440686741567,
+      "loss": 0.679,
+      "step": 4141
+    },
+    {
+      "epoch": 0.3185908776247981,
+      "learning_rate": 0.0023093389169611786,
+      "loss": 1.0136,
+      "step": 4142
+    },
+    {
+      "epoch": 0.31866779478501656,
+      "learning_rate": 0.002309033717990057,
+      "loss": 1.1826,
+      "step": 4143
+    },
+    {
+      "epoch": 0.31874471194523496,
+      "learning_rate": 0.002308728471778614,
+      "loss": 1.2552,
+      "step": 4144
+    },
+    {
+      "epoch": 0.3188216291054534,
+      "learning_rate": 0.0023084231783446723,
+      "loss": 1.3521,
+      "step": 4145
+    },
+    {
+      "epoch": 0.3188985462656719,
+      "learning_rate": 0.0023081178377060588,
+      "loss": 1.0351,
+      "step": 4146
+    },
+    {
+      "epoch": 0.31897546342589034,
+      "learning_rate": 0.0023078124498806026,
+      "loss": 1.1174,
+      "step": 4147
+    },
+    {
+      "epoch": 0.31905238058610874,
+      "learning_rate": 0.0023075070148861356,
+      "loss": 0.9617,
+      "step": 4148
+    },
+    {
+      "epoch": 0.3191292977463272,
+      "learning_rate": 0.0023072015327404924,
+      "loss": 1.0803,
+      "step": 4149
+    },
+    {
+      "epoch": 0.31920621490654566,
+      "learning_rate": 0.0023068960034615107,
+      "loss": 1.105,
+      "step": 4150
+    },
+    {
+      "epoch": 0.3192831320667641,
+      "learning_rate": 0.00230659042706703,
+      "loss": 0.7961,
+      "step": 4151
+    },
+    {
+      "epoch": 0.3193600492269825,
+      "learning_rate": 0.0023062848035748937,
+      "loss": 1.0708,
+      "step": 4152
+    },
+    {
+      "epoch": 0.319436966387201,
+      "learning_rate": 0.0023059791330029477,
+      "loss": 1.3105,
+      "step": 4153
+    },
+    {
+      "epoch": 0.31951388354741944,
+      "learning_rate": 0.0023056734153690398,
+      "loss": 1.6109,
+      "step": 4154
+    },
+    {
+      "epoch": 0.3195908007076379,
+      "learning_rate": 0.002305367650691022,
+      "loss": 1.0377,
+      "step": 4155
+    },
+    {
+      "epoch": 0.3196677178678563,
+      "learning_rate": 0.0023050618389867474,
+      "loss": 1.045,
+      "step": 4156
+    },
+    {
+      "epoch": 0.31974463502807476,
+      "learning_rate": 0.002304755980274073,
+      "loss": 0.939,
+      "step": 4157
+    },
+    {
+      "epoch": 0.3198215521882932,
+      "learning_rate": 0.0023044500745708586,
+      "loss": 0.9277,
+      "step": 4158
+    },
+    {
+      "epoch": 0.3198984693485116,
+      "learning_rate": 0.0023041441218949656,
+      "loss": 1.1165,
+      "step": 4159
+    },
+    {
+      "epoch": 0.3199753865087301,
+      "learning_rate": 0.0023038381222642594,
+      "loss": 1.1278,
+      "step": 4160
+    },
+    {
+      "epoch": 0.32005230366894855,
+      "learning_rate": 0.0023035320756966075,
+      "loss": 1.3572,
+      "step": 4161
+    },
+    {
+      "epoch": 0.320129220829167,
+      "learning_rate": 0.00230322598220988,
+      "loss": 1.1209,
+      "step": 4162
+    },
+    {
+      "epoch": 0.3202061379893854,
+      "learning_rate": 0.0023029198418219507,
+      "loss": 1.0094,
+      "step": 4163
+    },
+    {
+      "epoch": 0.32028305514960387,
+      "learning_rate": 0.002302613654550695,
+      "loss": 1.3123,
+      "step": 4164
+    },
+    {
+      "epoch": 0.3203599723098223,
+      "learning_rate": 0.0023023074204139913,
+      "loss": 1.2515,
+      "step": 4165
+    },
+    {
+      "epoch": 0.3204368894700408,
+      "learning_rate": 0.0023020011394297214,
+      "loss": 1.5766,
+      "step": 4166
+    },
+    {
+      "epoch": 0.3205138066302592,
+      "learning_rate": 0.002301694811615769,
+      "loss": 0.941,
+      "step": 4167
+    },
+    {
+      "epoch": 0.32059072379047765,
+      "learning_rate": 0.002301388436990021,
+      "loss": 1.3693,
+      "step": 4168
+    },
+    {
+      "epoch": 0.3206676409506961,
+      "learning_rate": 0.002301082015570367,
+      "loss": 1.0084,
+      "step": 4169
+    },
+    {
+      "epoch": 0.32074455811091457,
+      "learning_rate": 0.002300775547374699,
+      "loss": 1.3284,
+      "step": 4170
+    },
+    {
+      "epoch": 0.32082147527113297,
+      "learning_rate": 0.002300469032420913,
+      "loss": 1.178,
+      "step": 4171
+    },
+    {
+      "epoch": 0.32089839243135143,
+      "learning_rate": 0.0023001624707269055,
+      "loss": 1.3273,
+      "step": 4172
+    },
+    {
+      "epoch": 0.3209753095915699,
+      "learning_rate": 0.002299855862310577,
+      "loss": 1.3236,
+      "step": 4173
+    },
+    {
+      "epoch": 0.32105222675178835,
+      "learning_rate": 0.0022995492071898323,
+      "loss": 1.0124,
+      "step": 4174
+    },
+    {
+      "epoch": 0.32112914391200675,
+      "learning_rate": 0.002299242505382575,
+      "loss": 1.075,
+      "step": 4175
+    },
+    {
+      "epoch": 0.3212060610722252,
+      "learning_rate": 0.002298935756906715,
+      "loss": 1.2145,
+      "step": 4176
+    },
+    {
+      "epoch": 0.3212829782324437,
+      "learning_rate": 0.0022986289617801642,
+      "loss": 1.1242,
+      "step": 4177
+    },
+    {
+      "epoch": 0.3213598953926621,
+      "learning_rate": 0.002298322120020836,
+      "loss": 1.0867,
+      "step": 4178
+    },
+    {
+      "epoch": 0.32143681255288054,
+      "learning_rate": 0.0022980152316466463,
+      "loss": 0.7148,
+      "step": 4179
+    },
+    {
+      "epoch": 0.321513729713099,
+      "learning_rate": 0.002297708296675516,
+      "loss": 1.2394,
+      "step": 4180
+    },
+    {
+      "epoch": 0.32159064687331745,
+      "learning_rate": 0.002297401315125367,
+      "loss": 1.2044,
+      "step": 4181
+    },
+    {
+      "epoch": 0.32166756403353586,
+      "learning_rate": 0.0022970942870141242,
+      "loss": 0.8243,
+      "step": 4182
+    },
+    {
+      "epoch": 0.3217444811937543,
+      "learning_rate": 0.002296787212359715,
+      "loss": 1.3398,
+      "step": 4183
+    },
+    {
+      "epoch": 0.3218213983539728,
+      "learning_rate": 0.0022964800911800705,
+      "loss": 1.2309,
+      "step": 4184
+    },
+    {
+      "epoch": 0.32189831551419124,
+      "learning_rate": 0.002296172923493123,
+      "loss": 1.1032,
+      "step": 4185
+    },
+    {
+      "epoch": 0.32197523267440964,
+      "learning_rate": 0.002295865709316809,
+      "loss": 1.1799,
+      "step": 4186
+    },
+    {
+      "epoch": 0.3220521498346281,
+      "learning_rate": 0.002295558448669067,
+      "loss": 1.0306,
+      "step": 4187
+    },
+    {
+      "epoch": 0.32212906699484656,
+      "learning_rate": 0.0022952511415678376,
+      "loss": 0.8425,
+      "step": 4188
+    },
+    {
+      "epoch": 0.322205984155065,
+      "learning_rate": 0.002294943788031065,
+      "loss": 0.9645,
+      "step": 4189
+    },
+    {
+      "epoch": 0.3222829013152834,
+      "learning_rate": 0.002294636388076697,
+      "loss": 1.2359,
+      "step": 4190
+    },
+    {
+      "epoch": 0.3223598184755019,
+      "learning_rate": 0.0022943289417226816,
+      "loss": 0.9698,
+      "step": 4191
+    },
+    {
+      "epoch": 0.32243673563572034,
+      "learning_rate": 0.0022940214489869716,
+      "loss": 1.5443,
+      "step": 4192
+    },
+    {
+      "epoch": 0.3225136527959388,
+      "learning_rate": 0.002293713909887522,
+      "loss": 1.3285,
+      "step": 4193
+    },
+    {
+      "epoch": 0.3225905699561572,
+      "learning_rate": 0.00229340632444229,
+      "loss": 1.0976,
+      "step": 4194
+    },
+    {
+      "epoch": 0.32266748711637566,
+      "learning_rate": 0.002293098692669235,
+      "loss": 1.0602,
+      "step": 4195
+    },
+    {
+      "epoch": 0.3227444042765941,
+      "learning_rate": 0.0022927910145863215,
+      "loss": 0.9043,
+      "step": 4196
+    },
+    {
+      "epoch": 0.3228213214368126,
+      "learning_rate": 0.002292483290211514,
+      "loss": 1.242,
+      "step": 4197
+    },
+    {
+      "epoch": 0.322898238597031,
+      "learning_rate": 0.0022921755195627817,
+      "loss": 1.4223,
+      "step": 4198
+    },
+    {
+      "epoch": 0.32297515575724944,
+      "learning_rate": 0.0022918677026580953,
+      "loss": 1.2745,
+      "step": 4199
+    },
+    {
+      "epoch": 0.3230520729174679,
+      "learning_rate": 0.002291559839515428,
+      "loss": 1.0063,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3231289900776863,
+      "learning_rate": 0.0022912519301527576,
+      "loss": 1.0889,
+      "step": 4201
+    },
+    {
+      "epoch": 0.32320590723790477,
+      "learning_rate": 0.002290943974588061,
+      "loss": 0.9462,
+      "step": 4202
+    },
+    {
+      "epoch": 0.3232828243981232,
+      "learning_rate": 0.0022906359728393226,
+      "loss": 1.152,
+      "step": 4203
+    },
+    {
+      "epoch": 0.3233597415583417,
+      "learning_rate": 0.002290327924924525,
+      "loss": 1.0191,
+      "step": 4204
+    },
+    {
+      "epoch": 0.3234366587185601,
+      "learning_rate": 0.0022900198308616567,
+      "loss": 0.9533,
+      "step": 4205
+    },
+    {
+      "epoch": 0.32351357587877855,
+      "learning_rate": 0.002289711690668707,
+      "loss": 1.4973,
+      "step": 4206
+    },
+    {
+      "epoch": 0.323590493038997,
+      "learning_rate": 0.002289403504363669,
+      "loss": 1.248,
+      "step": 4207
+    },
+    {
+      "epoch": 0.32366741019921547,
+      "learning_rate": 0.002289095271964537,
+      "loss": 1.2784,
+      "step": 4208
+    },
+    {
+      "epoch": 0.32374432735943387,
+      "learning_rate": 0.00228878699348931,
+      "loss": 1.163,
+      "step": 4209
+    },
+    {
+      "epoch": 0.32382124451965233,
+      "learning_rate": 0.0022884786689559887,
+      "loss": 1.165,
+      "step": 4210
+    },
+    {
+      "epoch": 0.3238981616798708,
+      "learning_rate": 0.0022881702983825756,
+      "loss": 1.1404,
+      "step": 4211
+    },
+    {
+      "epoch": 0.32397507884008925,
+      "learning_rate": 0.002287861881787078,
+      "loss": 1.1549,
+      "step": 4212
+    },
+    {
+      "epoch": 0.32405199600030765,
+      "learning_rate": 0.002287553419187503,
+      "loss": 1.3102,
+      "step": 4213
+    },
+    {
+      "epoch": 0.3241289131605261,
+      "learning_rate": 0.0022872449106018642,
+      "loss": 1.3165,
+      "step": 4214
+    },
+    {
+      "epoch": 0.32420583032074457,
+      "learning_rate": 0.0022869363560481743,
+      "loss": 0.9043,
+      "step": 4215
+    },
+    {
+      "epoch": 0.32428274748096303,
+      "learning_rate": 0.00228662775554445,
+      "loss": 1.0413,
+      "step": 4216
+    },
+    {
+      "epoch": 0.32435966464118143,
+      "learning_rate": 0.0022863191091087123,
+      "loss": 1.1427,
+      "step": 4217
+    },
+    {
+      "epoch": 0.3244365818013999,
+      "learning_rate": 0.0022860104167589813,
+      "loss": 1.1331,
+      "step": 4218
+    },
+    {
+      "epoch": 0.32451349896161835,
+      "learning_rate": 0.002285701678513283,
+      "loss": 1.4227,
+      "step": 4219
+    },
+    {
+      "epoch": 0.32459041612183676,
+      "learning_rate": 0.0022853928943896455,
+      "loss": 1.3405,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3246673332820552,
+      "learning_rate": 0.0022850840644060982,
+      "loss": 1.2141,
+      "step": 4221
+    },
+    {
+      "epoch": 0.3247442504422737,
+      "learning_rate": 0.002284775188580674,
+      "loss": 0.99,
+      "step": 4222
+    },
+    {
+      "epoch": 0.32482116760249213,
+      "learning_rate": 0.0022844662669314086,
+      "loss": 1.253,
+      "step": 4223
+    },
+    {
+      "epoch": 0.32489808476271054,
+      "learning_rate": 0.00228415729947634,
+      "loss": 1.3627,
+      "step": 4224
+    },
+    {
+      "epoch": 0.324975001922929,
+      "learning_rate": 0.0022838482862335098,
+      "loss": 1.1932,
+      "step": 4225
+    },
+    {
+      "epoch": 0.32505191908314746,
+      "learning_rate": 0.002283539227220961,
+      "loss": 1.1084,
+      "step": 4226
+    },
+    {
+      "epoch": 0.3251288362433659,
+      "learning_rate": 0.0022832301224567407,
+      "loss": 0.7417,
+      "step": 4227
+    },
+    {
+      "epoch": 0.3252057534035843,
+      "learning_rate": 0.0022829209719588974,
+      "loss": 1.1925,
+      "step": 4228
+    },
+    {
+      "epoch": 0.3252826705638028,
+      "learning_rate": 0.0022826117757454817,
+      "loss": 1.0792,
+      "step": 4229
+    },
+    {
+      "epoch": 0.32535958772402124,
+      "learning_rate": 0.0022823025338345494,
+      "loss": 0.8884,
+      "step": 4230
+    },
+    {
+      "epoch": 0.3254365048842397,
+      "learning_rate": 0.0022819932462441565,
+      "loss": 1.1749,
+      "step": 4231
+    },
+    {
+      "epoch": 0.3255134220444581,
+      "learning_rate": 0.002281683912992364,
+      "loss": 0.7102,
+      "step": 4232
+    },
+    {
+      "epoch": 0.32559033920467656,
+      "learning_rate": 0.0022813745340972322,
+      "loss": 1.4558,
+      "step": 4233
+    },
+    {
+      "epoch": 0.325667256364895,
+      "learning_rate": 0.002281065109576827,
+      "loss": 0.8879,
+      "step": 4234
+    },
+    {
+      "epoch": 0.3257441735251135,
+      "learning_rate": 0.002280755639449216,
+      "loss": 0.9001,
+      "step": 4235
+    },
+    {
+      "epoch": 0.3258210906853319,
+      "learning_rate": 0.0022804461237324702,
+      "loss": 1.4333,
+      "step": 4236
+    },
+    {
+      "epoch": 0.32589800784555034,
+      "learning_rate": 0.0022801365624446615,
+      "loss": 1.1534,
+      "step": 4237
+    },
+    {
+      "epoch": 0.3259749250057688,
+      "learning_rate": 0.0022798269556038656,
+      "loss": 0.814,
+      "step": 4238
+    },
+    {
+      "epoch": 0.3260518421659872,
+      "learning_rate": 0.0022795173032281618,
+      "loss": 1.3223,
+      "step": 4239
+    },
+    {
+      "epoch": 0.32612875932620566,
+      "learning_rate": 0.0022792076053356294,
+      "loss": 0.7201,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3262056764864241,
+      "learning_rate": 0.002278897861944353,
+      "loss": 0.9512,
+      "step": 4241
+    },
+    {
+      "epoch": 0.3262825936466426,
+      "learning_rate": 0.002278588073072419,
+      "loss": 1.002,
+      "step": 4242
+    },
+    {
+      "epoch": 0.326359510806861,
+      "learning_rate": 0.0022782782387379165,
+      "loss": 1.1497,
+      "step": 4243
+    },
+    {
+      "epoch": 0.32643642796707945,
+      "learning_rate": 0.002277968358958936,
+      "loss": 0.7548,
+      "step": 4244
+    },
+    {
+      "epoch": 0.3265133451272979,
+      "learning_rate": 0.0022776584337535714,
+      "loss": 0.9711,
+      "step": 4245
+    },
+    {
+      "epoch": 0.32659026228751636,
+      "learning_rate": 0.0022773484631399216,
+      "loss": 1.067,
+      "step": 4246
+    },
+    {
+      "epoch": 0.32666717944773477,
+      "learning_rate": 0.0022770384471360846,
+      "loss": 0.9141,
+      "step": 4247
+    },
+    {
+      "epoch": 0.3267440966079532,
+      "learning_rate": 0.0022767283857601625,
+      "loss": 1.3968,
+      "step": 4248
+    },
+    {
+      "epoch": 0.3268210137681717,
+      "learning_rate": 0.002276418279030261,
+      "loss": 0.8375,
+      "step": 4249
+    },
+    {
+      "epoch": 0.32689793092839015,
+      "learning_rate": 0.0022761081269644864,
+      "loss": 1.1958,
+      "step": 4250
+    },
+    {
+      "epoch": 0.32697484808860855,
+      "learning_rate": 0.0022757979295809495,
+      "loss": 1.0153,
+      "step": 4251
+    },
+    {
+      "epoch": 0.327051765248827,
+      "learning_rate": 0.0022754876868977636,
+      "loss": 1.0296,
+      "step": 4252
+    },
+    {
+      "epoch": 0.32712868240904547,
+      "learning_rate": 0.002275177398933043,
+      "loss": 1.0223,
+      "step": 4253
+    },
+    {
+      "epoch": 0.3272055995692639,
+      "learning_rate": 0.0022748670657049062,
+      "loss": 1.2272,
+      "step": 4254
+    },
+    {
+      "epoch": 0.32728251672948233,
+      "learning_rate": 0.002274556687231474,
+      "loss": 1.0495,
+      "step": 4255
+    },
+    {
+      "epoch": 0.3273594338897008,
+      "learning_rate": 0.00227424626353087,
+      "loss": 1.1607,
+      "step": 4256
+    },
+    {
+      "epoch": 0.32743635104991925,
+      "learning_rate": 0.002273935794621219,
+      "loss": 1.1606,
+      "step": 4257
+    },
+    {
+      "epoch": 0.3275132682101377,
+      "learning_rate": 0.002273625280520651,
+      "loss": 1.4155,
+      "step": 4258
+    },
+    {
+      "epoch": 0.3275901853703561,
+      "learning_rate": 0.0022733147212472965,
+      "loss": 1.1348,
+      "step": 4259
+    },
+    {
+      "epoch": 0.32766710253057457,
+      "learning_rate": 0.0022730041168192894,
+      "loss": 1.1552,
+      "step": 4260
+    },
+    {
+      "epoch": 0.32774401969079303,
+      "learning_rate": 0.0022726934672547662,
+      "loss": 1.3827,
+      "step": 4261
+    },
+    {
+      "epoch": 0.32782093685101144,
+      "learning_rate": 0.002272382772571866,
+      "loss": 1.7051,
+      "step": 4262
+    },
+    {
+      "epoch": 0.3278978540112299,
+      "learning_rate": 0.0022720720327887313,
+      "loss": 1.2682,
+      "step": 4263
+    },
+    {
+      "epoch": 0.32797477117144835,
+      "learning_rate": 0.0022717612479235053,
+      "loss": 1.2746,
+      "step": 4264
+    },
+    {
+      "epoch": 0.3280516883316668,
+      "learning_rate": 0.0022714504179943366,
+      "loss": 1.46,
+      "step": 4265
+    },
+    {
+      "epoch": 0.3281286054918852,
+      "learning_rate": 0.002271139543019373,
+      "loss": 1.0588,
+      "step": 4266
+    },
+    {
+      "epoch": 0.3282055226521037,
+      "learning_rate": 0.0022708286230167687,
+      "loss": 1.1343,
+      "step": 4267
+    },
+    {
+      "epoch": 0.32828243981232214,
+      "learning_rate": 0.0022705176580046773,
+      "loss": 1.0675,
+      "step": 4268
+    },
+    {
+      "epoch": 0.3283593569725406,
+      "learning_rate": 0.002270206648001257,
+      "loss": 1.0478,
+      "step": 4269
+    },
+    {
+      "epoch": 0.328436274132759,
+      "learning_rate": 0.002269895593024668,
+      "loss": 1.1152,
+      "step": 4270
+    },
+    {
+      "epoch": 0.32851319129297746,
+      "learning_rate": 0.0022695844930930727,
+      "loss": 1.1935,
+      "step": 4271
+    },
+    {
+      "epoch": 0.3285901084531959,
+      "learning_rate": 0.0022692733482246367,
+      "loss": 1.316,
+      "step": 4272
+    },
+    {
+      "epoch": 0.3286670256134144,
+      "learning_rate": 0.0022689621584375285,
+      "loss": 1.1137,
+      "step": 4273
+    },
+    {
+      "epoch": 0.3287439427736328,
+      "learning_rate": 0.0022686509237499182,
+      "loss": 0.7445,
+      "step": 4274
+    },
+    {
+      "epoch": 0.32882085993385124,
+      "learning_rate": 0.0022683396441799787,
+      "loss": 1.191,
+      "step": 4275
+    },
+    {
+      "epoch": 0.3288977770940697,
+      "learning_rate": 0.002268028319745888,
+      "loss": 1.276,
+      "step": 4276
+    },
+    {
+      "epoch": 0.32897469425428816,
+      "learning_rate": 0.0022677169504658223,
+      "loss": 0.9966,
+      "step": 4277
+    },
+    {
+      "epoch": 0.32905161141450656,
+      "learning_rate": 0.0022674055363579647,
+      "loss": 1.2092,
+      "step": 4278
+    },
+    {
+      "epoch": 0.329128528574725,
+      "learning_rate": 0.0022670940774404975,
+      "loss": 1.024,
+      "step": 4279
+    },
+    {
+      "epoch": 0.3292054457349435,
+      "learning_rate": 0.002266782573731607,
+      "loss": 0.9996,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3292823628951619,
+      "learning_rate": 0.002266471025249484,
+      "loss": 1.3885,
+      "step": 4281
+    },
+    {
+      "epoch": 0.32935928005538034,
+      "learning_rate": 0.0022661594320123185,
+      "loss": 0.7952,
+      "step": 4282
+    },
+    {
+      "epoch": 0.3294361972155988,
+      "learning_rate": 0.0022658477940383057,
+      "loss": 1.1963,
+      "step": 4283
+    },
+    {
+      "epoch": 0.32951311437581726,
+      "learning_rate": 0.0022655361113456418,
+      "loss": 1.2864,
+      "step": 4284
+    },
+    {
+      "epoch": 0.32959003153603567,
+      "learning_rate": 0.0022652243839525266,
+      "loss": 1.033,
+      "step": 4285
+    },
+    {
+      "epoch": 0.3296669486962541,
+      "learning_rate": 0.002264912611877162,
+      "loss": 1.2902,
+      "step": 4286
+    },
+    {
+      "epoch": 0.3297438658564726,
+      "learning_rate": 0.002264600795137753,
+      "loss": 1.088,
+      "step": 4287
+    },
+    {
+      "epoch": 0.32982078301669104,
+      "learning_rate": 0.002264288933752506,
+      "loss": 1.2543,
+      "step": 4288
+    },
+    {
+      "epoch": 0.32989770017690945,
+      "learning_rate": 0.0022639770277396326,
+      "loss": 1.0899,
+      "step": 4289
+    },
+    {
+      "epoch": 0.3299746173371279,
+      "learning_rate": 0.002263665077117344,
+      "loss": 1.0596,
+      "step": 4290
+    },
+    {
+      "epoch": 0.33005153449734637,
+      "learning_rate": 0.0022633530819038557,
+      "loss": 1.5231,
+      "step": 4291
+    },
+    {
+      "epoch": 0.3301284516575648,
+      "learning_rate": 0.0022630410421173855,
+      "loss": 0.9046,
+      "step": 4292
+    },
+    {
+      "epoch": 0.33020536881778323,
+      "learning_rate": 0.002262728957776153,
+      "loss": 1.118,
+      "step": 4293
+    },
+    {
+      "epoch": 0.3302822859780017,
+      "learning_rate": 0.0022624168288983824,
+      "loss": 1.0512,
+      "step": 4294
+    },
+    {
+      "epoch": 0.33035920313822015,
+      "learning_rate": 0.0022621046555022983,
+      "loss": 0.9748,
+      "step": 4295
+    },
+    {
+      "epoch": 0.3304361202984386,
+      "learning_rate": 0.002261792437606129,
+      "loss": 1.0904,
+      "step": 4296
+    },
+    {
+      "epoch": 0.330513037458657,
+      "learning_rate": 0.002261480175228106,
+      "loss": 1.3217,
+      "step": 4297
+    },
+    {
+      "epoch": 0.33058995461887547,
+      "learning_rate": 0.0022611678683864614,
+      "loss": 1.1509,
+      "step": 4298
+    },
+    {
+      "epoch": 0.33066687177909393,
+      "learning_rate": 0.002260855517099432,
+      "loss": 1.2228,
+      "step": 4299
+    },
+    {
+      "epoch": 0.33074378893931233,
+      "learning_rate": 0.0022605431213852563,
+      "loss": 1.2046,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3308207060995308,
+      "learning_rate": 0.0022602306812621745,
+      "loss": 1.298,
+      "step": 4301
+    },
+    {
+      "epoch": 0.33089762325974925,
+      "learning_rate": 0.002259918196748431,
+      "loss": 1.1577,
+      "step": 4302
+    },
+    {
+      "epoch": 0.3309745404199677,
+      "learning_rate": 0.002259605667862272,
+      "loss": 1.0975,
+      "step": 4303
+    },
+    {
+      "epoch": 0.3310514575801861,
+      "learning_rate": 0.0022592930946219465,
+      "loss": 1.2118,
+      "step": 4304
+    },
+    {
+      "epoch": 0.3311283747404046,
+      "learning_rate": 0.0022589804770457057,
+      "loss": 0.7915,
+      "step": 4305
+    },
+    {
+      "epoch": 0.33120529190062303,
+      "learning_rate": 0.0022586678151518044,
+      "loss": 1.1003,
+      "step": 4306
+    },
+    {
+      "epoch": 0.3312822090608415,
+      "learning_rate": 0.0022583551089584986,
+      "loss": 1.1668,
+      "step": 4307
+    },
+    {
+      "epoch": 0.3313591262210599,
+      "learning_rate": 0.0022580423584840476,
+      "loss": 0.9239,
+      "step": 4308
+    },
+    {
+      "epoch": 0.33143604338127836,
+      "learning_rate": 0.002257729563746713,
+      "loss": 1.1236,
+      "step": 4309
+    },
+    {
+      "epoch": 0.3315129605414968,
+      "learning_rate": 0.0022574167247647594,
+      "loss": 1.1646,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3315898777017153,
+      "learning_rate": 0.002257103841556454,
+      "loss": 1.3413,
+      "step": 4311
+    },
+    {
+      "epoch": 0.3316667948619337,
+      "learning_rate": 0.0022567909141400664,
+      "loss": 1.2623,
+      "step": 4312
+    },
+    {
+      "epoch": 0.33174371202215214,
+      "learning_rate": 0.002256477942533869,
+      "loss": 1.1116,
+      "step": 4313
+    },
+    {
+      "epoch": 0.3318206291823706,
+      "learning_rate": 0.0022561649267561355,
+      "loss": 1.0319,
+      "step": 4314
+    },
+    {
+      "epoch": 0.33189754634258906,
+      "learning_rate": 0.0022558518668251443,
+      "loss": 0.8188,
+      "step": 4315
+    },
+    {
+      "epoch": 0.33197446350280746,
+      "learning_rate": 0.0022555387627591745,
+      "loss": 1.2136,
+      "step": 4316
+    },
+    {
+      "epoch": 0.3320513806630259,
+      "learning_rate": 0.0022552256145765086,
+      "loss": 1.2836,
+      "step": 4317
+    },
+    {
+      "epoch": 0.3321282978232444,
+      "learning_rate": 0.002254912422295433,
+      "loss": 1.3871,
+      "step": 4318
+    },
+    {
+      "epoch": 0.33220521498346284,
+      "learning_rate": 0.002254599185934233,
+      "loss": 1.4418,
+      "step": 4319
+    },
+    {
+      "epoch": 0.33228213214368124,
+      "learning_rate": 0.0022542859055112016,
+      "loss": 1.2481,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3323590493038997,
+      "learning_rate": 0.0022539725810446294,
+      "loss": 1.0429,
+      "step": 4321
+    },
+    {
+      "epoch": 0.33243596646411816,
+      "learning_rate": 0.0022536592125528125,
+      "loss": 1.0748,
+      "step": 4322
+    },
+    {
+      "epoch": 0.33251288362433656,
+      "learning_rate": 0.0022533458000540485,
+      "loss": 1.3121,
+      "step": 4323
+    },
+    {
+      "epoch": 0.332589800784555,
+      "learning_rate": 0.0022530323435666385,
+      "loss": 1.3557,
+      "step": 4324
+    },
+    {
+      "epoch": 0.3326667179447735,
+      "learning_rate": 0.0022527188431088847,
+      "loss": 1.1272,
+      "step": 4325
+    },
+    {
+      "epoch": 0.33274363510499194,
+      "learning_rate": 0.002252405298699093,
+      "loss": 1.029,
+      "step": 4326
+    },
+    {
+      "epoch": 0.33282055226521035,
+      "learning_rate": 0.0022520917103555724,
+      "loss": 1.1077,
+      "step": 4327
+    },
+    {
+      "epoch": 0.3328974694254288,
+      "learning_rate": 0.0022517780780966327,
+      "loss": 1.3373,
+      "step": 4328
+    },
+    {
+      "epoch": 0.33297438658564726,
+      "learning_rate": 0.0022514644019405875,
+      "loss": 1.39,
+      "step": 4329
+    },
+    {
+      "epoch": 0.3330513037458657,
+      "learning_rate": 0.0022511506819057527,
+      "loss": 1.1247,
+      "step": 4330
+    },
+    {
+      "epoch": 0.3331282209060841,
+      "learning_rate": 0.0022508369180104476,
+      "loss": 1.3734,
+      "step": 4331
+    },
+    {
+      "epoch": 0.3332051380663026,
+      "learning_rate": 0.002250523110272991,
+      "loss": 1.5231,
+      "step": 4332
+    },
+    {
+      "epoch": 0.33328205522652105,
+      "learning_rate": 0.002250209258711709,
+      "loss": 1.0011,
+      "step": 4333
+    },
+    {
+      "epoch": 0.3333589723867395,
+      "learning_rate": 0.0022498953633449258,
+      "loss": 1.0763,
+      "step": 4334
+    },
+    {
+      "epoch": 0.3334358895469579,
+      "learning_rate": 0.0022495814241909715,
+      "loss": 1.0338,
+      "step": 4335
+    },
+    {
+      "epoch": 0.33351280670717637,
+      "learning_rate": 0.0022492674412681758,
+      "loss": 1.3041,
+      "step": 4336
+    },
+    {
+      "epoch": 0.3335897238673948,
+      "learning_rate": 0.002248953414594874,
+      "loss": 1.0705,
+      "step": 4337
+    },
+    {
+      "epoch": 0.3336666410276133,
+      "learning_rate": 0.002248639344189401,
+      "loss": 1.1476,
+      "step": 4338
+    },
+    {
+      "epoch": 0.3337435581878317,
+      "learning_rate": 0.002248325230070097,
+      "loss": 0.6311,
+      "step": 4339
+    },
+    {
+      "epoch": 0.33382047534805015,
+      "learning_rate": 0.002248011072255303,
+      "loss": 1.1914,
+      "step": 4340
+    },
+    {
+      "epoch": 0.3338973925082686,
+      "learning_rate": 0.002247696870763363,
+      "loss": 1.354,
+      "step": 4341
+    },
+    {
+      "epoch": 0.333974309668487,
+      "learning_rate": 0.0022473826256126234,
+      "loss": 1.2951,
+      "step": 4342
+    },
+    {
+      "epoch": 0.33405122682870547,
+      "learning_rate": 0.0022470683368214332,
+      "loss": 0.8793,
+      "step": 4343
+    },
+    {
+      "epoch": 0.33412814398892393,
+      "learning_rate": 0.0022467540044081445,
+      "loss": 0.9412,
+      "step": 4344
+    },
+    {
+      "epoch": 0.3342050611491424,
+      "learning_rate": 0.0022464396283911107,
+      "loss": 1.0309,
+      "step": 4345
+    },
+    {
+      "epoch": 0.3342819783093608,
+      "learning_rate": 0.0022461252087886898,
+      "loss": 1.2234,
+      "step": 4346
+    },
+    {
+      "epoch": 0.33435889546957925,
+      "learning_rate": 0.00224581074561924,
+      "loss": 1.4772,
+      "step": 4347
+    },
+    {
+      "epoch": 0.3344358126297977,
+      "learning_rate": 0.0022454962389011237,
+      "loss": 1.2051,
+      "step": 4348
+    },
+    {
+      "epoch": 0.3345127297900162,
+      "learning_rate": 0.0022451816886527045,
+      "loss": 1.0304,
+      "step": 4349
+    },
+    {
+      "epoch": 0.3345896469502346,
+      "learning_rate": 0.0022448670948923503,
+      "loss": 1.6065,
+      "step": 4350
+    },
+    {
+      "epoch": 0.33466656411045304,
+      "learning_rate": 0.00224455245763843,
+      "loss": 0.9958,
+      "step": 4351
+    },
+    {
+      "epoch": 0.3347434812706715,
+      "learning_rate": 0.0022442377769093156,
+      "loss": 1.2912,
+      "step": 4352
+    },
+    {
+      "epoch": 0.33482039843088995,
+      "learning_rate": 0.002243923052723382,
+      "loss": 1.1904,
+      "step": 4353
+    },
+    {
+      "epoch": 0.33489731559110836,
+      "learning_rate": 0.0022436082850990067,
+      "loss": 1.496,
+      "step": 4354
+    },
+    {
+      "epoch": 0.3349742327513268,
+      "learning_rate": 0.002243293474054568,
+      "loss": 1.3317,
+      "step": 4355
+    },
+    {
+      "epoch": 0.3350511499115453,
+      "learning_rate": 0.002242978619608449,
+      "loss": 1.3624,
+      "step": 4356
+    },
+    {
+      "epoch": 0.33512806707176374,
+      "learning_rate": 0.0022426637217790335,
+      "loss": 1.3946,
+      "step": 4357
+    },
+    {
+      "epoch": 0.33520498423198214,
+      "learning_rate": 0.0022423487805847096,
+      "loss": 0.9476,
+      "step": 4358
+    },
+    {
+      "epoch": 0.3352819013922006,
+      "learning_rate": 0.0022420337960438667,
+      "loss": 1.0602,
+      "step": 4359
+    },
+    {
+      "epoch": 0.33535881855241906,
+      "learning_rate": 0.002241718768174897,
+      "loss": 1.3431,
+      "step": 4360
+    },
+    {
+      "epoch": 0.33543573571263746,
+      "learning_rate": 0.0022414036969961963,
+      "loss": 1.1943,
+      "step": 4361
+    },
+    {
+      "epoch": 0.3355126528728559,
+      "learning_rate": 0.00224108858252616,
+      "loss": 0.9365,
+      "step": 4362
+    },
+    {
+      "epoch": 0.3355895700330744,
+      "learning_rate": 0.0022407734247831897,
+      "loss": 1.1168,
+      "step": 4363
+    },
+    {
+      "epoch": 0.33566648719329284,
+      "learning_rate": 0.0022404582237856865,
+      "loss": 1.6628,
+      "step": 4364
+    },
+    {
+      "epoch": 0.33574340435351124,
+      "learning_rate": 0.0022401429795520563,
+      "loss": 1.1019,
+      "step": 4365
+    },
+    {
+      "epoch": 0.3358203215137297,
+      "learning_rate": 0.0022398276921007064,
+      "loss": 0.9463,
+      "step": 4366
+    },
+    {
+      "epoch": 0.33589723867394816,
+      "learning_rate": 0.002239512361450046,
+      "loss": 1.6078,
+      "step": 4367
+    },
+    {
+      "epoch": 0.3359741558341666,
+      "learning_rate": 0.002239196987618489,
+      "loss": 1.3525,
+      "step": 4368
+    },
+    {
+      "epoch": 0.336051072994385,
+      "learning_rate": 0.0022388815706244486,
+      "loss": 1.2302,
+      "step": 4369
+    },
+    {
+      "epoch": 0.3361279901546035,
+      "learning_rate": 0.002238566110486343,
+      "loss": 1.0367,
+      "step": 4370
+    },
+    {
+      "epoch": 0.33620490731482194,
+      "learning_rate": 0.0022382506072225936,
+      "loss": 1.1865,
+      "step": 4371
+    },
+    {
+      "epoch": 0.3362818244750404,
+      "learning_rate": 0.002237935060851621,
+      "loss": 1.2263,
+      "step": 4372
+    },
+    {
+      "epoch": 0.3363587416352588,
+      "learning_rate": 0.0022376194713918515,
+      "loss": 1.2354,
+      "step": 4373
+    },
+    {
+      "epoch": 0.33643565879547727,
+      "learning_rate": 0.0022373038388617122,
+      "loss": 0.9629,
+      "step": 4374
+    },
+    {
+      "epoch": 0.3365125759556957,
+      "learning_rate": 0.002236988163279634,
+      "loss": 1.2289,
+      "step": 4375
+    },
+    {
+      "epoch": 0.3365894931159142,
+      "learning_rate": 0.0022366724446640476,
+      "loss": 1.3058,
+      "step": 4376
+    },
+    {
+      "epoch": 0.3366664102761326,
+      "learning_rate": 0.0022363566830333904,
+      "loss": 1.0201,
+      "step": 4377
+    },
+    {
+      "epoch": 0.33674332743635105,
+      "learning_rate": 0.0022360408784060986,
+      "loss": 0.8795,
+      "step": 4378
+    },
+    {
+      "epoch": 0.3368202445965695,
+      "learning_rate": 0.002235725030800613,
+      "loss": 1.0216,
+      "step": 4379
+    },
+    {
+      "epoch": 0.33689716175678797,
+      "learning_rate": 0.002235409140235376,
+      "loss": 0.9301,
+      "step": 4380
+    },
+    {
+      "epoch": 0.33697407891700637,
+      "learning_rate": 0.002235093206728833,
+      "loss": 1.1659,
+      "step": 4381
+    },
+    {
+      "epoch": 0.33705099607722483,
+      "learning_rate": 0.0022347772302994317,
+      "loss": 1.1632,
+      "step": 4382
+    },
+    {
+      "epoch": 0.3371279132374433,
+      "learning_rate": 0.002234461210965622,
+      "loss": 1.0076,
+      "step": 4383
+    },
+    {
+      "epoch": 0.3372048303976617,
+      "learning_rate": 0.002234145148745857,
+      "loss": 1.1557,
+      "step": 4384
+    },
+    {
+      "epoch": 0.33728174755788015,
+      "learning_rate": 0.002233829043658591,
+      "loss": 1.2187,
+      "step": 4385
+    },
+    {
+      "epoch": 0.3373586647180986,
+      "learning_rate": 0.0022335128957222832,
+      "loss": 1.0021,
+      "step": 4386
+    },
+    {
+      "epoch": 0.33743558187831707,
+      "learning_rate": 0.002233196704955392,
+      "loss": 0.8605,
+      "step": 4387
+    },
+    {
+      "epoch": 0.3375124990385355,
+      "learning_rate": 0.002232880471376382,
+      "loss": 0.8381,
+      "step": 4388
+    },
+    {
+      "epoch": 0.33758941619875393,
+      "learning_rate": 0.002232564195003717,
+      "loss": 1.1981,
+      "step": 4389
+    },
+    {
+      "epoch": 0.3376663333589724,
+      "learning_rate": 0.0022322478758558654,
+      "loss": 0.9065,
+      "step": 4390
+    },
+    {
+      "epoch": 0.33774325051919085,
+      "learning_rate": 0.002231931513951297,
+      "loss": 0.9244,
+      "step": 4391
+    },
+    {
+      "epoch": 0.33782016767940926,
+      "learning_rate": 0.0022316151093084847,
+      "loss": 1.1886,
+      "step": 4392
+    },
+    {
+      "epoch": 0.3378970848396277,
+      "learning_rate": 0.002231298661945904,
+      "loss": 1.2158,
+      "step": 4393
+    },
+    {
+      "epoch": 0.3379740019998462,
+      "learning_rate": 0.002230982171882032,
+      "loss": 1.2558,
+      "step": 4394
+    },
+    {
+      "epoch": 0.33805091916006463,
+      "learning_rate": 0.0022306656391353494,
+      "loss": 1.4155,
+      "step": 4395
+    },
+    {
+      "epoch": 0.33812783632028304,
+      "learning_rate": 0.0022303490637243383,
+      "loss": 1.0438,
+      "step": 4396
+    },
+    {
+      "epoch": 0.3382047534805015,
+      "learning_rate": 0.0022300324456674845,
+      "loss": 1.3169,
+      "step": 4397
+    },
+    {
+      "epoch": 0.33828167064071996,
+      "learning_rate": 0.002229715784983275,
+      "loss": 0.7401,
+      "step": 4398
+    },
+    {
+      "epoch": 0.3383585878009384,
+      "learning_rate": 0.0022293990816902006,
+      "loss": 1.0843,
+      "step": 4399
+    },
+    {
+      "epoch": 0.3384355049611568,
+      "learning_rate": 0.0022290823358067542,
+      "loss": 1.296,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3385124221213753,
+      "learning_rate": 0.0022287655473514295,
+      "loss": 1.1755,
+      "step": 4401
+    },
+    {
+      "epoch": 0.33858933928159374,
+      "learning_rate": 0.0022284487163427256,
+      "loss": 1.2204,
+      "step": 4402
+    },
+    {
+      "epoch": 0.33866625644181214,
+      "learning_rate": 0.0022281318427991416,
+      "loss": 1.1812,
+      "step": 4403
+    },
+    {
+      "epoch": 0.3387431736020306,
+      "learning_rate": 0.0022278149267391805,
+      "loss": 1.5273,
+      "step": 4404
+    },
+    {
+      "epoch": 0.33882009076224906,
+      "learning_rate": 0.0022274979681813473,
+      "loss": 1.298,
+      "step": 4405
+    },
+    {
+      "epoch": 0.3388970079224675,
+      "learning_rate": 0.00222718096714415,
+      "loss": 1.2752,
+      "step": 4406
+    },
+    {
+      "epoch": 0.3389739250826859,
+      "learning_rate": 0.0022268639236460975,
+      "loss": 1.1704,
+      "step": 4407
+    },
+    {
+      "epoch": 0.3390508422429044,
+      "learning_rate": 0.0022265468377057037,
+      "loss": 0.9175,
+      "step": 4408
+    },
+    {
+      "epoch": 0.33912775940312284,
+      "learning_rate": 0.0022262297093414824,
+      "loss": 1.1228,
+      "step": 4409
+    },
+    {
+      "epoch": 0.3392046765633413,
+      "learning_rate": 0.0022259125385719514,
+      "loss": 1.1993,
+      "step": 4410
+    },
+    {
+      "epoch": 0.3392815937235597,
+      "learning_rate": 0.002225595325415631,
+      "loss": 0.747,
+      "step": 4411
+    },
+    {
+      "epoch": 0.33935851088377816,
+      "learning_rate": 0.0022252780698910433,
+      "loss": 1.2871,
+      "step": 4412
+    },
+    {
+      "epoch": 0.3394354280439966,
+      "learning_rate": 0.002224960772016713,
+      "loss": 1.3381,
+      "step": 4413
+    },
+    {
+      "epoch": 0.3395123452042151,
+      "learning_rate": 0.0022246434318111685,
+      "loss": 1.0126,
+      "step": 4414
+    },
+    {
+      "epoch": 0.3395892623644335,
+      "learning_rate": 0.0022243260492929378,
+      "loss": 1.4034,
+      "step": 4415
+    },
+    {
+      "epoch": 0.33966617952465195,
+      "learning_rate": 0.0022240086244805545,
+      "loss": 1.2425,
+      "step": 4416
+    },
+    {
+      "epoch": 0.3397430966848704,
+      "learning_rate": 0.0022236911573925533,
+      "loss": 1.7546,
+      "step": 4417
+    },
+    {
+      "epoch": 0.33982001384508886,
+      "learning_rate": 0.002223373648047471,
+      "loss": 1.0383,
+      "step": 4418
+    },
+    {
+      "epoch": 0.33989693100530727,
+      "learning_rate": 0.002223056096463848,
+      "loss": 1.3741,
+      "step": 4419
+    },
+    {
+      "epoch": 0.3399738481655257,
+      "learning_rate": 0.0022227385026602252,
+      "loss": 0.7326,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3400507653257442,
+      "learning_rate": 0.0022224208666551484,
+      "loss": 1.4531,
+      "step": 4421
+    },
+    {
+      "epoch": 0.34012768248596265,
+      "learning_rate": 0.0022221031884671646,
+      "loss": 1.0761,
+      "step": 4422
+    },
+    {
+      "epoch": 0.34020459964618105,
+      "learning_rate": 0.002221785468114823,
+      "loss": 1.086,
+      "step": 4423
+    },
+    {
+      "epoch": 0.3402815168063995,
+      "learning_rate": 0.002221467705616675,
+      "loss": 1.0769,
+      "step": 4424
+    },
+    {
+      "epoch": 0.34035843396661797,
+      "learning_rate": 0.0022211499009912763,
+      "loss": 1.2885,
+      "step": 4425
+    },
+    {
+      "epoch": 0.34043535112683637,
+      "learning_rate": 0.0022208320542571837,
+      "loss": 1.1621,
+      "step": 4426
+    },
+    {
+      "epoch": 0.34051226828705483,
+      "learning_rate": 0.0022205141654329554,
+      "loss": 1.306,
+      "step": 4427
+    },
+    {
+      "epoch": 0.3405891854472733,
+      "learning_rate": 0.0022201962345371543,
+      "loss": 1.0948,
+      "step": 4428
+    },
+    {
+      "epoch": 0.34066610260749175,
+      "learning_rate": 0.0022198782615883445,
+      "loss": 1.179,
+      "step": 4429
+    },
+    {
+      "epoch": 0.34074301976771015,
+      "learning_rate": 0.0022195602466050933,
+      "loss": 1.0467,
+      "step": 4430
+    },
+    {
+      "epoch": 0.3408199369279286,
+      "learning_rate": 0.002219242189605969,
+      "loss": 1.0063,
+      "step": 4431
+    },
+    {
+      "epoch": 0.34089685408814707,
+      "learning_rate": 0.002218924090609544,
+      "loss": 0.9891,
+      "step": 4432
+    },
+    {
+      "epoch": 0.34097377124836553,
+      "learning_rate": 0.0022186059496343913,
+      "loss": 1.3963,
+      "step": 4433
+    },
+    {
+      "epoch": 0.34105068840858394,
+      "learning_rate": 0.0022182877666990885,
+      "loss": 1.4076,
+      "step": 4434
+    },
+    {
+      "epoch": 0.3411276055688024,
+      "learning_rate": 0.002217969541822215,
+      "loss": 1.2832,
+      "step": 4435
+    },
+    {
+      "epoch": 0.34120452272902085,
+      "learning_rate": 0.002217651275022352,
+      "loss": 1.2185,
+      "step": 4436
+    },
+    {
+      "epoch": 0.3412814398892393,
+      "learning_rate": 0.002217332966318082,
+      "loss": 1.4446,
+      "step": 4437
+    },
+    {
+      "epoch": 0.3413583570494577,
+      "learning_rate": 0.0022170146157279927,
+      "loss": 1.1896,
+      "step": 4438
+    },
+    {
+      "epoch": 0.3414352742096762,
+      "learning_rate": 0.0022166962232706734,
+      "loss": 1.1249,
+      "step": 4439
+    },
+    {
+      "epoch": 0.34151219136989464,
+      "learning_rate": 0.002216377788964714,
+      "loss": 1.2731,
+      "step": 4440
+    },
+    {
+      "epoch": 0.3415891085301131,
+      "learning_rate": 0.00221605931282871,
+      "loss": 1.2442,
+      "step": 4441
+    },
+    {
+      "epoch": 0.3416660256903315,
+      "learning_rate": 0.0022157407948812558,
+      "loss": 1.1938,
+      "step": 4442
+    },
+    {
+      "epoch": 0.34174294285054996,
+      "learning_rate": 0.0022154222351409504,
+      "loss": 0.8131,
+      "step": 4443
+    },
+    {
+      "epoch": 0.3418198600107684,
+      "learning_rate": 0.002215103633626395,
+      "loss": 1.3119,
+      "step": 4444
+    },
+    {
+      "epoch": 0.3418967771709868,
+      "learning_rate": 0.0022147849903561933,
+      "loss": 1.3002,
+      "step": 4445
+    },
+    {
+      "epoch": 0.3419736943312053,
+      "learning_rate": 0.002214466305348952,
+      "loss": 1.6118,
+      "step": 4446
+    },
+    {
+      "epoch": 0.34205061149142374,
+      "learning_rate": 0.0022141475786232774,
+      "loss": 1.2568,
+      "step": 4447
+    },
+    {
+      "epoch": 0.3421275286516422,
+      "learning_rate": 0.0022138288101977814,
+      "loss": 1.3951,
+      "step": 4448
+    },
+    {
+      "epoch": 0.3422044458118606,
+      "learning_rate": 0.002213510000091078,
+      "loss": 1.1839,
+      "step": 4449
+    },
+    {
+      "epoch": 0.34228136297207906,
+      "learning_rate": 0.0022131911483217817,
+      "loss": 0.766,
+      "step": 4450
+    },
+    {
+      "epoch": 0.3423582801322975,
+      "learning_rate": 0.002212872254908511,
+      "loss": 1.0955,
+      "step": 4451
+    },
+    {
+      "epoch": 0.342435197292516,
+      "learning_rate": 0.002212553319869886,
+      "loss": 1.0926,
+      "step": 4452
+    },
+    {
+      "epoch": 0.3425121144527344,
+      "learning_rate": 0.002212234343224531,
+      "loss": 1.3951,
+      "step": 4453
+    },
+    {
+      "epoch": 0.34258903161295284,
+      "learning_rate": 0.0022119153249910697,
+      "loss": 1.2506,
+      "step": 4454
+    },
+    {
+      "epoch": 0.3426659487731713,
+      "learning_rate": 0.0022115962651881307,
+      "loss": 1.0962,
+      "step": 4455
+    },
+    {
+      "epoch": 0.34274286593338976,
+      "learning_rate": 0.0022112771638343443,
+      "loss": 0.963,
+      "step": 4456
+    },
+    {
+      "epoch": 0.34281978309360817,
+      "learning_rate": 0.002210958020948343,
+      "loss": 1.0749,
+      "step": 4457
+    },
+    {
+      "epoch": 0.3428967002538266,
+      "learning_rate": 0.002210638836548762,
+      "loss": 0.9101,
+      "step": 4458
+    },
+    {
+      "epoch": 0.3429736174140451,
+      "learning_rate": 0.0022103196106542386,
+      "loss": 0.7482,
+      "step": 4459
+    },
+    {
+      "epoch": 0.34305053457426354,
+      "learning_rate": 0.0022100003432834127,
+      "loss": 1.1059,
+      "step": 4460
+    },
+    {
+      "epoch": 0.34312745173448195,
+      "learning_rate": 0.0022096810344549274,
+      "loss": 1.025,
+      "step": 4461
+    },
+    {
+      "epoch": 0.3432043688947004,
+      "learning_rate": 0.0022093616841874263,
+      "loss": 1.4481,
+      "step": 4462
+    },
+    {
+      "epoch": 0.34328128605491887,
+      "learning_rate": 0.0022090422924995576,
+      "loss": 1.3447,
+      "step": 4463
+    },
+    {
+      "epoch": 0.34335820321513727,
+      "learning_rate": 0.00220872285940997,
+      "loss": 0.6945,
+      "step": 4464
+    },
+    {
+      "epoch": 0.34343512037535573,
+      "learning_rate": 0.0022084033849373167,
+      "loss": 1.2389,
+      "step": 4465
+    },
+    {
+      "epoch": 0.3435120375355742,
+      "learning_rate": 0.0022080838691002505,
+      "loss": 0.7735,
+      "step": 4466
+    },
+    {
+      "epoch": 0.34358895469579265,
+      "learning_rate": 0.0022077643119174303,
+      "loss": 1.4623,
+      "step": 4467
+    },
+    {
+      "epoch": 0.34366587185601105,
+      "learning_rate": 0.002207444713407514,
+      "loss": 1.0244,
+      "step": 4468
+    },
+    {
+      "epoch": 0.3437427890162295,
+      "learning_rate": 0.002207125073589163,
+      "loss": 0.9686,
+      "step": 4469
+    },
+    {
+      "epoch": 0.34381970617644797,
+      "learning_rate": 0.0022068053924810425,
+      "loss": 1.1557,
+      "step": 4470
+    },
+    {
+      "epoch": 0.34389662333666643,
+      "learning_rate": 0.0022064856701018184,
+      "loss": 1.211,
+      "step": 4471
+    },
+    {
+      "epoch": 0.34397354049688483,
+      "learning_rate": 0.00220616590647016,
+      "loss": 1.4653,
+      "step": 4472
+    },
+    {
+      "epoch": 0.3440504576571033,
+      "learning_rate": 0.002205846101604738,
+      "loss": 0.9713,
+      "step": 4473
+    },
+    {
+      "epoch": 0.34412737481732175,
+      "learning_rate": 0.002205526255524227,
+      "loss": 1.053,
+      "step": 4474
+    },
+    {
+      "epoch": 0.3442042919775402,
+      "learning_rate": 0.002205206368247302,
+      "loss": 0.9828,
+      "step": 4475
+    },
+    {
+      "epoch": 0.3442812091377586,
+      "learning_rate": 0.002204886439792643,
+      "loss": 1.0335,
+      "step": 4476
+    },
+    {
+      "epoch": 0.3443581262979771,
+      "learning_rate": 0.0022045664701789295,
+      "loss": 1.0159,
+      "step": 4477
+    },
+    {
+      "epoch": 0.34443504345819553,
+      "learning_rate": 0.002204246459424846,
+      "loss": 0.9004,
+      "step": 4478
+    },
+    {
+      "epoch": 0.344511960618414,
+      "learning_rate": 0.002203926407549077,
+      "loss": 1.3003,
+      "step": 4479
+    },
+    {
+      "epoch": 0.3445888777786324,
+      "learning_rate": 0.002203606314570312,
+      "loss": 1.1376,
+      "step": 4480
+    },
+    {
+      "epoch": 0.34466579493885086,
+      "learning_rate": 0.002203286180507241,
+      "loss": 1.2093,
+      "step": 4481
+    },
+    {
+      "epoch": 0.3447427120990693,
+      "learning_rate": 0.0022029660053785568,
+      "loss": 0.9017,
+      "step": 4482
+    },
+    {
+      "epoch": 0.3448196292592878,
+      "learning_rate": 0.002202645789202955,
+      "loss": 1.3401,
+      "step": 4483
+    },
+    {
+      "epoch": 0.3448965464195062,
+      "learning_rate": 0.0022023255319991338,
+      "loss": 0.975,
+      "step": 4484
+    },
+    {
+      "epoch": 0.34497346357972464,
+      "learning_rate": 0.0022020052337857923,
+      "loss": 0.9787,
+      "step": 4485
+    },
+    {
+      "epoch": 0.3450503807399431,
+      "learning_rate": 0.0022016848945816334,
+      "loss": 0.9551,
+      "step": 4486
+    },
+    {
+      "epoch": 0.3451272979001615,
+      "learning_rate": 0.0022013645144053633,
+      "loss": 0.9308,
+      "step": 4487
+    },
+    {
+      "epoch": 0.34520421506037996,
+      "learning_rate": 0.002201044093275687,
+      "loss": 1.2331,
+      "step": 4488
+    },
+    {
+      "epoch": 0.3452811322205984,
+      "learning_rate": 0.002200723631211316,
+      "loss": 0.8028,
+      "step": 4489
+    },
+    {
+      "epoch": 0.3453580493808169,
+      "learning_rate": 0.002200403128230962,
+      "loss": 1.0846,
+      "step": 4490
+    },
+    {
+      "epoch": 0.3454349665410353,
+      "learning_rate": 0.0022000825843533395,
+      "loss": 1.3261,
+      "step": 4491
+    },
+    {
+      "epoch": 0.34551188370125374,
+      "learning_rate": 0.0021997619995971648,
+      "loss": 0.9967,
+      "step": 4492
+    },
+    {
+      "epoch": 0.3455888008614722,
+      "learning_rate": 0.0021994413739811576,
+      "loss": 1.2422,
+      "step": 4493
+    },
+    {
+      "epoch": 0.34566571802169066,
+      "learning_rate": 0.00219912070752404,
+      "loss": 1.0371,
+      "step": 4494
+    },
+    {
+      "epoch": 0.34574263518190906,
+      "learning_rate": 0.002198800000244536,
+      "loss": 1.0514,
+      "step": 4495
+    },
+    {
+      "epoch": 0.3458195523421275,
+      "learning_rate": 0.0021984792521613714,
+      "loss": 1.101,
+      "step": 4496
+    },
+    {
+      "epoch": 0.345896469502346,
+      "learning_rate": 0.002198158463293275,
+      "loss": 0.8815,
+      "step": 4497
+    },
+    {
+      "epoch": 0.34597338666256444,
+      "learning_rate": 0.002197837633658979,
+      "loss": 1.0826,
+      "step": 4498
+    },
+    {
+      "epoch": 0.34605030382278285,
+      "learning_rate": 0.002197516763277216,
+      "loss": 1.1511,
+      "step": 4499
+    },
+    {
+      "epoch": 0.3461272209830013,
+      "learning_rate": 0.002197195852166722,
+      "loss": 0.7556,
+      "step": 4500
+    },
+    {
+      "epoch": 0.34620413814321976,
+      "learning_rate": 0.0021968749003462356,
+      "loss": 1.0577,
+      "step": 4501
+    },
+    {
+      "epoch": 0.3462810553034382,
+      "learning_rate": 0.0021965539078344976,
+      "loss": 1.1914,
+      "step": 4502
+    },
+    {
+      "epoch": 0.3463579724636566,
+      "learning_rate": 0.0021962328746502513,
+      "loss": 0.974,
+      "step": 4503
+    },
+    {
+      "epoch": 0.3464348896238751,
+      "learning_rate": 0.002195911800812241,
+      "loss": 1.6516,
+      "step": 4504
+    },
+    {
+      "epoch": 0.34651180678409355,
+      "learning_rate": 0.0021955906863392162,
+      "loss": 1.2723,
+      "step": 4505
+    },
+    {
+      "epoch": 0.34658872394431195,
+      "learning_rate": 0.0021952695312499266,
+      "loss": 1.1857,
+      "step": 4506
+    },
+    {
+      "epoch": 0.3466656411045304,
+      "learning_rate": 0.002194948335563124,
+      "loss": 1.0173,
+      "step": 4507
+    },
+    {
+      "epoch": 0.34674255826474887,
+      "learning_rate": 0.0021946270992975634,
+      "loss": 1.3723,
+      "step": 4508
+    },
+    {
+      "epoch": 0.3468194754249673,
+      "learning_rate": 0.002194305822472003,
+      "loss": 1.0406,
+      "step": 4509
+    },
+    {
+      "epoch": 0.34689639258518573,
+      "learning_rate": 0.0021939845051052024,
+      "loss": 0.9578,
+      "step": 4510
+    },
+    {
+      "epoch": 0.3469733097454042,
+      "learning_rate": 0.0021936631472159224,
+      "loss": 1.8563,
+      "step": 4511
+    },
+    {
+      "epoch": 0.34705022690562265,
+      "learning_rate": 0.002193341748822929,
+      "loss": 1.5305,
+      "step": 4512
+    },
+    {
+      "epoch": 0.3471271440658411,
+      "learning_rate": 0.002193020309944988,
+      "loss": 1.223,
+      "step": 4513
+    },
+    {
+      "epoch": 0.3472040612260595,
+      "learning_rate": 0.002192698830600869,
+      "loss": 1.1849,
+      "step": 4514
+    },
+    {
+      "epoch": 0.34728097838627797,
+      "learning_rate": 0.0021923773108093432,
+      "loss": 1.4335,
+      "step": 4515
+    },
+    {
+      "epoch": 0.34735789554649643,
+      "learning_rate": 0.002192055750589185,
+      "loss": 0.8563,
+      "step": 4516
+    },
+    {
+      "epoch": 0.3474348127067149,
+      "learning_rate": 0.0021917341499591695,
+      "loss": 0.8439,
+      "step": 4517
+    },
+    {
+      "epoch": 0.3475117298669333,
+      "learning_rate": 0.0021914125089380772,
+      "loss": 0.7949,
+      "step": 4518
+    },
+    {
+      "epoch": 0.34758864702715175,
+      "learning_rate": 0.0021910908275446874,
+      "loss": 1.1507,
+      "step": 4519
+    },
+    {
+      "epoch": 0.3476655641873702,
+      "learning_rate": 0.0021907691057977837,
+      "loss": 1.2038,
+      "step": 4520
+    },
+    {
+      "epoch": 0.34774248134758867,
+      "learning_rate": 0.002190447343716152,
+      "loss": 1.1221,
+      "step": 4521
+    },
+    {
+      "epoch": 0.3478193985078071,
+      "learning_rate": 0.002190125541318581,
+      "loss": 1.1422,
+      "step": 4522
+    },
+    {
+      "epoch": 0.34789631566802554,
+      "learning_rate": 0.00218980369862386,
+      "loss": 1.2733,
+      "step": 4523
+    },
+    {
+      "epoch": 0.347973232828244,
+      "learning_rate": 0.002189481815650782,
+      "loss": 1.0066,
+      "step": 4524
+    },
+    {
+      "epoch": 0.3480501499884624,
+      "learning_rate": 0.0021891598924181424,
+      "loss": 1.1536,
+      "step": 4525
+    },
+    {
+      "epoch": 0.34812706714868086,
+      "learning_rate": 0.0021888379289447383,
+      "loss": 1.15,
+      "step": 4526
+    },
+    {
+      "epoch": 0.3482039843088993,
+      "learning_rate": 0.0021885159252493696,
+      "loss": 1.2677,
+      "step": 4527
+    },
+    {
+      "epoch": 0.3482809014691178,
+      "learning_rate": 0.0021881938813508384,
+      "loss": 1.1407,
+      "step": 4528
+    },
+    {
+      "epoch": 0.3483578186293362,
+      "learning_rate": 0.0021878717972679497,
+      "loss": 0.9848,
+      "step": 4529
+    },
+    {
+      "epoch": 0.34843473578955464,
+      "learning_rate": 0.0021875496730195095,
+      "loss": 1.2132,
+      "step": 4530
+    },
+    {
+      "epoch": 0.3485116529497731,
+      "learning_rate": 0.0021872275086243275,
+      "loss": 1.2289,
+      "step": 4531
+    },
+    {
+      "epoch": 0.34858857010999156,
+      "learning_rate": 0.002186905304101215,
+      "loss": 1.279,
+      "step": 4532
+    },
+    {
+      "epoch": 0.34866548727020996,
+      "learning_rate": 0.002186583059468986,
+      "loss": 1.3782,
+      "step": 4533
+    },
+    {
+      "epoch": 0.3487424044304284,
+      "learning_rate": 0.0021862607747464564,
+      "loss": 1.2075,
+      "step": 4534
+    },
+    {
+      "epoch": 0.3488193215906469,
+      "learning_rate": 0.002185938449952445,
+      "loss": 1.2174,
+      "step": 4535
+    },
+    {
+      "epoch": 0.34889623875086534,
+      "learning_rate": 0.0021856160851057727,
+      "loss": 1.2171,
+      "step": 4536
+    },
+    {
+      "epoch": 0.34897315591108374,
+      "learning_rate": 0.002185293680225263,
+      "loss": 0.6313,
+      "step": 4537
+    },
+    {
+      "epoch": 0.3490500730713022,
+      "learning_rate": 0.002184971235329741,
+      "loss": 0.7088,
+      "step": 4538
+    },
+    {
+      "epoch": 0.34912699023152066,
+      "learning_rate": 0.002184648750438034,
+      "loss": 1.198,
+      "step": 4539
+    },
+    {
+      "epoch": 0.3492039073917391,
+      "learning_rate": 0.0021843262255689735,
+      "loss": 0.9856,
+      "step": 4540
+    },
+    {
+      "epoch": 0.3492808245519575,
+      "learning_rate": 0.0021840036607413916,
+      "loss": 1.2084,
+      "step": 4541
+    },
+    {
+      "epoch": 0.349357741712176,
+      "learning_rate": 0.002183681055974123,
+      "loss": 1.125,
+      "step": 4542
+    },
+    {
+      "epoch": 0.34943465887239444,
+      "learning_rate": 0.002183358411286005,
+      "loss": 1.0228,
+      "step": 4543
+    },
+    {
+      "epoch": 0.3495115760326129,
+      "learning_rate": 0.002183035726695877,
+      "loss": 1.4435,
+      "step": 4544
+    },
+    {
+      "epoch": 0.3495884931928313,
+      "learning_rate": 0.0021827130022225812,
+      "loss": 1.1466,
+      "step": 4545
+    },
+    {
+      "epoch": 0.34966541035304977,
+      "learning_rate": 0.002182390237884962,
+      "loss": 0.8455,
+      "step": 4546
+    },
+    {
+      "epoch": 0.3497423275132682,
+      "learning_rate": 0.0021820674337018654,
+      "loss": 1.5281,
+      "step": 4547
+    },
+    {
+      "epoch": 0.34981924467348663,
+      "learning_rate": 0.0021817445896921404,
+      "loss": 0.7684,
+      "step": 4548
+    },
+    {
+      "epoch": 0.3498961618337051,
+      "learning_rate": 0.0021814217058746385,
+      "loss": 1.2232,
+      "step": 4549
+    },
+    {
+      "epoch": 0.34997307899392355,
+      "learning_rate": 0.0021810987822682127,
+      "loss": 1.0272,
+      "step": 4550
+    },
+    {
+      "epoch": 0.350049996154142,
+      "learning_rate": 0.0021807758188917194,
+      "loss": 0.9636,
+      "step": 4551
+    },
+    {
+      "epoch": 0.3501269133143604,
+      "learning_rate": 0.002180452815764017,
+      "loss": 1.3657,
+      "step": 4552
+    },
+    {
+      "epoch": 0.35020383047457887,
+      "learning_rate": 0.0021801297729039655,
+      "loss": 1.1887,
+      "step": 4553
+    },
+    {
+      "epoch": 0.35028074763479733,
+      "learning_rate": 0.0021798066903304268,
+      "loss": 1.1966,
+      "step": 4554
+    },
+    {
+      "epoch": 0.3503576647950158,
+      "learning_rate": 0.0021794835680622677,
+      "loss": 1.034,
+      "step": 4555
+    },
+    {
+      "epoch": 0.3504345819552342,
+      "learning_rate": 0.0021791604061183545,
+      "loss": 1.2712,
+      "step": 4556
+    },
+    {
+      "epoch": 0.35051149911545265,
+      "learning_rate": 0.0021788372045175577,
+      "loss": 0.8929,
+      "step": 4557
+    },
+    {
+      "epoch": 0.3505884162756711,
+      "learning_rate": 0.0021785139632787493,
+      "loss": 1.1922,
+      "step": 4558
+    },
+    {
+      "epoch": 0.35066533343588957,
+      "learning_rate": 0.0021781906824208026,
+      "loss": 0.8638,
+      "step": 4559
+    },
+    {
+      "epoch": 0.350742250596108,
+      "learning_rate": 0.002177867361962596,
+      "loss": 1.1992,
+      "step": 4560
+    },
+    {
+      "epoch": 0.35081916775632643,
+      "learning_rate": 0.0021775440019230073,
+      "loss": 1.2379,
+      "step": 4561
+    },
+    {
+      "epoch": 0.3508960849165449,
+      "learning_rate": 0.0021772206023209185,
+      "loss": 1.1376,
+      "step": 4562
+    },
+    {
+      "epoch": 0.35097300207676335,
+      "learning_rate": 0.002176897163175212,
+      "loss": 1.2832,
+      "step": 4563
+    },
+    {
+      "epoch": 0.35104991923698176,
+      "learning_rate": 0.0021765736845047764,
+      "loss": 1.3489,
+      "step": 4564
+    },
+    {
+      "epoch": 0.3511268363972002,
+      "learning_rate": 0.0021762501663284968,
+      "loss": 0.9143,
+      "step": 4565
+    },
+    {
+      "epoch": 0.3512037535574187,
+      "learning_rate": 0.002175926608665266,
+      "loss": 1.1331,
+      "step": 4566
+    },
+    {
+      "epoch": 0.3512806707176371,
+      "learning_rate": 0.002175603011533976,
+      "loss": 1.1869,
+      "step": 4567
+    },
+    {
+      "epoch": 0.35135758787785554,
+      "learning_rate": 0.0021752793749535214,
+      "loss": 0.9946,
+      "step": 4568
+    },
+    {
+      "epoch": 0.351434505038074,
+      "learning_rate": 0.002174955698942801,
+      "loss": 1.334,
+      "step": 4569
+    },
+    {
+      "epoch": 0.35151142219829246,
+      "learning_rate": 0.0021746319835207134,
+      "loss": 0.83,
+      "step": 4570
+    },
+    {
+      "epoch": 0.35158833935851086,
+      "learning_rate": 0.002174308228706162,
+      "loss": 1.3723,
+      "step": 4571
+    },
+    {
+      "epoch": 0.3516652565187293,
+      "learning_rate": 0.00217398443451805,
+      "loss": 1.5225,
+      "step": 4572
+    },
+    {
+      "epoch": 0.3517421736789478,
+      "learning_rate": 0.002173660600975284,
+      "loss": 1.3333,
+      "step": 4573
+    },
+    {
+      "epoch": 0.35181909083916624,
+      "learning_rate": 0.002173336728096774,
+      "loss": 0.8326,
+      "step": 4574
+    },
+    {
+      "epoch": 0.35189600799938464,
+      "learning_rate": 0.002173012815901431,
+      "loss": 1.3905,
+      "step": 4575
+    },
+    {
+      "epoch": 0.3519729251596031,
+      "learning_rate": 0.002172688864408167,
+      "loss": 0.7822,
+      "step": 4576
+    },
+    {
+      "epoch": 0.35204984231982156,
+      "learning_rate": 0.0021723648736358997,
+      "loss": 1.1313,
+      "step": 4577
+    },
+    {
+      "epoch": 0.35212675948004,
+      "learning_rate": 0.002172040843603547,
+      "loss": 1.4056,
+      "step": 4578
+    },
+    {
+      "epoch": 0.3522036766402584,
+      "learning_rate": 0.002171716774330029,
+      "loss": 1.1409,
+      "step": 4579
+    },
+    {
+      "epoch": 0.3522805938004769,
+      "learning_rate": 0.0021713926658342687,
+      "loss": 1.4108,
+      "step": 4580
+    },
+    {
+      "epoch": 0.35235751096069534,
+      "learning_rate": 0.0021710685181351905,
+      "loss": 1.1529,
+      "step": 4581
+    },
+    {
+      "epoch": 0.3524344281209138,
+      "learning_rate": 0.0021707443312517223,
+      "loss": 1.3354,
+      "step": 4582
+    },
+    {
+      "epoch": 0.3525113452811322,
+      "learning_rate": 0.0021704201052027937,
+      "loss": 1.0511,
+      "step": 4583
+    },
+    {
+      "epoch": 0.35258826244135066,
+      "learning_rate": 0.0021700958400073366,
+      "loss": 1.4036,
+      "step": 4584
+    },
+    {
+      "epoch": 0.3526651796015691,
+      "learning_rate": 0.0021697715356842847,
+      "loss": 1.0696,
+      "step": 4585
+    },
+    {
+      "epoch": 0.3527420967617876,
+      "learning_rate": 0.002169447192252575,
+      "loss": 1.4884,
+      "step": 4586
+    },
+    {
+      "epoch": 0.352819013922006,
+      "learning_rate": 0.002169122809731146,
+      "loss": 1.3654,
+      "step": 4587
+    },
+    {
+      "epoch": 0.35289593108222445,
+      "learning_rate": 0.002168798388138939,
+      "loss": 0.849,
+      "step": 4588
+    },
+    {
+      "epoch": 0.3529728482424429,
+      "learning_rate": 0.0021684739274948966,
+      "loss": 1.448,
+      "step": 4589
+    },
+    {
+      "epoch": 0.3530497654026613,
+      "learning_rate": 0.002168149427817966,
+      "loss": 0.9463,
+      "step": 4590
+    },
+    {
+      "epoch": 0.35312668256287977,
+      "learning_rate": 0.0021678248891270928,
+      "loss": 1.4599,
+      "step": 4591
+    },
+    {
+      "epoch": 0.3532035997230982,
+      "learning_rate": 0.0021675003114412284,
+      "loss": 1.377,
+      "step": 4592
+    },
+    {
+      "epoch": 0.3532805168833167,
+      "learning_rate": 0.0021671756947793256,
+      "loss": 1.0482,
+      "step": 4593
+    },
+    {
+      "epoch": 0.3533574340435351,
+      "learning_rate": 0.0021668510391603386,
+      "loss": 1.351,
+      "step": 4594
+    },
+    {
+      "epoch": 0.35343435120375355,
+      "learning_rate": 0.0021665263446032246,
+      "loss": 0.9677,
+      "step": 4595
+    },
+    {
+      "epoch": 0.353511268363972,
+      "learning_rate": 0.0021662016111269424,
+      "loss": 1.4628,
+      "step": 4596
+    },
+    {
+      "epoch": 0.35358818552419047,
+      "learning_rate": 0.0021658768387504542,
+      "loss": 1.2718,
+      "step": 4597
+    },
+    {
+      "epoch": 0.35366510268440887,
+      "learning_rate": 0.0021655520274927232,
+      "loss": 1.4732,
+      "step": 4598
+    },
+    {
+      "epoch": 0.35374201984462733,
+      "learning_rate": 0.0021652271773727155,
+      "loss": 0.6943,
+      "step": 4599
+    },
+    {
+      "epoch": 0.3538189370048458,
+      "learning_rate": 0.0021649022884094,
+      "loss": 1.192,
+      "step": 4600
+    },
+    {
+      "epoch": 0.35389585416506425,
+      "learning_rate": 0.0021645773606217464,
+      "loss": 1.2953,
+      "step": 4601
+    },
+    {
+      "epoch": 0.35397277132528265,
+      "learning_rate": 0.0021642523940287287,
+      "loss": 1.2194,
+      "step": 4602
+    },
+    {
+      "epoch": 0.3540496884855011,
+      "learning_rate": 0.002163927388649321,
+      "loss": 1.2613,
+      "step": 4603
+    },
+    {
+      "epoch": 0.35412660564571957,
+      "learning_rate": 0.0021636023445025013,
+      "loss": 0.9221,
+      "step": 4604
+    },
+    {
+      "epoch": 0.35420352280593803,
+      "learning_rate": 0.002163277261607249,
+      "loss": 1.1065,
+      "step": 4605
+    },
+    {
+      "epoch": 0.35428043996615644,
+      "learning_rate": 0.002162952139982547,
+      "loss": 1.0068,
+      "step": 4606
+    },
+    {
+      "epoch": 0.3543573571263749,
+      "learning_rate": 0.0021626269796473775,
+      "loss": 1.3094,
+      "step": 4607
+    },
+    {
+      "epoch": 0.35443427428659335,
+      "learning_rate": 0.0021623017806207284,
+      "loss": 1.2667,
+      "step": 4608
+    },
+    {
+      "epoch": 0.35451119144681176,
+      "learning_rate": 0.0021619765429215882,
+      "loss": 1.2654,
+      "step": 4609
+    },
+    {
+      "epoch": 0.3545881086070302,
+      "learning_rate": 0.002161651266568948,
+      "loss": 0.8474,
+      "step": 4610
+    },
+    {
+      "epoch": 0.3546650257672487,
+      "learning_rate": 0.0021613259515818007,
+      "loss": 1.1862,
+      "step": 4611
+    },
+    {
+      "epoch": 0.35474194292746714,
+      "learning_rate": 0.002161000597979141,
+      "loss": 1.1444,
+      "step": 4612
+    },
+    {
+      "epoch": 0.35481886008768554,
+      "learning_rate": 0.002160675205779969,
+      "loss": 1.1831,
+      "step": 4613
+    },
+    {
+      "epoch": 0.354895777247904,
+      "learning_rate": 0.0021603497750032822,
+      "loss": 1.1047,
+      "step": 4614
+    },
+    {
+      "epoch": 0.35497269440812246,
+      "learning_rate": 0.002160024305668084,
+      "loss": 1.4083,
+      "step": 4615
+    },
+    {
+      "epoch": 0.3550496115683409,
+      "learning_rate": 0.002159698797793379,
+      "loss": 0.9487,
+      "step": 4616
+    },
+    {
+      "epoch": 0.3551265287285593,
+      "learning_rate": 0.0021593732513981736,
+      "loss": 1.0903,
+      "step": 4617
+    },
+    {
+      "epoch": 0.3552034458887778,
+      "learning_rate": 0.0021590476665014767,
+      "loss": 1.258,
+      "step": 4618
+    },
+    {
+      "epoch": 0.35528036304899624,
+      "learning_rate": 0.0021587220431222996,
+      "loss": 1.1773,
+      "step": 4619
+    },
+    {
+      "epoch": 0.3553572802092147,
+      "learning_rate": 0.002158396381279656,
+      "loss": 1.1955,
+      "step": 4620
+    },
+    {
+      "epoch": 0.3554341973694331,
+      "learning_rate": 0.0021580706809925616,
+      "loss": 0.899,
+      "step": 4621
+    },
+    {
+      "epoch": 0.35551111452965156,
+      "learning_rate": 0.0021577449422800344,
+      "loss": 0.905,
+      "step": 4622
+    },
+    {
+      "epoch": 0.35558803168987,
+      "learning_rate": 0.0021574191651610943,
+      "loss": 0.9677,
+      "step": 4623
+    },
+    {
+      "epoch": 0.3556649488500885,
+      "learning_rate": 0.002157093349654764,
+      "loss": 1.1556,
+      "step": 4624
+    },
+    {
+      "epoch": 0.3557418660103069,
+      "learning_rate": 0.002156767495780069,
+      "loss": 1.2715,
+      "step": 4625
+    },
+    {
+      "epoch": 0.35581878317052534,
+      "learning_rate": 0.0021564416035560344,
+      "loss": 1.2876,
+      "step": 4626
+    },
+    {
+      "epoch": 0.3558957003307438,
+      "learning_rate": 0.002156115673001691,
+      "loss": 1.1753,
+      "step": 4627
+    },
+    {
+      "epoch": 0.3559726174909622,
+      "learning_rate": 0.0021557897041360697,
+      "loss": 1.2324,
+      "step": 4628
+    },
+    {
+      "epoch": 0.35604953465118067,
+      "learning_rate": 0.0021554636969782037,
+      "loss": 0.9744,
+      "step": 4629
+    },
+    {
+      "epoch": 0.3561264518113991,
+      "learning_rate": 0.0021551376515471297,
+      "loss": 1.1218,
+      "step": 4630
+    },
+    {
+      "epoch": 0.3562033689716176,
+      "learning_rate": 0.002154811567861885,
+      "loss": 1.1404,
+      "step": 4631
+    },
+    {
+      "epoch": 0.356280286131836,
+      "learning_rate": 0.002154485445941511,
+      "loss": 1.2168,
+      "step": 4632
+    },
+    {
+      "epoch": 0.35635720329205445,
+      "learning_rate": 0.0021541592858050494,
+      "loss": 0.9979,
+      "step": 4633
+    },
+    {
+      "epoch": 0.3564341204522729,
+      "learning_rate": 0.002153833087471545,
+      "loss": 0.9219,
+      "step": 4634
+    },
+    {
+      "epoch": 0.35651103761249137,
+      "learning_rate": 0.002153506850960046,
+      "loss": 1.2117,
+      "step": 4635
+    },
+    {
+      "epoch": 0.35658795477270977,
+      "learning_rate": 0.0021531805762896004,
+      "loss": 1.0654,
+      "step": 4636
+    },
+    {
+      "epoch": 0.35666487193292823,
+      "learning_rate": 0.0021528542634792606,
+      "loss": 1.3072,
+      "step": 4637
+    },
+    {
+      "epoch": 0.3567417890931467,
+      "learning_rate": 0.0021525279125480796,
+      "loss": 1.0846,
+      "step": 4638
+    },
+    {
+      "epoch": 0.35681870625336515,
+      "learning_rate": 0.002152201523515114,
+      "loss": 1.2969,
+      "step": 4639
+    },
+    {
+      "epoch": 0.35689562341358355,
+      "learning_rate": 0.0021518750963994216,
+      "loss": 1.1594,
+      "step": 4640
+    },
+    {
+      "epoch": 0.356972540573802,
+      "learning_rate": 0.0021515486312200633,
+      "loss": 1.2355,
+      "step": 4641
+    },
+    {
+      "epoch": 0.35704945773402047,
+      "learning_rate": 0.0021512221279961,
+      "loss": 0.7651,
+      "step": 4642
+    },
+    {
+      "epoch": 0.35712637489423893,
+      "learning_rate": 0.0021508955867465995,
+      "loss": 1.0978,
+      "step": 4643
+    },
+    {
+      "epoch": 0.35720329205445733,
+      "learning_rate": 0.0021505690074906272,
+      "loss": 1.047,
+      "step": 4644
+    },
+    {
+      "epoch": 0.3572802092146758,
+      "learning_rate": 0.0021502423902472517,
+      "loss": 0.964,
+      "step": 4645
+    },
+    {
+      "epoch": 0.35735712637489425,
+      "learning_rate": 0.0021499157350355464,
+      "loss": 1.0864,
+      "step": 4646
+    },
+    {
+      "epoch": 0.3574340435351127,
+      "learning_rate": 0.002149589041874583,
+      "loss": 1.4225,
+      "step": 4647
+    },
+    {
+      "epoch": 0.3575109606953311,
+      "learning_rate": 0.0021492623107834397,
+      "loss": 1.3418,
+      "step": 4648
+    },
+    {
+      "epoch": 0.3575878778555496,
+      "learning_rate": 0.0021489355417811918,
+      "loss": 1.0769,
+      "step": 4649
+    },
+    {
+      "epoch": 0.35766479501576803,
+      "learning_rate": 0.0021486087348869224,
+      "loss": 1.0327,
+      "step": 4650
+    },
+    {
+      "epoch": 0.35774171217598644,
+      "learning_rate": 0.002148281890119713,
+      "loss": 1.295,
+      "step": 4651
+    },
+    {
+      "epoch": 0.3578186293362049,
+      "learning_rate": 0.002147955007498648,
+      "loss": 1.1204,
+      "step": 4652
+    },
+    {
+      "epoch": 0.35789554649642336,
+      "learning_rate": 0.0021476280870428147,
+      "loss": 1.094,
+      "step": 4653
+    },
+    {
+      "epoch": 0.3579724636566418,
+      "learning_rate": 0.002147301128771303,
+      "loss": 1.2706,
+      "step": 4654
+    },
+    {
+      "epoch": 0.3580493808168602,
+      "learning_rate": 0.0021469741327032037,
+      "loss": 1.2103,
+      "step": 4655
+    },
+    {
+      "epoch": 0.3581262979770787,
+      "learning_rate": 0.0021466470988576103,
+      "loss": 1.4753,
+      "step": 4656
+    },
+    {
+      "epoch": 0.35820321513729714,
+      "learning_rate": 0.002146320027253619,
+      "loss": 1.0931,
+      "step": 4657
+    },
+    {
+      "epoch": 0.3582801322975156,
+      "learning_rate": 0.0021459929179103274,
+      "loss": 1.0186,
+      "step": 4658
+    },
+    {
+      "epoch": 0.358357049457734,
+      "learning_rate": 0.002145665770846837,
+      "loss": 1.0179,
+      "step": 4659
+    },
+    {
+      "epoch": 0.35843396661795246,
+      "learning_rate": 0.0021453385860822484,
+      "loss": 1.1274,
+      "step": 4660
+    },
+    {
+      "epoch": 0.3585108837781709,
+      "learning_rate": 0.0021450113636356676,
+      "loss": 0.8537,
+      "step": 4661
+    },
+    {
+      "epoch": 0.3585878009383894,
+      "learning_rate": 0.002144684103526201,
+      "loss": 1.2804,
+      "step": 4662
+    },
+    {
+      "epoch": 0.3586647180986078,
+      "learning_rate": 0.0021443568057729575,
+      "loss": 0.9907,
+      "step": 4663
+    },
+    {
+      "epoch": 0.35874163525882624,
+      "learning_rate": 0.002144029470395049,
+      "loss": 0.9742,
+      "step": 4664
+    },
+    {
+      "epoch": 0.3588185524190447,
+      "learning_rate": 0.0021437020974115887,
+      "loss": 1.071,
+      "step": 4665
+    },
+    {
+      "epoch": 0.35889546957926316,
+      "learning_rate": 0.002143374686841691,
+      "loss": 1.1727,
+      "step": 4666
+    },
+    {
+      "epoch": 0.35897238673948156,
+      "learning_rate": 0.002143047238704476,
+      "loss": 0.862,
+      "step": 4667
+    },
+    {
+      "epoch": 0.3590493038997,
+      "learning_rate": 0.002142719753019062,
+      "loss": 1.2982,
+      "step": 4668
+    },
+    {
+      "epoch": 0.3591262210599185,
+      "learning_rate": 0.002142392229804572,
+      "loss": 0.9353,
+      "step": 4669
+    },
+    {
+      "epoch": 0.3592031382201369,
+      "learning_rate": 0.0021420646690801302,
+      "loss": 1.0221,
+      "step": 4670
+    },
+    {
+      "epoch": 0.35928005538035535,
+      "learning_rate": 0.0021417370708648634,
+      "loss": 1.4891,
+      "step": 4671
+    },
+    {
+      "epoch": 0.3593569725405738,
+      "learning_rate": 0.0021414094351778997,
+      "loss": 0.89,
+      "step": 4672
+    },
+    {
+      "epoch": 0.35943388970079226,
+      "learning_rate": 0.0021410817620383703,
+      "loss": 0.9063,
+      "step": 4673
+    },
+    {
+      "epoch": 0.35951080686101067,
+      "learning_rate": 0.0021407540514654095,
+      "loss": 0.9539,
+      "step": 4674
+    },
+    {
+      "epoch": 0.3595877240212291,
+      "learning_rate": 0.002140426303478152,
+      "loss": 0.9472,
+      "step": 4675
+    },
+    {
+      "epoch": 0.3596646411814476,
+      "learning_rate": 0.0021400985180957347,
+      "loss": 1.0589,
+      "step": 4676
+    },
+    {
+      "epoch": 0.35974155834166605,
+      "learning_rate": 0.0021397706953372978,
+      "loss": 1.1529,
+      "step": 4677
+    },
+    {
+      "epoch": 0.35981847550188445,
+      "learning_rate": 0.0021394428352219837,
+      "loss": 1.4477,
+      "step": 4678
+    },
+    {
+      "epoch": 0.3598953926621029,
+      "learning_rate": 0.0021391149377689355,
+      "loss": 1.0915,
+      "step": 4679
+    },
+    {
+      "epoch": 0.35997230982232137,
+      "learning_rate": 0.0021387870029973004,
+      "loss": 1.4642,
+      "step": 4680
+    },
+    {
+      "epoch": 0.3600492269825398,
+      "learning_rate": 0.002138459030926226,
+      "loss": 0.9607,
+      "step": 4681
+    },
+    {
+      "epoch": 0.36012614414275823,
+      "learning_rate": 0.0021381310215748644,
+      "loss": 1.2426,
+      "step": 4682
+    },
+    {
+      "epoch": 0.3602030613029767,
+      "learning_rate": 0.0021378029749623663,
+      "loss": 1.3444,
+      "step": 4683
+    },
+    {
+      "epoch": 0.36027997846319515,
+      "learning_rate": 0.0021374748911078883,
+      "loss": 1.2366,
+      "step": 4684
+    },
+    {
+      "epoch": 0.3603568956234136,
+      "learning_rate": 0.0021371467700305874,
+      "loss": 1.1309,
+      "step": 4685
+    },
+    {
+      "epoch": 0.360433812783632,
+      "learning_rate": 0.002136818611749622,
+      "loss": 0.8353,
+      "step": 4686
+    },
+    {
+      "epoch": 0.36051072994385047,
+      "learning_rate": 0.0021364904162841543,
+      "loss": 1.1354,
+      "step": 4687
+    },
+    {
+      "epoch": 0.36058764710406893,
+      "learning_rate": 0.0021361621836533478,
+      "loss": 1.1264,
+      "step": 4688
+    },
+    {
+      "epoch": 0.36066456426428734,
+      "learning_rate": 0.0021358339138763684,
+      "loss": 0.9663,
+      "step": 4689
+    },
+    {
+      "epoch": 0.3607414814245058,
+      "learning_rate": 0.0021355056069723845,
+      "loss": 1.0954,
+      "step": 4690
+    },
+    {
+      "epoch": 0.36081839858472425,
+      "learning_rate": 0.0021351772629605655,
+      "loss": 1.057,
+      "step": 4691
+    },
+    {
+      "epoch": 0.3608953157449427,
+      "learning_rate": 0.0021348488818600845,
+      "loss": 1.1182,
+      "step": 4692
+    },
+    {
+      "epoch": 0.3609722329051611,
+      "learning_rate": 0.0021345204636901154,
+      "loss": 1.0405,
+      "step": 4693
+    },
+    {
+      "epoch": 0.3610491500653796,
+      "learning_rate": 0.002134192008469835,
+      "loss": 1.2886,
+      "step": 4694
+    },
+    {
+      "epoch": 0.36112606722559804,
+      "learning_rate": 0.0021338635162184223,
+      "loss": 1.0584,
+      "step": 4695
+    },
+    {
+      "epoch": 0.3612029843858165,
+      "learning_rate": 0.0021335349869550588,
+      "loss": 1.389,
+      "step": 4696
+    },
+    {
+      "epoch": 0.3612799015460349,
+      "learning_rate": 0.002133206420698927,
+      "loss": 1.0482,
+      "step": 4697
+    },
+    {
+      "epoch": 0.36135681870625336,
+      "learning_rate": 0.002132877817469212,
+      "loss": 0.8862,
+      "step": 4698
+    },
+    {
+      "epoch": 0.3614337358664718,
+      "learning_rate": 0.0021325491772851016,
+      "loss": 1.2062,
+      "step": 4699
+    },
+    {
+      "epoch": 0.3615106530266903,
+      "learning_rate": 0.0021322205001657857,
+      "loss": 1.4097,
+      "step": 4700
+    },
+    {
+      "epoch": 0.3615875701869087,
+      "learning_rate": 0.0021318917861304566,
+      "loss": 1.3983,
+      "step": 4701
+    },
+    {
+      "epoch": 0.36166448734712714,
+      "learning_rate": 0.002131563035198307,
+      "loss": 1.1813,
+      "step": 4702
+    },
+    {
+      "epoch": 0.3617414045073456,
+      "learning_rate": 0.0021312342473885333,
+      "loss": 1.4315,
+      "step": 4703
+    },
+    {
+      "epoch": 0.36181832166756406,
+      "learning_rate": 0.0021309054227203338,
+      "loss": 1.0769,
+      "step": 4704
+    },
+    {
+      "epoch": 0.36189523882778246,
+      "learning_rate": 0.0021305765612129103,
+      "loss": 1.2647,
+      "step": 4705
+    },
+    {
+      "epoch": 0.3619721559880009,
+      "learning_rate": 0.002130247662885463,
+      "loss": 1.2099,
+      "step": 4706
+    },
+    {
+      "epoch": 0.3620490731482194,
+      "learning_rate": 0.002129918727757199,
+      "loss": 1.0531,
+      "step": 4707
+    },
+    {
+      "epoch": 0.36212599030843784,
+      "learning_rate": 0.002129589755847323,
+      "loss": 1.0395,
+      "step": 4708
+    },
+    {
+      "epoch": 0.36220290746865624,
+      "learning_rate": 0.002129260747175046,
+      "loss": 1.4338,
+      "step": 4709
+    },
+    {
+      "epoch": 0.3622798246288747,
+      "learning_rate": 0.0021289317017595772,
+      "loss": 1.1014,
+      "step": 4710
+    },
+    {
+      "epoch": 0.36235674178909316,
+      "learning_rate": 0.002128602619620131,
+      "loss": 1.3059,
+      "step": 4711
+    },
+    {
+      "epoch": 0.36243365894931157,
+      "learning_rate": 0.0021282735007759233,
+      "loss": 1.2196,
+      "step": 4712
+    },
+    {
+      "epoch": 0.36251057610953,
+      "learning_rate": 0.0021279443452461707,
+      "loss": 1.1374,
+      "step": 4713
+    },
+    {
+      "epoch": 0.3625874932697485,
+      "learning_rate": 0.0021276151530500937,
+      "loss": 1.1438,
+      "step": 4714
+    },
+    {
+      "epoch": 0.36266441042996694,
+      "learning_rate": 0.0021272859242069133,
+      "loss": 1.2911,
+      "step": 4715
+    },
+    {
+      "epoch": 0.36274132759018535,
+      "learning_rate": 0.0021269566587358547,
+      "loss": 1.1752,
+      "step": 4716
+    },
+    {
+      "epoch": 0.3628182447504038,
+      "learning_rate": 0.002126627356656143,
+      "loss": 1.3372,
+      "step": 4717
+    },
+    {
+      "epoch": 0.36289516191062227,
+      "learning_rate": 0.0021262980179870064,
+      "loss": 1.0532,
+      "step": 4718
+    },
+    {
+      "epoch": 0.3629720790708407,
+      "learning_rate": 0.0021259686427476765,
+      "loss": 1.0698,
+      "step": 4719
+    },
+    {
+      "epoch": 0.36304899623105913,
+      "learning_rate": 0.002125639230957385,
+      "loss": 0.8697,
+      "step": 4720
+    },
+    {
+      "epoch": 0.3631259133912776,
+      "learning_rate": 0.002125309782635367,
+      "loss": 1.317,
+      "step": 4721
+    },
+    {
+      "epoch": 0.36320283055149605,
+      "learning_rate": 0.0021249802978008585,
+      "loss": 1.3131,
+      "step": 4722
+    },
+    {
+      "epoch": 0.3632797477117145,
+      "learning_rate": 0.0021246507764731,
+      "loss": 1.5172,
+      "step": 4723
+    },
+    {
+      "epoch": 0.3633566648719329,
+      "learning_rate": 0.002124321218671331,
+      "loss": 1.3545,
+      "step": 4724
+    },
+    {
+      "epoch": 0.36343358203215137,
+      "learning_rate": 0.0021239916244147955,
+      "loss": 1.3479,
+      "step": 4725
+    },
+    {
+      "epoch": 0.36351049919236983,
+      "learning_rate": 0.0021236619937227384,
+      "loss": 1.0363,
+      "step": 4726
+    },
+    {
+      "epoch": 0.3635874163525883,
+      "learning_rate": 0.002123332326614408,
+      "loss": 0.9226,
+      "step": 4727
+    },
+    {
+      "epoch": 0.3636643335128067,
+      "learning_rate": 0.002123002623109053,
+      "loss": 1.0268,
+      "step": 4728
+    },
+    {
+      "epoch": 0.36374125067302515,
+      "learning_rate": 0.0021226728832259264,
+      "loss": 1.3165,
+      "step": 4729
+    },
+    {
+      "epoch": 0.3638181678332436,
+      "learning_rate": 0.0021223431069842804,
+      "loss": 1.081,
+      "step": 4730
+    },
+    {
+      "epoch": 0.363895084993462,
+      "learning_rate": 0.002122013294403372,
+      "loss": 1.0528,
+      "step": 4731
+    },
+    {
+      "epoch": 0.3639720021536805,
+      "learning_rate": 0.0021216834455024592,
+      "loss": 1.3662,
+      "step": 4732
+    },
+    {
+      "epoch": 0.36404891931389893,
+      "learning_rate": 0.002121353560300802,
+      "loss": 1.4345,
+      "step": 4733
+    },
+    {
+      "epoch": 0.3641258364741174,
+      "learning_rate": 0.0021210236388176634,
+      "loss": 1.5348,
+      "step": 4734
+    },
+    {
+      "epoch": 0.3642027536343358,
+      "learning_rate": 0.002120693681072307,
+      "loss": 1.0566,
+      "step": 4735
+    },
+    {
+      "epoch": 0.36427967079455426,
+      "learning_rate": 0.0021203636870839996,
+      "loss": 1.2474,
+      "step": 4736
+    },
+    {
+      "epoch": 0.3643565879547727,
+      "learning_rate": 0.00212003365687201,
+      "loss": 1.013,
+      "step": 4737
+    },
+    {
+      "epoch": 0.3644335051149912,
+      "learning_rate": 0.0021197035904556086,
+      "loss": 0.9083,
+      "step": 4738
+    },
+    {
+      "epoch": 0.3645104222752096,
+      "learning_rate": 0.002119373487854069,
+      "loss": 1.1171,
+      "step": 4739
+    },
+    {
+      "epoch": 0.36458733943542804,
+      "learning_rate": 0.002119043349086666,
+      "loss": 1.1389,
+      "step": 4740
+    },
+    {
+      "epoch": 0.3646642565956465,
+      "learning_rate": 0.0021187131741726766,
+      "loss": 1.0876,
+      "step": 4741
+    },
+    {
+      "epoch": 0.36474117375586496,
+      "learning_rate": 0.0021183829631313805,
+      "loss": 0.8338,
+      "step": 4742
+    },
+    {
+      "epoch": 0.36481809091608336,
+      "learning_rate": 0.0021180527159820584,
+      "loss": 1.3758,
+      "step": 4743
+    },
+    {
+      "epoch": 0.3648950080763018,
+      "learning_rate": 0.002117722432743994,
+      "loss": 1.0835,
+      "step": 4744
+    },
+    {
+      "epoch": 0.3649719252365203,
+      "learning_rate": 0.002117392113436473,
+      "loss": 0.727,
+      "step": 4745
+    },
+    {
+      "epoch": 0.36504884239673874,
+      "learning_rate": 0.002117061758078783,
+      "loss": 1.5522,
+      "step": 4746
+    },
+    {
+      "epoch": 0.36512575955695714,
+      "learning_rate": 0.0021167313666902144,
+      "loss": 1.0084,
+      "step": 4747
+    },
+    {
+      "epoch": 0.3652026767171756,
+      "learning_rate": 0.002116400939290058,
+      "loss": 0.9914,
+      "step": 4748
+    },
+    {
+      "epoch": 0.36527959387739406,
+      "learning_rate": 0.002116070475897609,
+      "loss": 1.3308,
+      "step": 4749
+    },
+    {
+      "epoch": 0.36535651103761246,
+      "learning_rate": 0.0021157399765321626,
+      "loss": 1.1763,
+      "step": 4750
+    },
+    {
+      "epoch": 0.3654334281978309,
+      "learning_rate": 0.0021154094412130166,
+      "loss": 1.0322,
+      "step": 4751
+    },
+    {
+      "epoch": 0.3655103453580494,
+      "learning_rate": 0.002115078869959473,
+      "loss": 1.1166,
+      "step": 4752
+    },
+    {
+      "epoch": 0.36558726251826784,
+      "learning_rate": 0.002114748262790832,
+      "loss": 1.0376,
+      "step": 4753
+    },
+    {
+      "epoch": 0.36566417967848625,
+      "learning_rate": 0.0021144176197264004,
+      "loss": 0.983,
+      "step": 4754
+    },
+    {
+      "epoch": 0.3657410968387047,
+      "learning_rate": 0.0021140869407854828,
+      "loss": 1.2964,
+      "step": 4755
+    },
+    {
+      "epoch": 0.36581801399892316,
+      "learning_rate": 0.002113756225987389,
+      "loss": 1.2203,
+      "step": 4756
+    },
+    {
+      "epoch": 0.3658949311591416,
+      "learning_rate": 0.0021134254753514296,
+      "loss": 1.4343,
+      "step": 4757
+    },
+    {
+      "epoch": 0.36597184831936,
+      "learning_rate": 0.0021130946888969176,
+      "loss": 1.0308,
+      "step": 4758
+    },
+    {
+      "epoch": 0.3660487654795785,
+      "learning_rate": 0.002112763866643167,
+      "loss": 1.3455,
+      "step": 4759
+    },
+    {
+      "epoch": 0.36612568263979695,
+      "learning_rate": 0.0021124330086094967,
+      "loss": 1.2272,
+      "step": 4760
+    },
+    {
+      "epoch": 0.3662025998000154,
+      "learning_rate": 0.002112102114815224,
+      "loss": 1.3347,
+      "step": 4761
+    },
+    {
+      "epoch": 0.3662795169602338,
+      "learning_rate": 0.002111771185279671,
+      "loss": 1.0937,
+      "step": 4762
+    },
+    {
+      "epoch": 0.36635643412045227,
+      "learning_rate": 0.0021114402200221615,
+      "loss": 1.0203,
+      "step": 4763
+    },
+    {
+      "epoch": 0.3664333512806707,
+      "learning_rate": 0.0021111092190620193,
+      "loss": 1.4053,
+      "step": 4764
+    },
+    {
+      "epoch": 0.3665102684408892,
+      "learning_rate": 0.002110778182418574,
+      "loss": 1.3736,
+      "step": 4765
+    },
+    {
+      "epoch": 0.3665871856011076,
+      "learning_rate": 0.0021104471101111534,
+      "loss": 0.801,
+      "step": 4766
+    },
+    {
+      "epoch": 0.36666410276132605,
+      "learning_rate": 0.00211011600215909,
+      "loss": 1.0116,
+      "step": 4767
+    },
+    {
+      "epoch": 0.3667410199215445,
+      "learning_rate": 0.0021097848585817174,
+      "loss": 0.9977,
+      "step": 4768
+    },
+    {
+      "epoch": 0.36681793708176297,
+      "learning_rate": 0.0021094536793983716,
+      "loss": 1.1082,
+      "step": 4769
+    },
+    {
+      "epoch": 0.36689485424198137,
+      "learning_rate": 0.00210912246462839,
+      "loss": 1.3029,
+      "step": 4770
+    },
+    {
+      "epoch": 0.36697177140219983,
+      "learning_rate": 0.002108791214291113,
+      "loss": 1.1388,
+      "step": 4771
+    },
+    {
+      "epoch": 0.3670486885624183,
+      "learning_rate": 0.0021084599284058826,
+      "loss": 1.2999,
+      "step": 4772
+    },
+    {
+      "epoch": 0.3671256057226367,
+      "learning_rate": 0.002108128606992043,
+      "loss": 1.3385,
+      "step": 4773
+    },
+    {
+      "epoch": 0.36720252288285515,
+      "learning_rate": 0.00210779725006894,
+      "loss": 1.3433,
+      "step": 4774
+    },
+    {
+      "epoch": 0.3672794400430736,
+      "learning_rate": 0.0021074658576559217,
+      "loss": 0.8713,
+      "step": 4775
+    },
+    {
+      "epoch": 0.36735635720329207,
+      "learning_rate": 0.0021071344297723398,
+      "loss": 1.1533,
+      "step": 4776
+    },
+    {
+      "epoch": 0.3674332743635105,
+      "learning_rate": 0.0021068029664375455,
+      "loss": 1.1066,
+      "step": 4777
+    },
+    {
+      "epoch": 0.36751019152372894,
+      "learning_rate": 0.0021064714676708933,
+      "loss": 0.8034,
+      "step": 4778
+    },
+    {
+      "epoch": 0.3675871086839474,
+      "learning_rate": 0.0021061399334917404,
+      "loss": 1.0343,
+      "step": 4779
+    },
+    {
+      "epoch": 0.36766402584416585,
+      "learning_rate": 0.002105808363919445,
+      "loss": 1.1599,
+      "step": 4780
+    },
+    {
+      "epoch": 0.36774094300438426,
+      "learning_rate": 0.002105476758973368,
+      "loss": 1.3468,
+      "step": 4781
+    },
+    {
+      "epoch": 0.3678178601646027,
+      "learning_rate": 0.0021051451186728715,
+      "loss": 1.0666,
+      "step": 4782
+    },
+    {
+      "epoch": 0.3678947773248212,
+      "learning_rate": 0.002104813443037321,
+      "loss": 1.1095,
+      "step": 4783
+    },
+    {
+      "epoch": 0.36797169448503964,
+      "learning_rate": 0.0021044817320860837,
+      "loss": 1.1295,
+      "step": 4784
+    },
+    {
+      "epoch": 0.36804861164525804,
+      "learning_rate": 0.0021041499858385276,
+      "loss": 1.027,
+      "step": 4785
+    },
+    {
+      "epoch": 0.3681255288054765,
+      "learning_rate": 0.0021038182043140244,
+      "loss": 0.867,
+      "step": 4786
+    },
+    {
+      "epoch": 0.36820244596569496,
+      "learning_rate": 0.002103486387531947,
+      "loss": 1.0818,
+      "step": 4787
+    },
+    {
+      "epoch": 0.3682793631259134,
+      "learning_rate": 0.00210315453551167,
+      "loss": 1.0079,
+      "step": 4788
+    },
+    {
+      "epoch": 0.3683562802861318,
+      "learning_rate": 0.002102822648272572,
+      "loss": 1.2499,
+      "step": 4789
+    },
+    {
+      "epoch": 0.3684331974463503,
+      "learning_rate": 0.0021024907258340307,
+      "loss": 1.4547,
+      "step": 4790
+    },
+    {
+      "epoch": 0.36851011460656874,
+      "learning_rate": 0.0021021587682154286,
+      "loss": 0.6778,
+      "step": 4791
+    },
+    {
+      "epoch": 0.36858703176678714,
+      "learning_rate": 0.002101826775436148,
+      "loss": 0.8478,
+      "step": 4792
+    },
+    {
+      "epoch": 0.3686639489270056,
+      "learning_rate": 0.002101494747515575,
+      "loss": 1.2081,
+      "step": 4793
+    },
+    {
+      "epoch": 0.36874086608722406,
+      "learning_rate": 0.0021011626844730964,
+      "loss": 1.1856,
+      "step": 4794
+    },
+    {
+      "epoch": 0.3688177832474425,
+      "learning_rate": 0.002100830586328102,
+      "loss": 1.2232,
+      "step": 4795
+    },
+    {
+      "epoch": 0.3688947004076609,
+      "learning_rate": 0.0021004984530999843,
+      "loss": 1.1324,
+      "step": 4796
+    },
+    {
+      "epoch": 0.3689716175678794,
+      "learning_rate": 0.0021001662848081355,
+      "loss": 1.1762,
+      "step": 4797
+    },
+    {
+      "epoch": 0.36904853472809784,
+      "learning_rate": 0.002099834081471952,
+      "loss": 1.0931,
+      "step": 4798
+    },
+    {
+      "epoch": 0.3691254518883163,
+      "learning_rate": 0.0020995018431108313,
+      "loss": 1.0817,
+      "step": 4799
+    },
+    {
+      "epoch": 0.3692023690485347,
+      "learning_rate": 0.0020991695697441734,
+      "loss": 1.2891,
+      "step": 4800
+    },
+    {
+      "epoch": 0.36927928620875317,
+      "learning_rate": 0.0020988372613913796,
+      "loss": 1.1474,
+      "step": 4801
+    },
+    {
+      "epoch": 0.3693562033689716,
+      "learning_rate": 0.0020985049180718544,
+      "loss": 1.2557,
+      "step": 4802
+    },
+    {
+      "epoch": 0.3694331205291901,
+      "learning_rate": 0.002098172539805003,
+      "loss": 1.2381,
+      "step": 4803
+    },
+    {
+      "epoch": 0.3695100376894085,
+      "learning_rate": 0.0020978401266102337,
+      "loss": 1.2017,
+      "step": 4804
+    },
+    {
+      "epoch": 0.36958695484962695,
+      "learning_rate": 0.002097507678506956,
+      "loss": 1.2147,
+      "step": 4805
+    },
+    {
+      "epoch": 0.3696638720098454,
+      "learning_rate": 0.002097175195514583,
+      "loss": 1.2062,
+      "step": 4806
+    },
+    {
+      "epoch": 0.36974078917006387,
+      "learning_rate": 0.002096842677652527,
+      "loss": 1.3095,
+      "step": 4807
+    },
+    {
+      "epoch": 0.36981770633028227,
+      "learning_rate": 0.002096510124940206,
+      "loss": 1.2723,
+      "step": 4808
+    },
+    {
+      "epoch": 0.36989462349050073,
+      "learning_rate": 0.0020961775373970364,
+      "loss": 0.704,
+      "step": 4809
+    },
+    {
+      "epoch": 0.3699715406507192,
+      "learning_rate": 0.0020958449150424393,
+      "loss": 1.3423,
+      "step": 4810
+    },
+    {
+      "epoch": 0.37004845781093765,
+      "learning_rate": 0.002095512257895837,
+      "loss": 0.9298,
+      "step": 4811
+    },
+    {
+      "epoch": 0.37012537497115605,
+      "learning_rate": 0.002095179565976653,
+      "loss": 1.1521,
+      "step": 4812
+    },
+    {
+      "epoch": 0.3702022921313745,
+      "learning_rate": 0.002094846839304314,
+      "loss": 1.0047,
+      "step": 4813
+    },
+    {
+      "epoch": 0.37027920929159297,
+      "learning_rate": 0.0020945140778982478,
+      "loss": 0.7626,
+      "step": 4814
+    },
+    {
+      "epoch": 0.3703561264518114,
+      "learning_rate": 0.0020941812817778854,
+      "loss": 1.0414,
+      "step": 4815
+    },
+    {
+      "epoch": 0.37043304361202983,
+      "learning_rate": 0.0020938484509626584,
+      "loss": 1.0078,
+      "step": 4816
+    },
+    {
+      "epoch": 0.3705099607722483,
+      "learning_rate": 0.0020935155854720014,
+      "loss": 1.2156,
+      "step": 4817
+    },
+    {
+      "epoch": 0.37058687793246675,
+      "learning_rate": 0.0020931826853253514,
+      "loss": 1.2674,
+      "step": 4818
+    },
+    {
+      "epoch": 0.37066379509268516,
+      "learning_rate": 0.0020928497505421457,
+      "loss": 1.275,
+      "step": 4819
+    },
+    {
+      "epoch": 0.3707407122529036,
+      "learning_rate": 0.0020925167811418257,
+      "loss": 1.058,
+      "step": 4820
+    },
+    {
+      "epoch": 0.3708176294131221,
+      "learning_rate": 0.0020921837771438324,
+      "loss": 1.2275,
+      "step": 4821
+    },
+    {
+      "epoch": 0.37089454657334053,
+      "learning_rate": 0.0020918507385676117,
+      "loss": 1.0988,
+      "step": 4822
+    },
+    {
+      "epoch": 0.37097146373355894,
+      "learning_rate": 0.00209151766543261,
+      "loss": 1.2692,
+      "step": 4823
+    },
+    {
+      "epoch": 0.3710483808937774,
+      "learning_rate": 0.0020911845577582747,
+      "loss": 1.3823,
+      "step": 4824
+    },
+    {
+      "epoch": 0.37112529805399586,
+      "learning_rate": 0.002090851415564057,
+      "loss": 1.0632,
+      "step": 4825
+    },
+    {
+      "epoch": 0.3712022152142143,
+      "learning_rate": 0.00209051823886941,
+      "loss": 0.8388,
+      "step": 4826
+    },
+    {
+      "epoch": 0.3712791323744327,
+      "learning_rate": 0.0020901850276937874,
+      "loss": 0.7805,
+      "step": 4827
+    },
+    {
+      "epoch": 0.3713560495346512,
+      "learning_rate": 0.0020898517820566458,
+      "loss": 1.2459,
+      "step": 4828
+    },
+    {
+      "epoch": 0.37143296669486964,
+      "learning_rate": 0.002089518501977444,
+      "loss": 0.8033,
+      "step": 4829
+    },
+    {
+      "epoch": 0.3715098838550881,
+      "learning_rate": 0.002089185187475642,
+      "loss": 1.2512,
+      "step": 4830
+    },
+    {
+      "epoch": 0.3715868010153065,
+      "learning_rate": 0.0020888518385707034,
+      "loss": 1.0718,
+      "step": 4831
+    },
+    {
+      "epoch": 0.37166371817552496,
+      "learning_rate": 0.002088518455282092,
+      "loss": 1.4192,
+      "step": 4832
+    },
+    {
+      "epoch": 0.3717406353357434,
+      "learning_rate": 0.002088185037629275,
+      "loss": 0.953,
+      "step": 4833
+    },
+    {
+      "epoch": 0.3718175524959618,
+      "learning_rate": 0.00208785158563172,
+      "loss": 0.9038,
+      "step": 4834
+    },
+    {
+      "epoch": 0.3718944696561803,
+      "learning_rate": 0.002087518099308899,
+      "loss": 0.8231,
+      "step": 4835
+    },
+    {
+      "epoch": 0.37197138681639874,
+      "learning_rate": 0.0020871845786802836,
+      "loss": 1.0362,
+      "step": 4836
+    },
+    {
+      "epoch": 0.3720483039766172,
+      "learning_rate": 0.0020868510237653484,
+      "loss": 0.929,
+      "step": 4837
+    },
+    {
+      "epoch": 0.3721252211368356,
+      "learning_rate": 0.002086517434583571,
+      "loss": 1.0692,
+      "step": 4838
+    },
+    {
+      "epoch": 0.37220213829705406,
+      "learning_rate": 0.0020861838111544285,
+      "loss": 1.5116,
+      "step": 4839
+    },
+    {
+      "epoch": 0.3722790554572725,
+      "learning_rate": 0.002085850153497403,
+      "loss": 1.0373,
+      "step": 4840
+    },
+    {
+      "epoch": 0.372355972617491,
+      "learning_rate": 0.002085516461631976,
+      "loss": 1.1705,
+      "step": 4841
+    },
+    {
+      "epoch": 0.3724328897777094,
+      "learning_rate": 0.0020851827355776336,
+      "loss": 1.1292,
+      "step": 4842
+    },
+    {
+      "epoch": 0.37250980693792785,
+      "learning_rate": 0.0020848489753538604,
+      "loss": 0.9182,
+      "step": 4843
+    },
+    {
+      "epoch": 0.3725867240981463,
+      "learning_rate": 0.002084515180980146,
+      "loss": 1.3056,
+      "step": 4844
+    },
+    {
+      "epoch": 0.37266364125836476,
+      "learning_rate": 0.002084181352475981,
+      "loss": 1.1036,
+      "step": 4845
+    },
+    {
+      "epoch": 0.37274055841858317,
+      "learning_rate": 0.002083847489860859,
+      "loss": 1.0138,
+      "step": 4846
+    },
+    {
+      "epoch": 0.3728174755788016,
+      "learning_rate": 0.0020835135931542726,
+      "loss": 1.2371,
+      "step": 4847
+    },
+    {
+      "epoch": 0.3728943927390201,
+      "learning_rate": 0.00208317966237572,
+      "loss": 1.3178,
+      "step": 4848
+    },
+    {
+      "epoch": 0.37297130989923855,
+      "learning_rate": 0.0020828456975446985,
+      "loss": 1.6001,
+      "step": 4849
+    },
+    {
+      "epoch": 0.37304822705945695,
+      "learning_rate": 0.0020825116986807096,
+      "loss": 0.8634,
+      "step": 4850
+    },
+    {
+      "epoch": 0.3731251442196754,
+      "learning_rate": 0.002082177665803256,
+      "loss": 0.7771,
+      "step": 4851
+    },
+    {
+      "epoch": 0.37320206137989387,
+      "learning_rate": 0.0020818435989318406,
+      "loss": 1.1008,
+      "step": 4852
+    },
+    {
+      "epoch": 0.37327897854011227,
+      "learning_rate": 0.0020815094980859723,
+      "loss": 1.4486,
+      "step": 4853
+    },
+    {
+      "epoch": 0.37335589570033073,
+      "learning_rate": 0.002081175363285158,
+      "loss": 1.0082,
+      "step": 4854
+    },
+    {
+      "epoch": 0.3734328128605492,
+      "learning_rate": 0.0020808411945489086,
+      "loss": 1.3112,
+      "step": 4855
+    },
+    {
+      "epoch": 0.37350973002076765,
+      "learning_rate": 0.002080506991896736,
+      "loss": 0.7854,
+      "step": 4856
+    },
+    {
+      "epoch": 0.37358664718098605,
+      "learning_rate": 0.002080172755348156,
+      "loss": 1.2396,
+      "step": 4857
+    },
+    {
+      "epoch": 0.3736635643412045,
+      "learning_rate": 0.0020798384849226833,
+      "loss": 0.8316,
+      "step": 4858
+    },
+    {
+      "epoch": 0.37374048150142297,
+      "learning_rate": 0.0020795041806398375,
+      "loss": 0.9986,
+      "step": 4859
+    },
+    {
+      "epoch": 0.37381739866164143,
+      "learning_rate": 0.0020791698425191383,
+      "loss": 1.3407,
+      "step": 4860
+    },
+    {
+      "epoch": 0.37389431582185984,
+      "learning_rate": 0.002078835470580109,
+      "loss": 1.4374,
+      "step": 4861
+    },
+    {
+      "epoch": 0.3739712329820783,
+      "learning_rate": 0.0020785010648422734,
+      "loss": 1.0916,
+      "step": 4862
+    },
+    {
+      "epoch": 0.37404815014229675,
+      "learning_rate": 0.0020781666253251573,
+      "loss": 1.3573,
+      "step": 4863
+    },
+    {
+      "epoch": 0.3741250673025152,
+      "learning_rate": 0.0020778321520482893,
+      "loss": 1.072,
+      "step": 4864
+    },
+    {
+      "epoch": 0.3742019844627336,
+      "learning_rate": 0.0020774976450312,
+      "loss": 1.1448,
+      "step": 4865
+    },
+    {
+      "epoch": 0.3742789016229521,
+      "learning_rate": 0.0020771631042934215,
+      "loss": 1.396,
+      "step": 4866
+    },
+    {
+      "epoch": 0.37435581878317054,
+      "learning_rate": 0.0020768285298544873,
+      "loss": 1.3147,
+      "step": 4867
+    },
+    {
+      "epoch": 0.374432735943389,
+      "learning_rate": 0.002076493921733935,
+      "loss": 1.1093,
+      "step": 4868
+    },
+    {
+      "epoch": 0.3745096531036074,
+      "learning_rate": 0.002076159279951301,
+      "loss": 1.1783,
+      "step": 4869
+    },
+    {
+      "epoch": 0.37458657026382586,
+      "learning_rate": 0.0020758246045261266,
+      "loss": 0.8997,
+      "step": 4870
+    },
+    {
+      "epoch": 0.3746634874240443,
+      "learning_rate": 0.002075489895477953,
+      "loss": 1.4723,
+      "step": 4871
+    },
+    {
+      "epoch": 0.3747404045842628,
+      "learning_rate": 0.002075155152826325,
+      "loss": 1.376,
+      "step": 4872
+    },
+    {
+      "epoch": 0.3748173217444812,
+      "learning_rate": 0.002074820376590788,
+      "loss": 1.0331,
+      "step": 4873
+    },
+    {
+      "epoch": 0.37489423890469964,
+      "learning_rate": 0.0020744855667908903,
+      "loss": 0.9642,
+      "step": 4874
+    },
+    {
+      "epoch": 0.3749711560649181,
+      "learning_rate": 0.0020741507234461813,
+      "loss": 0.8382,
+      "step": 4875
+    },
+    {
+      "epoch": 0.3750480732251365,
+      "learning_rate": 0.002073815846576214,
+      "loss": 1.3105,
+      "step": 4876
+    },
+    {
+      "epoch": 0.37512499038535496,
+      "learning_rate": 0.0020734809362005406,
+      "loss": 1.418,
+      "step": 4877
+    },
+    {
+      "epoch": 0.3752019075455734,
+      "learning_rate": 0.002073145992338718,
+      "loss": 1.347,
+      "step": 4878
+    },
+    {
+      "epoch": 0.3752788247057919,
+      "learning_rate": 0.0020728110150103033,
+      "loss": 1.0774,
+      "step": 4879
+    },
+    {
+      "epoch": 0.3753557418660103,
+      "learning_rate": 0.002072476004234857,
+      "loss": 1.0817,
+      "step": 4880
+    },
+    {
+      "epoch": 0.37543265902622874,
+      "learning_rate": 0.0020721409600319403,
+      "loss": 0.6908,
+      "step": 4881
+    },
+    {
+      "epoch": 0.3755095761864472,
+      "learning_rate": 0.0020718058824211157,
+      "loss": 1.1856,
+      "step": 4882
+    },
+    {
+      "epoch": 0.37558649334666566,
+      "learning_rate": 0.0020714707714219504,
+      "loss": 1.329,
+      "step": 4883
+    },
+    {
+      "epoch": 0.37566341050688407,
+      "learning_rate": 0.0020711356270540116,
+      "loss": 1.0417,
+      "step": 4884
+    },
+    {
+      "epoch": 0.3757403276671025,
+      "learning_rate": 0.002070800449336867,
+      "loss": 1.2722,
+      "step": 4885
+    },
+    {
+      "epoch": 0.375817244827321,
+      "learning_rate": 0.0020704652382900904,
+      "loss": 1.1854,
+      "step": 4886
+    },
+    {
+      "epoch": 0.37589416198753944,
+      "learning_rate": 0.002070129993933254,
+      "loss": 1.0445,
+      "step": 4887
+    },
+    {
+      "epoch": 0.37597107914775785,
+      "learning_rate": 0.0020697947162859327,
+      "loss": 1.4291,
+      "step": 4888
+    },
+    {
+      "epoch": 0.3760479963079763,
+      "learning_rate": 0.002069459405367704,
+      "loss": 1.4924,
+      "step": 4889
+    },
+    {
+      "epoch": 0.37612491346819477,
+      "learning_rate": 0.002069124061198147,
+      "loss": 1.0078,
+      "step": 4890
+    },
+    {
+      "epoch": 0.3762018306284132,
+      "learning_rate": 0.0020687886837968436,
+      "loss": 1.2222,
+      "step": 4891
+    },
+    {
+      "epoch": 0.37627874778863163,
+      "learning_rate": 0.0020684532731833753,
+      "loss": 1.2222,
+      "step": 4892
+    },
+    {
+      "epoch": 0.3763556649488501,
+      "learning_rate": 0.002068117829377329,
+      "loss": 1.1769,
+      "step": 4893
+    },
+    {
+      "epoch": 0.37643258210906855,
+      "learning_rate": 0.00206778235239829,
+      "loss": 1.2531,
+      "step": 4894
+    },
+    {
+      "epoch": 0.37650949926928695,
+      "learning_rate": 0.0020674468422658475,
+      "loss": 1.5151,
+      "step": 4895
+    },
+    {
+      "epoch": 0.3765864164295054,
+      "learning_rate": 0.002067111298999593,
+      "loss": 1.4172,
+      "step": 4896
+    },
+    {
+      "epoch": 0.37666333358972387,
+      "learning_rate": 0.0020667757226191185,
+      "loss": 1.1472,
+      "step": 4897
+    },
+    {
+      "epoch": 0.37674025074994233,
+      "learning_rate": 0.0020664401131440186,
+      "loss": 1.3908,
+      "step": 4898
+    },
+    {
+      "epoch": 0.37681716791016073,
+      "learning_rate": 0.0020661044705938907,
+      "loss": 0.8493,
+      "step": 4899
+    },
+    {
+      "epoch": 0.3768940850703792,
+      "learning_rate": 0.0020657687949883325,
+      "loss": 1.1832,
+      "step": 4900
+    },
+    {
+      "epoch": 0.37697100223059765,
+      "learning_rate": 0.0020654330863469446,
+      "loss": 1.0361,
+      "step": 4901
+    },
+    {
+      "epoch": 0.3770479193908161,
+      "learning_rate": 0.0020650973446893293,
+      "loss": 1.26,
+      "step": 4902
+    },
+    {
+      "epoch": 0.3771248365510345,
+      "learning_rate": 0.0020647615700350916,
+      "loss": 0.8732,
+      "step": 4903
+    },
+    {
+      "epoch": 0.377201753711253,
+      "learning_rate": 0.002064425762403837,
+      "loss": 0.8881,
+      "step": 4904
+    },
+    {
+      "epoch": 0.37727867087147143,
+      "learning_rate": 0.002064089921815174,
+      "loss": 1.1377,
+      "step": 4905
+    },
+    {
+      "epoch": 0.3773555880316899,
+      "learning_rate": 0.0020637540482887125,
+      "loss": 1.2488,
+      "step": 4906
+    },
+    {
+      "epoch": 0.3774325051919083,
+      "learning_rate": 0.002063418141844065,
+      "loss": 1.2511,
+      "step": 4907
+    },
+    {
+      "epoch": 0.37750942235212676,
+      "learning_rate": 0.0020630822025008446,
+      "loss": 1.0231,
+      "step": 4908
+    },
+    {
+      "epoch": 0.3775863395123452,
+      "learning_rate": 0.002062746230278667,
+      "loss": 0.8397,
+      "step": 4909
+    },
+    {
+      "epoch": 0.3776632566725637,
+      "learning_rate": 0.002062410225197152,
+      "loss": 1.0901,
+      "step": 4910
+    },
+    {
+      "epoch": 0.3777401738327821,
+      "learning_rate": 0.0020620741872759167,
+      "loss": 1.2004,
+      "step": 4911
+    },
+    {
+      "epoch": 0.37781709099300054,
+      "learning_rate": 0.0020617381165345845,
+      "loss": 1.1119,
+      "step": 4912
+    },
+    {
+      "epoch": 0.377894008153219,
+      "learning_rate": 0.0020614020129927774,
+      "loss": 1.1782,
+      "step": 4913
+    },
+    {
+      "epoch": 0.3779709253134374,
+      "learning_rate": 0.0020610658766701223,
+      "loss": 1.2043,
+      "step": 4914
+    },
+    {
+      "epoch": 0.37804784247365586,
+      "learning_rate": 0.002060729707586246,
+      "loss": 1.001,
+      "step": 4915
+    },
+    {
+      "epoch": 0.3781247596338743,
+      "learning_rate": 0.002060393505760777,
+      "loss": 1.0096,
+      "step": 4916
+    },
+    {
+      "epoch": 0.3782016767940928,
+      "learning_rate": 0.002060057271213348,
+      "loss": 1.0906,
+      "step": 4917
+    },
+    {
+      "epoch": 0.3782785939543112,
+      "learning_rate": 0.0020597210039635908,
+      "loss": 1.4346,
+      "step": 4918
+    },
+    {
+      "epoch": 0.37835551111452964,
+      "learning_rate": 0.002059384704031141,
+      "loss": 1.2484,
+      "step": 4919
+    },
+    {
+      "epoch": 0.3784324282747481,
+      "learning_rate": 0.0020590483714356352,
+      "loss": 0.989,
+      "step": 4920
+    },
+    {
+      "epoch": 0.37850934543496656,
+      "learning_rate": 0.0020587120061967127,
+      "loss": 1.3396,
+      "step": 4921
+    },
+    {
+      "epoch": 0.37858626259518496,
+      "learning_rate": 0.0020583756083340137,
+      "loss": 1.2484,
+      "step": 4922
+    },
+    {
+      "epoch": 0.3786631797554034,
+      "learning_rate": 0.002058039177867181,
+      "loss": 1.3944,
+      "step": 4923
+    },
+    {
+      "epoch": 0.3787400969156219,
+      "learning_rate": 0.002057702714815859,
+      "loss": 1.0276,
+      "step": 4924
+    },
+    {
+      "epoch": 0.37881701407584034,
+      "learning_rate": 0.002057366219199694,
+      "loss": 1.53,
+      "step": 4925
+    },
+    {
+      "epoch": 0.37889393123605875,
+      "learning_rate": 0.002057029691038335,
+      "loss": 0.8265,
+      "step": 4926
+    },
+    {
+      "epoch": 0.3789708483962772,
+      "learning_rate": 0.002056693130351432,
+      "loss": 1.056,
+      "step": 4927
+    },
+    {
+      "epoch": 0.37904776555649566,
+      "learning_rate": 0.0020563565371586367,
+      "loss": 1.2299,
+      "step": 4928
+    },
+    {
+      "epoch": 0.3791246827167141,
+      "learning_rate": 0.0020560199114796032,
+      "loss": 0.9996,
+      "step": 4929
+    },
+    {
+      "epoch": 0.3792015998769325,
+      "learning_rate": 0.0020556832533339883,
+      "loss": 1.2088,
+      "step": 4930
+    },
+    {
+      "epoch": 0.379278517037151,
+      "learning_rate": 0.002055346562741448,
+      "loss": 0.9918,
+      "step": 4931
+    },
+    {
+      "epoch": 0.37935543419736945,
+      "learning_rate": 0.0020550098397216442,
+      "loss": 1.0546,
+      "step": 4932
+    },
+    {
+      "epoch": 0.3794323513575879,
+      "learning_rate": 0.002054673084294237,
+      "loss": 1.1932,
+      "step": 4933
+    },
+    {
+      "epoch": 0.3795092685178063,
+      "learning_rate": 0.0020543362964788904,
+      "loss": 1.0308,
+      "step": 4934
+    },
+    {
+      "epoch": 0.37958618567802477,
+      "learning_rate": 0.002053999476295269,
+      "loss": 1.1372,
+      "step": 4935
+    },
+    {
+      "epoch": 0.3796631028382432,
+      "learning_rate": 0.002053662623763042,
+      "loss": 1.0867,
+      "step": 4936
+    },
+    {
+      "epoch": 0.37974001999846163,
+      "learning_rate": 0.0020533257389018767,
+      "loss": 0.9925,
+      "step": 4937
+    },
+    {
+      "epoch": 0.3798169371586801,
+      "learning_rate": 0.002052988821731445,
+      "loss": 1.2389,
+      "step": 4938
+    },
+    {
+      "epoch": 0.37989385431889855,
+      "learning_rate": 0.0020526518722714195,
+      "loss": 1.3574,
+      "step": 4939
+    },
+    {
+      "epoch": 0.379970771479117,
+      "learning_rate": 0.002052314890541475,
+      "loss": 1.0758,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3800476886393354,
+      "learning_rate": 0.0020519778765612897,
+      "loss": 1.0655,
+      "step": 4941
+    },
+    {
+      "epoch": 0.38012460579955387,
+      "learning_rate": 0.00205164083035054,
+      "loss": 1.2907,
+      "step": 4942
+    },
+    {
+      "epoch": 0.38020152295977233,
+      "learning_rate": 0.002051303751928907,
+      "loss": 1.4228,
+      "step": 4943
+    },
+    {
+      "epoch": 0.3802784401199908,
+      "learning_rate": 0.0020509666413160736,
+      "loss": 1.1147,
+      "step": 4944
+    },
+    {
+      "epoch": 0.3803553572802092,
+      "learning_rate": 0.0020506294985317246,
+      "loss": 1.4504,
+      "step": 4945
+    },
+    {
+      "epoch": 0.38043227444042765,
+      "learning_rate": 0.0020502923235955445,
+      "loss": 1.242,
+      "step": 4946
+    },
+    {
+      "epoch": 0.3805091916006461,
+      "learning_rate": 0.0020499551165272223,
+      "loss": 0.7567,
+      "step": 4947
+    },
+    {
+      "epoch": 0.38058610876086457,
+      "learning_rate": 0.0020496178773464473,
+      "loss": 0.9129,
+      "step": 4948
+    },
+    {
+      "epoch": 0.380663025921083,
+      "learning_rate": 0.002049280606072912,
+      "loss": 1.1387,
+      "step": 4949
+    },
+    {
+      "epoch": 0.38073994308130144,
+      "learning_rate": 0.00204894330272631,
+      "loss": 1.184,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3808168602415199,
+      "learning_rate": 0.002048605967326336,
+      "loss": 1.222,
+      "step": 4951
+    },
+    {
+      "epoch": 0.38089377740173835,
+      "learning_rate": 0.002048268599892688,
+      "loss": 0.8492,
+      "step": 4952
+    },
+    {
+      "epoch": 0.38097069456195676,
+      "learning_rate": 0.0020479312004450653,
+      "loss": 1.3435,
+      "step": 4953
+    },
+    {
+      "epoch": 0.3810476117221752,
+      "learning_rate": 0.0020475937690031683,
+      "loss": 1.1365,
+      "step": 4954
+    },
+    {
+      "epoch": 0.3811245288823937,
+      "learning_rate": 0.002047256305586701,
+      "loss": 1.2877,
+      "step": 4955
+    },
+    {
+      "epoch": 0.3812014460426121,
+      "learning_rate": 0.0020469188102153675,
+      "loss": 0.9223,
+      "step": 4956
+    },
+    {
+      "epoch": 0.38127836320283054,
+      "learning_rate": 0.0020465812829088743,
+      "loss": 1.0167,
+      "step": 4957
+    },
+    {
+      "epoch": 0.381355280363049,
+      "learning_rate": 0.0020462437236869304,
+      "loss": 1.1759,
+      "step": 4958
+    },
+    {
+      "epoch": 0.38143219752326746,
+      "learning_rate": 0.0020459061325692465,
+      "loss": 1.2252,
+      "step": 4959
+    },
+    {
+      "epoch": 0.38150911468348586,
+      "learning_rate": 0.0020455685095755343,
+      "loss": 0.9734,
+      "step": 4960
+    },
+    {
+      "epoch": 0.3815860318437043,
+      "learning_rate": 0.0020452308547255086,
+      "loss": 1.2894,
+      "step": 4961
+    },
+    {
+      "epoch": 0.3816629490039228,
+      "learning_rate": 0.0020448931680388843,
+      "loss": 1.0756,
+      "step": 4962
+    },
+    {
+      "epoch": 0.38173986616414124,
+      "learning_rate": 0.002044555449535381,
+      "loss": 0.8025,
+      "step": 4963
+    },
+    {
+      "epoch": 0.38181678332435964,
+      "learning_rate": 0.002044217699234717,
+      "loss": 1.1754,
+      "step": 4964
+    },
+    {
+      "epoch": 0.3818937004845781,
+      "learning_rate": 0.0020438799171566147,
+      "loss": 1.1561,
+      "step": 4965
+    },
+    {
+      "epoch": 0.38197061764479656,
+      "learning_rate": 0.002043542103320797,
+      "loss": 1.0023,
+      "step": 4966
+    },
+    {
+      "epoch": 0.382047534805015,
+      "learning_rate": 0.002043204257746989,
+      "loss": 1.3553,
+      "step": 4967
+    },
+    {
+      "epoch": 0.3821244519652334,
+      "learning_rate": 0.002042866380454919,
+      "loss": 1.31,
+      "step": 4968
+    },
+    {
+      "epoch": 0.3822013691254519,
+      "learning_rate": 0.002042528471464315,
+      "loss": 1.6954,
+      "step": 4969
+    },
+    {
+      "epoch": 0.38227828628567034,
+      "learning_rate": 0.002042190530794908,
+      "loss": 1.036,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3823552034458888,
+      "learning_rate": 0.002041852558466431,
+      "loss": 1.1154,
+      "step": 4971
+    },
+    {
+      "epoch": 0.3824321206061072,
+      "learning_rate": 0.002041514554498619,
+      "loss": 1.462,
+      "step": 4972
+    },
+    {
+      "epoch": 0.38250903776632567,
+      "learning_rate": 0.0020411765189112066,
+      "loss": 1.0796,
+      "step": 4973
+    },
+    {
+      "epoch": 0.3825859549265441,
+      "learning_rate": 0.002040838451723934,
+      "loss": 1.0786,
+      "step": 4974
+    },
+    {
+      "epoch": 0.3826628720867626,
+      "learning_rate": 0.00204050035295654,
+      "loss": 1.1372,
+      "step": 4975
+    },
+    {
+      "epoch": 0.382739789246981,
+      "learning_rate": 0.0020401622226287677,
+      "loss": 0.897,
+      "step": 4976
+    },
+    {
+      "epoch": 0.38281670640719945,
+      "learning_rate": 0.00203982406076036,
+      "loss": 1.1475,
+      "step": 4977
+    },
+    {
+      "epoch": 0.3828936235674179,
+      "learning_rate": 0.002039485867371063,
+      "loss": 0.601,
+      "step": 4978
+    },
+    {
+      "epoch": 0.3829705407276363,
+      "learning_rate": 0.002039147642480624,
+      "loss": 0.9801,
+      "step": 4979
+    },
+    {
+      "epoch": 0.38304745788785477,
+      "learning_rate": 0.002038809386108792,
+      "loss": 1.0897,
+      "step": 4980
+    },
+    {
+      "epoch": 0.38312437504807323,
+      "learning_rate": 0.0020384710982753185,
+      "loss": 0.9211,
+      "step": 4981
+    },
+    {
+      "epoch": 0.3832012922082917,
+      "learning_rate": 0.002038132778999956,
+      "loss": 1.1034,
+      "step": 4982
+    },
+    {
+      "epoch": 0.3832782093685101,
+      "learning_rate": 0.0020377944283024607,
+      "loss": 0.9202,
+      "step": 4983
+    },
+    {
+      "epoch": 0.38335512652872855,
+      "learning_rate": 0.0020374560462025873,
+      "loss": 1.1294,
+      "step": 4984
+    },
+    {
+      "epoch": 0.383432043688947,
+      "learning_rate": 0.0020371176327200957,
+      "loss": 1.0268,
+      "step": 4985
+    },
+    {
+      "epoch": 0.38350896084916547,
+      "learning_rate": 0.0020367791878747456,
+      "loss": 1.0809,
+      "step": 4986
+    },
+    {
+      "epoch": 0.3835858780093839,
+      "learning_rate": 0.0020364407116862998,
+      "loss": 1.0792,
+      "step": 4987
+    },
+    {
+      "epoch": 0.38366279516960233,
+      "learning_rate": 0.002036102204174521,
+      "loss": 1.3723,
+      "step": 4988
+    },
+    {
+      "epoch": 0.3837397123298208,
+      "learning_rate": 0.002035763665359176,
+      "loss": 1.2951,
+      "step": 4989
+    },
+    {
+      "epoch": 0.38381662949003925,
+      "learning_rate": 0.0020354250952600325,
+      "loss": 1.113,
+      "step": 4990
+    },
+    {
+      "epoch": 0.38389354665025766,
+      "learning_rate": 0.0020350864938968594,
+      "loss": 1.2646,
+      "step": 4991
+    },
+    {
+      "epoch": 0.3839704638104761,
+      "learning_rate": 0.0020347478612894286,
+      "loss": 1.4198,
+      "step": 4992
+    },
+    {
+      "epoch": 0.3840473809706946,
+      "learning_rate": 0.002034409197457512,
+      "loss": 1.0156,
+      "step": 4993
+    },
+    {
+      "epoch": 0.38412429813091303,
+      "learning_rate": 0.0020340705024208864,
+      "loss": 0.9134,
+      "step": 4994
+    },
+    {
+      "epoch": 0.38420121529113144,
+      "learning_rate": 0.002033731776199328,
+      "loss": 1.1205,
+      "step": 4995
+    },
+    {
+      "epoch": 0.3842781324513499,
+      "learning_rate": 0.002033393018812614,
+      "loss": 1.2747,
+      "step": 4996
+    },
+    {
+      "epoch": 0.38435504961156836,
+      "learning_rate": 0.0020330542302805254,
+      "loss": 1.4218,
+      "step": 4997
+    },
+    {
+      "epoch": 0.38443196677178676,
+      "learning_rate": 0.0020327154106228457,
+      "loss": 1.263,
+      "step": 4998
+    },
+    {
+      "epoch": 0.3845088839320052,
+      "learning_rate": 0.0020323765598593575,
+      "loss": 1.347,
+      "step": 4999
+    },
+    {
+      "epoch": 0.3845858010922237,
+      "learning_rate": 0.0020320376780098475,
+      "loss": 1.3136,
+      "step": 5000
+    },
+    {
+      "epoch": 0.38466271825244214,
+      "learning_rate": 0.0020316987650941024,
+      "loss": 1.0214,
+      "step": 5001
+    },
+    {
+      "epoch": 0.38473963541266054,
+      "learning_rate": 0.002031359821131913,
+      "loss": 1.0561,
+      "step": 5002
+    },
+    {
+      "epoch": 0.384816552572879,
+      "learning_rate": 0.0020310208461430694,
+      "loss": 1.1358,
+      "step": 5003
+    },
+    {
+      "epoch": 0.38489346973309746,
+      "learning_rate": 0.0020306818401473648,
+      "loss": 0.9896,
+      "step": 5004
+    },
+    {
+      "epoch": 0.3849703868933159,
+      "learning_rate": 0.0020303428031645952,
+      "loss": 0.966,
+      "step": 5005
+    },
+    {
+      "epoch": 0.3850473040535343,
+      "learning_rate": 0.0020300037352145566,
+      "loss": 1.275,
+      "step": 5006
+    },
+    {
+      "epoch": 0.3851242212137528,
+      "learning_rate": 0.0020296646363170473,
+      "loss": 1.2206,
+      "step": 5007
+    },
+    {
+      "epoch": 0.38520113837397124,
+      "learning_rate": 0.0020293255064918678,
+      "loss": 1.3341,
+      "step": 5008
+    },
+    {
+      "epoch": 0.3852780555341897,
+      "learning_rate": 0.0020289863457588204,
+      "loss": 1.0249,
+      "step": 5009
+    },
+    {
+      "epoch": 0.3853549726944081,
+      "learning_rate": 0.002028647154137709,
+      "loss": 0.8251,
+      "step": 5010
+    },
+    {
+      "epoch": 0.38543188985462656,
+      "learning_rate": 0.0020283079316483392,
+      "loss": 1.453,
+      "step": 5011
+    },
+    {
+      "epoch": 0.385508807014845,
+      "learning_rate": 0.0020279686783105186,
+      "loss": 1.4294,
+      "step": 5012
+    },
+    {
+      "epoch": 0.3855857241750635,
+      "learning_rate": 0.002027629394144057,
+      "loss": 1.3009,
+      "step": 5013
+    },
+    {
+      "epoch": 0.3856626413352819,
+      "learning_rate": 0.0020272900791687655,
+      "loss": 1.0109,
+      "step": 5014
+    },
+    {
+      "epoch": 0.38573955849550035,
+      "learning_rate": 0.0020269507334044557,
+      "loss": 1.0238,
+      "step": 5015
+    },
+    {
+      "epoch": 0.3858164756557188,
+      "learning_rate": 0.0020266113568709445,
+      "loss": 1.1444,
+      "step": 5016
+    },
+    {
+      "epoch": 0.3858933928159372,
+      "learning_rate": 0.0020262719495880467,
+      "loss": 1.1523,
+      "step": 5017
+    },
+    {
+      "epoch": 0.38597030997615567,
+      "learning_rate": 0.0020259325115755816,
+      "loss": 1.1241,
+      "step": 5018
+    },
+    {
+      "epoch": 0.3860472271363741,
+      "learning_rate": 0.0020255930428533684,
+      "loss": 1.1907,
+      "step": 5019
+    },
+    {
+      "epoch": 0.3861241442965926,
+      "learning_rate": 0.0020252535434412302,
+      "loss": 1.1154,
+      "step": 5020
+    },
+    {
+      "epoch": 0.386201061456811,
+      "learning_rate": 0.00202491401335899,
+      "loss": 1.5667,
+      "step": 5021
+    },
+    {
+      "epoch": 0.38627797861702945,
+      "learning_rate": 0.0020245744526264744,
+      "loss": 1.1757,
+      "step": 5022
+    },
+    {
+      "epoch": 0.3863548957772479,
+      "learning_rate": 0.0020242348612635086,
+      "loss": 1.249,
+      "step": 5023
+    },
+    {
+      "epoch": 0.38643181293746637,
+      "learning_rate": 0.0020238952392899233,
+      "loss": 1.3065,
+      "step": 5024
+    },
+    {
+      "epoch": 0.38650873009768477,
+      "learning_rate": 0.0020235555867255495,
+      "loss": 1.2946,
+      "step": 5025
+    },
+    {
+      "epoch": 0.38658564725790323,
+      "learning_rate": 0.0020232159035902184,
+      "loss": 1.0711,
+      "step": 5026
+    },
+    {
+      "epoch": 0.3866625644181217,
+      "learning_rate": 0.002022876189903766,
+      "loss": 1.0078,
+      "step": 5027
+    },
+    {
+      "epoch": 0.38673948157834015,
+      "learning_rate": 0.0020225364456860277,
+      "loss": 1.0949,
+      "step": 5028
+    },
+    {
+      "epoch": 0.38681639873855855,
+      "learning_rate": 0.002022196670956842,
+      "loss": 1.0695,
+      "step": 5029
+    },
+    {
+      "epoch": 0.386893315898777,
+      "learning_rate": 0.0020218568657360486,
+      "loss": 1.1325,
+      "step": 5030
+    },
+    {
+      "epoch": 0.38697023305899547,
+      "learning_rate": 0.0020215170300434885,
+      "loss": 1.1046,
+      "step": 5031
+    },
+    {
+      "epoch": 0.38704715021921393,
+      "learning_rate": 0.0020211771638990058,
+      "loss": 1.2303,
+      "step": 5032
+    },
+    {
+      "epoch": 0.38712406737943234,
+      "learning_rate": 0.0020208372673224456,
+      "loss": 0.9923,
+      "step": 5033
+    },
+    {
+      "epoch": 0.3872009845396508,
+      "learning_rate": 0.002020497340333654,
+      "loss": 0.8864,
+      "step": 5034
+    },
+    {
+      "epoch": 0.38727790169986925,
+      "learning_rate": 0.0020201573829524803,
+      "loss": 1.0268,
+      "step": 5035
+    },
+    {
+      "epoch": 0.3873548188600877,
+      "learning_rate": 0.0020198173951987753,
+      "loss": 1.0813,
+      "step": 5036
+    },
+    {
+      "epoch": 0.3874317360203061,
+      "learning_rate": 0.0020194773770923907,
+      "loss": 1.0885,
+      "step": 5037
+    },
+    {
+      "epoch": 0.3875086531805246,
+      "learning_rate": 0.0020191373286531805,
+      "loss": 1.3219,
+      "step": 5038
+    },
+    {
+      "epoch": 0.38758557034074304,
+      "learning_rate": 0.0020187972499010006,
+      "loss": 0.8751,
+      "step": 5039
+    },
+    {
+      "epoch": 0.38766248750096144,
+      "learning_rate": 0.0020184571408557095,
+      "loss": 1.1717,
+      "step": 5040
+    },
+    {
+      "epoch": 0.3877394046611799,
+      "learning_rate": 0.0020181170015371643,
+      "loss": 1.1097,
+      "step": 5041
+    },
+    {
+      "epoch": 0.38781632182139836,
+      "learning_rate": 0.0020177768319652283,
+      "loss": 1.1087,
+      "step": 5042
+    },
+    {
+      "epoch": 0.3878932389816168,
+      "learning_rate": 0.0020174366321597625,
+      "loss": 0.9766,
+      "step": 5043
+    },
+    {
+      "epoch": 0.3879701561418352,
+      "learning_rate": 0.0020170964021406336,
+      "loss": 1.1057,
+      "step": 5044
+    },
+    {
+      "epoch": 0.3880470733020537,
+      "learning_rate": 0.0020167561419277063,
+      "loss": 1.1334,
+      "step": 5045
+    },
+    {
+      "epoch": 0.38812399046227214,
+      "learning_rate": 0.0020164158515408497,
+      "loss": 0.9277,
+      "step": 5046
+    },
+    {
+      "epoch": 0.3882009076224906,
+      "learning_rate": 0.002016075530999933,
+      "loss": 1.266,
+      "step": 5047
+    },
+    {
+      "epoch": 0.388277824782709,
+      "learning_rate": 0.0020157351803248283,
+      "loss": 0.7329,
+      "step": 5048
+    },
+    {
+      "epoch": 0.38835474194292746,
+      "learning_rate": 0.002015394799535409,
+      "loss": 1.0342,
+      "step": 5049
+    },
+    {
+      "epoch": 0.3884316591031459,
+      "learning_rate": 0.00201505438865155,
+      "loss": 1.0407,
+      "step": 5050
+    },
+    {
+      "epoch": 0.3885085762633644,
+      "learning_rate": 0.0020147139476931284,
+      "loss": 0.8504,
+      "step": 5051
+    },
+    {
+      "epoch": 0.3885854934235828,
+      "learning_rate": 0.0020143734766800234,
+      "loss": 1.3714,
+      "step": 5052
+    },
+    {
+      "epoch": 0.38866241058380124,
+      "learning_rate": 0.002014032975632115,
+      "loss": 1.4464,
+      "step": 5053
+    },
+    {
+      "epoch": 0.3887393277440197,
+      "learning_rate": 0.002013692444569285,
+      "loss": 1.2104,
+      "step": 5054
+    },
+    {
+      "epoch": 0.38881624490423816,
+      "learning_rate": 0.002013351883511418,
+      "loss": 1.2731,
+      "step": 5055
+    },
+    {
+      "epoch": 0.38889316206445657,
+      "learning_rate": 0.0020130112924784,
+      "loss": 0.982,
+      "step": 5056
+    },
+    {
+      "epoch": 0.388970079224675,
+      "learning_rate": 0.002012670671490117,
+      "loss": 1.4885,
+      "step": 5057
+    },
+    {
+      "epoch": 0.3890469963848935,
+      "learning_rate": 0.00201233002056646,
+      "loss": 1.1467,
+      "step": 5058
+    },
+    {
+      "epoch": 0.3891239135451119,
+      "learning_rate": 0.002011989339727319,
+      "loss": 0.7181,
+      "step": 5059
+    },
+    {
+      "epoch": 0.38920083070533035,
+      "learning_rate": 0.0020116486289925867,
+      "loss": 1.0285,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3892777478655488,
+      "learning_rate": 0.0020113078883821577,
+      "loss": 1.1083,
+      "step": 5061
+    },
+    {
+      "epoch": 0.38935466502576727,
+      "learning_rate": 0.0020109671179159285,
+      "loss": 1.218,
+      "step": 5062
+    },
+    {
+      "epoch": 0.38943158218598567,
+      "learning_rate": 0.0020106263176137965,
+      "loss": 1.3801,
+      "step": 5063
+    },
+    {
+      "epoch": 0.38950849934620413,
+      "learning_rate": 0.002010285487495662,
+      "loss": 0.8797,
+      "step": 5064
+    },
+    {
+      "epoch": 0.3895854165064226,
+      "learning_rate": 0.0020099446275814257,
+      "loss": 1.1166,
+      "step": 5065
+    },
+    {
+      "epoch": 0.38966233366664105,
+      "learning_rate": 0.0020096037378909914,
+      "loss": 1.5569,
+      "step": 5066
+    },
+    {
+      "epoch": 0.38973925082685945,
+      "learning_rate": 0.0020092628184442637,
+      "loss": 1.3894,
+      "step": 5067
+    },
+    {
+      "epoch": 0.3898161679870779,
+      "learning_rate": 0.002008921869261149,
+      "loss": 1.2409,
+      "step": 5068
+    },
+    {
+      "epoch": 0.38989308514729637,
+      "learning_rate": 0.0020085808903615564,
+      "loss": 1.0281,
+      "step": 5069
+    },
+    {
+      "epoch": 0.38997000230751483,
+      "learning_rate": 0.002008239881765395,
+      "loss": 1.0496,
+      "step": 5070
+    },
+    {
+      "epoch": 0.39004691946773323,
+      "learning_rate": 0.002007898843492578,
+      "loss": 1.1025,
+      "step": 5071
+    },
+    {
+      "epoch": 0.3901238366279517,
+      "learning_rate": 0.0020075577755630175,
+      "loss": 1.0,
+      "step": 5072
+    },
+    {
+      "epoch": 0.39020075378817015,
+      "learning_rate": 0.00200721667799663,
+      "loss": 1.55,
+      "step": 5073
+    },
+    {
+      "epoch": 0.3902776709483886,
+      "learning_rate": 0.0020068755508133316,
+      "loss": 1.0485,
+      "step": 5074
+    },
+    {
+      "epoch": 0.390354588108607,
+      "learning_rate": 0.0020065343940330422,
+      "loss": 1.0761,
+      "step": 5075
+    },
+    {
+      "epoch": 0.3904315052688255,
+      "learning_rate": 0.002006193207675681,
+      "loss": 1.1752,
+      "step": 5076
+    },
+    {
+      "epoch": 0.39050842242904393,
+      "learning_rate": 0.0020058519917611714,
+      "loss": 1.0596,
+      "step": 5077
+    },
+    {
+      "epoch": 0.39058533958926234,
+      "learning_rate": 0.002005510746309437,
+      "loss": 1.1141,
+      "step": 5078
+    },
+    {
+      "epoch": 0.3906622567494808,
+      "learning_rate": 0.0020051694713404023,
+      "loss": 0.8887,
+      "step": 5079
+    },
+    {
+      "epoch": 0.39073917390969926,
+      "learning_rate": 0.002004828166873996,
+      "loss": 1.158,
+      "step": 5080
+    },
+    {
+      "epoch": 0.3908160910699177,
+      "learning_rate": 0.002004486832930147,
+      "loss": 1.6662,
+      "step": 5081
+    },
+    {
+      "epoch": 0.3908930082301361,
+      "learning_rate": 0.0020041454695287867,
+      "loss": 1.2052,
+      "step": 5082
+    },
+    {
+      "epoch": 0.3909699253903546,
+      "learning_rate": 0.0020038040766898464,
+      "loss": 1.0947,
+      "step": 5083
+    },
+    {
+      "epoch": 0.39104684255057304,
+      "learning_rate": 0.002003462654433261,
+      "loss": 1.421,
+      "step": 5084
+    },
+    {
+      "epoch": 0.3911237597107915,
+      "learning_rate": 0.0020031212027789664,
+      "loss": 1.3049,
+      "step": 5085
+    },
+    {
+      "epoch": 0.3912006768710099,
+      "learning_rate": 0.002002779721746901,
+      "loss": 0.9061,
+      "step": 5086
+    },
+    {
+      "epoch": 0.39127759403122836,
+      "learning_rate": 0.002002438211357003,
+      "loss": 0.6874,
+      "step": 5087
+    },
+    {
+      "epoch": 0.3913545111914468,
+      "learning_rate": 0.002002096671629215,
+      "loss": 1.3376,
+      "step": 5088
+    },
+    {
+      "epoch": 0.3914314283516653,
+      "learning_rate": 0.0020017551025834786,
+      "loss": 1.5181,
+      "step": 5089
+    },
+    {
+      "epoch": 0.3915083455118837,
+      "learning_rate": 0.0020014135042397386,
+      "loss": 1.1969,
+      "step": 5090
+    },
+    {
+      "epoch": 0.39158526267210214,
+      "learning_rate": 0.002001071876617942,
+      "loss": 1.165,
+      "step": 5091
+    },
+    {
+      "epoch": 0.3916621798323206,
+      "learning_rate": 0.002000730219738036,
+      "loss": 0.9387,
+      "step": 5092
+    },
+    {
+      "epoch": 0.39173909699253906,
+      "learning_rate": 0.0020003885336199708,
+      "loss": 1.0339,
+      "step": 5093
+    },
+    {
+      "epoch": 0.39181601415275746,
+      "learning_rate": 0.0020000468182836974,
+      "loss": 0.985,
+      "step": 5094
+    },
+    {
+      "epoch": 0.3918929313129759,
+      "learning_rate": 0.0019997050737491696,
+      "loss": 1.1823,
+      "step": 5095
+    },
+    {
+      "epoch": 0.3919698484731944,
+      "learning_rate": 0.0019993633000363414,
+      "loss": 1.2862,
+      "step": 5096
+    },
+    {
+      "epoch": 0.39204676563341284,
+      "learning_rate": 0.00199902149716517,
+      "loss": 0.7466,
+      "step": 5097
+    },
+    {
+      "epoch": 0.39212368279363125,
+      "learning_rate": 0.0019986796651556126,
+      "loss": 1.0475,
+      "step": 5098
+    },
+    {
+      "epoch": 0.3922005999538497,
+      "learning_rate": 0.0019983378040276307,
+      "loss": 1.203,
+      "step": 5099
+    },
+    {
+      "epoch": 0.39227751711406816,
+      "learning_rate": 0.0019979959138011847,
+      "loss": 1.3668,
+      "step": 5100
+    },
+    {
+      "epoch": 0.39235443427428657,
+      "learning_rate": 0.001997653994496238,
+      "loss": 1.1902,
+      "step": 5101
+    },
+    {
+      "epoch": 0.392431351434505,
+      "learning_rate": 0.0019973120461327567,
+      "loss": 1.1621,
+      "step": 5102
+    },
+    {
+      "epoch": 0.3925082685947235,
+      "learning_rate": 0.0019969700687307054,
+      "loss": 1.4672,
+      "step": 5103
+    },
+    {
+      "epoch": 0.39258518575494195,
+      "learning_rate": 0.001996628062310055,
+      "loss": 1.1216,
+      "step": 5104
+    },
+    {
+      "epoch": 0.39266210291516035,
+      "learning_rate": 0.0019962860268907742,
+      "loss": 1.1515,
+      "step": 5105
+    },
+    {
+      "epoch": 0.3927390200753788,
+      "learning_rate": 0.001995943962492835,
+      "loss": 1.3502,
+      "step": 5106
+    },
+    {
+      "epoch": 0.39281593723559727,
+      "learning_rate": 0.0019956018691362114,
+      "loss": 1.1658,
+      "step": 5107
+    },
+    {
+      "epoch": 0.3928928543958157,
+      "learning_rate": 0.0019952597468408774,
+      "loss": 1.0952,
+      "step": 5108
+    },
+    {
+      "epoch": 0.39296977155603413,
+      "learning_rate": 0.001994917595626811,
+      "loss": 1.0106,
+      "step": 5109
+    },
+    {
+      "epoch": 0.3930466887162526,
+      "learning_rate": 0.0019945754155139906,
+      "loss": 1.2692,
+      "step": 5110
+    },
+    {
+      "epoch": 0.39312360587647105,
+      "learning_rate": 0.001994233206522396,
+      "loss": 0.8599,
+      "step": 5111
+    },
+    {
+      "epoch": 0.3932005230366895,
+      "learning_rate": 0.0019938909686720097,
+      "loss": 1.0223,
+      "step": 5112
+    },
+    {
+      "epoch": 0.3932774401969079,
+      "learning_rate": 0.0019935487019828147,
+      "loss": 0.9486,
+      "step": 5113
+    },
+    {
+      "epoch": 0.39335435735712637,
+      "learning_rate": 0.0019932064064747965,
+      "loss": 1.1028,
+      "step": 5114
+    },
+    {
+      "epoch": 0.39343127451734483,
+      "learning_rate": 0.0019928640821679426,
+      "loss": 1.1744,
+      "step": 5115
+    },
+    {
+      "epoch": 0.3935081916775633,
+      "learning_rate": 0.0019925217290822405,
+      "loss": 1.1037,
+      "step": 5116
+    },
+    {
+      "epoch": 0.3935851088377817,
+      "learning_rate": 0.0019921793472376824,
+      "loss": 0.9918,
+      "step": 5117
+    },
+    {
+      "epoch": 0.39366202599800015,
+      "learning_rate": 0.001991836936654259,
+      "loss": 1.2017,
+      "step": 5118
+    },
+    {
+      "epoch": 0.3937389431582186,
+      "learning_rate": 0.0019914944973519638,
+      "loss": 1.2374,
+      "step": 5119
+    },
+    {
+      "epoch": 0.393815860318437,
+      "learning_rate": 0.001991152029350793,
+      "loss": 1.2245,
+      "step": 5120
+    },
+    {
+      "epoch": 0.3938927774786555,
+      "learning_rate": 0.0019908095326707435,
+      "loss": 0.9647,
+      "step": 5121
+    },
+    {
+      "epoch": 0.39396969463887394,
+      "learning_rate": 0.0019904670073318136,
+      "loss": 1.2698,
+      "step": 5122
+    },
+    {
+      "epoch": 0.3940466117990924,
+      "learning_rate": 0.001990124453354004,
+      "loss": 0.8463,
+      "step": 5123
+    },
+    {
+      "epoch": 0.3941235289593108,
+      "learning_rate": 0.001989781870757317,
+      "loss": 1.1296,
+      "step": 5124
+    },
+    {
+      "epoch": 0.39420044611952926,
+      "learning_rate": 0.001989439259561756,
+      "loss": 1.0608,
+      "step": 5125
+    },
+    {
+      "epoch": 0.3942773632797477,
+      "learning_rate": 0.001989096619787326,
+      "loss": 1.0291,
+      "step": 5126
+    },
+    {
+      "epoch": 0.3943542804399662,
+      "learning_rate": 0.0019887539514540356,
+      "loss": 0.788,
+      "step": 5127
+    },
+    {
+      "epoch": 0.3944311976001846,
+      "learning_rate": 0.0019884112545818927,
+      "loss": 0.974,
+      "step": 5128
+    },
+    {
+      "epoch": 0.39450811476040304,
+      "learning_rate": 0.0019880685291909068,
+      "loss": 1.3288,
+      "step": 5129
+    },
+    {
+      "epoch": 0.3945850319206215,
+      "learning_rate": 0.001987725775301091,
+      "loss": 1.181,
+      "step": 5130
+    },
+    {
+      "epoch": 0.39466194908083996,
+      "learning_rate": 0.001987382992932459,
+      "loss": 1.2275,
+      "step": 5131
+    },
+    {
+      "epoch": 0.39473886624105836,
+      "learning_rate": 0.001987040182105026,
+      "loss": 1.5413,
+      "step": 5132
+    },
+    {
+      "epoch": 0.3948157834012768,
+      "learning_rate": 0.00198669734283881,
+      "loss": 1.2808,
+      "step": 5133
+    },
+    {
+      "epoch": 0.3948927005614953,
+      "learning_rate": 0.001986354475153828,
+      "loss": 0.9555,
+      "step": 5134
+    },
+    {
+      "epoch": 0.39496961772171374,
+      "learning_rate": 0.001986011579070101,
+      "loss": 0.8047,
+      "step": 5135
+    },
+    {
+      "epoch": 0.39504653488193214,
+      "learning_rate": 0.0019856686546076523,
+      "loss": 1.4716,
+      "step": 5136
+    },
+    {
+      "epoch": 0.3951234520421506,
+      "learning_rate": 0.001985325701786505,
+      "loss": 0.8409,
+      "step": 5137
+    },
+    {
+      "epoch": 0.39520036920236906,
+      "learning_rate": 0.001984982720626683,
+      "loss": 0.8102,
+      "step": 5138
+    },
+    {
+      "epoch": 0.39527728636258747,
+      "learning_rate": 0.0019846397111482152,
+      "loss": 1.1077,
+      "step": 5139
+    },
+    {
+      "epoch": 0.3953542035228059,
+      "learning_rate": 0.0019842966733711297,
+      "loss": 1.0789,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3954311206830244,
+      "learning_rate": 0.0019839536073154563,
+      "loss": 0.9913,
+      "step": 5141
+    },
+    {
+      "epoch": 0.39550803784324284,
+      "learning_rate": 0.0019836105130012277,
+      "loss": 1.097,
+      "step": 5142
+    },
+    {
+      "epoch": 0.39558495500346125,
+      "learning_rate": 0.001983267390448477,
+      "loss": 1.2007,
+      "step": 5143
+    },
+    {
+      "epoch": 0.3956618721636797,
+      "learning_rate": 0.00198292423967724,
+      "loss": 0.8776,
+      "step": 5144
+    },
+    {
+      "epoch": 0.39573878932389817,
+      "learning_rate": 0.001982581060707553,
+      "loss": 1.526,
+      "step": 5145
+    },
+    {
+      "epoch": 0.3958157064841166,
+      "learning_rate": 0.0019822378535594553,
+      "loss": 0.8039,
+      "step": 5146
+    },
+    {
+      "epoch": 0.39589262364433503,
+      "learning_rate": 0.0019818946182529867,
+      "loss": 0.8459,
+      "step": 5147
+    },
+    {
+      "epoch": 0.3959695408045535,
+      "learning_rate": 0.001981551354808189,
+      "loss": 1.148,
+      "step": 5148
+    },
+    {
+      "epoch": 0.39604645796477195,
+      "learning_rate": 0.0019812080632451053,
+      "loss": 1.0183,
+      "step": 5149
+    },
+    {
+      "epoch": 0.3961233751249904,
+      "learning_rate": 0.0019808647435837824,
+      "loss": 1.1046,
+      "step": 5150
+    },
+    {
+      "epoch": 0.3962002922852088,
+      "learning_rate": 0.001980521395844265,
+      "loss": 0.8268,
+      "step": 5151
+    },
+    {
+      "epoch": 0.39627720944542727,
+      "learning_rate": 0.0019801780200466032,
+      "loss": 0.7725,
+      "step": 5152
+    },
+    {
+      "epoch": 0.39635412660564573,
+      "learning_rate": 0.001979834616210846,
+      "loss": 1.1379,
+      "step": 5153
+    },
+    {
+      "epoch": 0.3964310437658642,
+      "learning_rate": 0.0019794911843570456,
+      "loss": 1.2747,
+      "step": 5154
+    },
+    {
+      "epoch": 0.3965079609260826,
+      "learning_rate": 0.0019791477245052554,
+      "loss": 1.1457,
+      "step": 5155
+    },
+    {
+      "epoch": 0.39658487808630105,
+      "learning_rate": 0.0019788042366755297,
+      "loss": 1.4812,
+      "step": 5156
+    },
+    {
+      "epoch": 0.3966617952465195,
+      "learning_rate": 0.001978460720887926,
+      "loss": 1.0438,
+      "step": 5157
+    },
+    {
+      "epoch": 0.39673871240673797,
+      "learning_rate": 0.001978117177162502,
+      "loss": 1.4135,
+      "step": 5158
+    },
+    {
+      "epoch": 0.3968156295669564,
+      "learning_rate": 0.0019777736055193182,
+      "loss": 1.7645,
+      "step": 5159
+    },
+    {
+      "epoch": 0.39689254672717483,
+      "learning_rate": 0.001977430005978435,
+      "loss": 1.4837,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3969694638873933,
+      "learning_rate": 0.0019770863785599168,
+      "loss": 1.1554,
+      "step": 5161
+    },
+    {
+      "epoch": 0.3970463810476117,
+      "learning_rate": 0.001976742723283827,
+      "loss": 1.0866,
+      "step": 5162
+    },
+    {
+      "epoch": 0.39712329820783016,
+      "learning_rate": 0.0019763990401702337,
+      "loss": 1.3217,
+      "step": 5163
+    },
+    {
+      "epoch": 0.3972002153680486,
+      "learning_rate": 0.001976055329239203,
+      "loss": 1.1724,
+      "step": 5164
+    },
+    {
+      "epoch": 0.3972771325282671,
+      "learning_rate": 0.0019757115905108063,
+      "loss": 0.9498,
+      "step": 5165
+    },
+    {
+      "epoch": 0.3973540496884855,
+      "learning_rate": 0.0019753678240051137,
+      "loss": 1.0203,
+      "step": 5166
+    },
+    {
+      "epoch": 0.39743096684870394,
+      "learning_rate": 0.001975024029742198,
+      "loss": 1.16,
+      "step": 5167
+    },
+    {
+      "epoch": 0.3975078840089224,
+      "learning_rate": 0.0019746802077421346,
+      "loss": 1.2596,
+      "step": 5168
+    },
+    {
+      "epoch": 0.39758480116914086,
+      "learning_rate": 0.001974336358024999,
+      "loss": 1.1585,
+      "step": 5169
+    },
+    {
+      "epoch": 0.39766171832935926,
+      "learning_rate": 0.0019739924806108696,
+      "loss": 1.3912,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3977386354895777,
+      "learning_rate": 0.0019736485755198247,
+      "loss": 1.162,
+      "step": 5171
+    },
+    {
+      "epoch": 0.3978155526497962,
+      "learning_rate": 0.0019733046427719463,
+      "loss": 0.9619,
+      "step": 5172
+    },
+    {
+      "epoch": 0.39789246981001464,
+      "learning_rate": 0.001972960682387316,
+      "loss": 1.134,
+      "step": 5173
+    },
+    {
+      "epoch": 0.39796938697023304,
+      "learning_rate": 0.001972616694386019,
+      "loss": 1.0416,
+      "step": 5174
+    },
+    {
+      "epoch": 0.3980463041304515,
+      "learning_rate": 0.0019722726787881405,
+      "loss": 1.3365,
+      "step": 5175
+    },
+    {
+      "epoch": 0.39812322129066996,
+      "learning_rate": 0.001971928635613768,
+      "loss": 1.395,
+      "step": 5176
+    },
+    {
+      "epoch": 0.3982001384508884,
+      "learning_rate": 0.001971584564882991,
+      "loss": 1.0601,
+      "step": 5177
+    },
+    {
+      "epoch": 0.3982770556111068,
+      "learning_rate": 0.0019712404666158995,
+      "loss": 0.9681,
+      "step": 5178
+    },
+    {
+      "epoch": 0.3983539727713253,
+      "learning_rate": 0.001970896340832586,
+      "loss": 1.0862,
+      "step": 5179
+    },
+    {
+      "epoch": 0.39843088993154374,
+      "learning_rate": 0.0019705521875531444,
+      "loss": 1.1627,
+      "step": 5180
+    },
+    {
+      "epoch": 0.39850780709176215,
+      "learning_rate": 0.0019702080067976703,
+      "loss": 1.2503,
+      "step": 5181
+    },
+    {
+      "epoch": 0.3985847242519806,
+      "learning_rate": 0.001969863798586261,
+      "loss": 1.2246,
+      "step": 5182
+    },
+    {
+      "epoch": 0.39866164141219906,
+      "learning_rate": 0.0019695195629390143,
+      "loss": 1.2009,
+      "step": 5183
+    },
+    {
+      "epoch": 0.3987385585724175,
+      "learning_rate": 0.0019691752998760308,
+      "loss": 1.3125,
+      "step": 5184
+    },
+    {
+      "epoch": 0.3988154757326359,
+      "learning_rate": 0.0019688310094174133,
+      "loss": 0.9761,
+      "step": 5185
+    },
+    {
+      "epoch": 0.3988923928928544,
+      "learning_rate": 0.001968486691583264,
+      "loss": 0.8742,
+      "step": 5186
+    },
+    {
+      "epoch": 0.39896931005307285,
+      "learning_rate": 0.001968142346393689,
+      "loss": 1.2871,
+      "step": 5187
+    },
+    {
+      "epoch": 0.3990462272132913,
+      "learning_rate": 0.0019677979738687936,
+      "loss": 1.4258,
+      "step": 5188
+    },
+    {
+      "epoch": 0.3991231443735097,
+      "learning_rate": 0.001967453574028688,
+      "loss": 1.2961,
+      "step": 5189
+    },
+    {
+      "epoch": 0.39920006153372817,
+      "learning_rate": 0.0019671091468934803,
+      "loss": 0.9601,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3992769786939466,
+      "learning_rate": 0.001966764692483283,
+      "loss": 1.4248,
+      "step": 5191
+    },
+    {
+      "epoch": 0.3993538958541651,
+      "learning_rate": 0.001966420210818209,
+      "loss": 1.3774,
+      "step": 5192
+    },
+    {
+      "epoch": 0.3994308130143835,
+      "learning_rate": 0.0019660757019183727,
+      "loss": 1.2007,
+      "step": 5193
+    },
+    {
+      "epoch": 0.39950773017460195,
+      "learning_rate": 0.0019657311658038902,
+      "loss": 1.1962,
+      "step": 5194
+    },
+    {
+      "epoch": 0.3995846473348204,
+      "learning_rate": 0.00196538660249488,
+      "loss": 0.8821,
+      "step": 5195
+    },
+    {
+      "epoch": 0.39966156449503887,
+      "learning_rate": 0.0019650420120114606,
+      "loss": 1.1384,
+      "step": 5196
+    },
+    {
+      "epoch": 0.39973848165525727,
+      "learning_rate": 0.0019646973943737538,
+      "loss": 0.992,
+      "step": 5197
+    },
+    {
+      "epoch": 0.39981539881547573,
+      "learning_rate": 0.0019643527496018822,
+      "loss": 1.0595,
+      "step": 5198
+    },
+    {
+      "epoch": 0.3998923159756942,
+      "learning_rate": 0.001964008077715969,
+      "loss": 1.2635,
+      "step": 5199
+    },
+    {
+      "epoch": 0.39996923313591265,
+      "learning_rate": 0.00196366337873614,
+      "loss": 1.101,
+      "step": 5200
+    },
+    {
+      "epoch": 0.40004615029613105,
+      "learning_rate": 0.0019633186526825243,
+      "loss": 0.9079,
+      "step": 5201
+    },
+    {
+      "epoch": 0.4001230674563495,
+      "learning_rate": 0.0019629738995752487,
+      "loss": 1.1907,
+      "step": 5202
+    },
+    {
+      "epoch": 0.40019998461656797,
+      "learning_rate": 0.0019626291194344454,
+      "loss": 1.1628,
+      "step": 5203
+    },
+    {
+      "epoch": 0.4002769017767864,
+      "learning_rate": 0.001962284312280245,
+      "loss": 1.7953,
+      "step": 5204
+    },
+    {
+      "epoch": 0.40035381893700484,
+      "learning_rate": 0.0019619394781327827,
+      "loss": 1.1781,
+      "step": 5205
+    },
+    {
+      "epoch": 0.4004307360972233,
+      "learning_rate": 0.001961594617012192,
+      "loss": 1.2836,
+      "step": 5206
+    },
+    {
+      "epoch": 0.40050765325744175,
+      "learning_rate": 0.001961249728938611,
+      "loss": 1.0594,
+      "step": 5207
+    },
+    {
+      "epoch": 0.40058457041766016,
+      "learning_rate": 0.0019609048139321776,
+      "loss": 0.2582,
+      "step": 5208
+    },
+    {
+      "epoch": 0.4006614875778786,
+      "learning_rate": 0.001960559872013031,
+      "loss": 1.0238,
+      "step": 5209
+    },
+    {
+      "epoch": 0.4007384047380971,
+      "learning_rate": 0.001960214903201314,
+      "loss": 1.1124,
+      "step": 5210
+    },
+    {
+      "epoch": 0.40081532189831554,
+      "learning_rate": 0.0019598699075171694,
+      "loss": 1.0666,
+      "step": 5211
+    },
+    {
+      "epoch": 0.40089223905853394,
+      "learning_rate": 0.0019595248849807417,
+      "loss": 1.2561,
+      "step": 5212
+    },
+    {
+      "epoch": 0.4009691562187524,
+      "learning_rate": 0.001959179835612177,
+      "loss": 1.0237,
+      "step": 5213
+    },
+    {
+      "epoch": 0.40104607337897086,
+      "learning_rate": 0.001958834759431623,
+      "loss": 1.1575,
+      "step": 5214
+    },
+    {
+      "epoch": 0.4011229905391893,
+      "learning_rate": 0.001958489656459229,
+      "loss": 1.2128,
+      "step": 5215
+    },
+    {
+      "epoch": 0.4011999076994077,
+      "learning_rate": 0.001958144526715147,
+      "loss": 1.1695,
+      "step": 5216
+    },
+    {
+      "epoch": 0.4012768248596262,
+      "learning_rate": 0.0019577993702195274,
+      "loss": 1.2952,
+      "step": 5217
+    },
+    {
+      "epoch": 0.40135374201984464,
+      "learning_rate": 0.0019574541869925265,
+      "loss": 1.0493,
+      "step": 5218
+    },
+    {
+      "epoch": 0.4014306591800631,
+      "learning_rate": 0.001957108977054298,
+      "loss": 1.0179,
+      "step": 5219
+    },
+    {
+      "epoch": 0.4015075763402815,
+      "learning_rate": 0.0019567637404250006,
+      "loss": 1.3156,
+      "step": 5220
+    },
+    {
+      "epoch": 0.40158449350049996,
+      "learning_rate": 0.001956418477124792,
+      "loss": 1.2617,
+      "step": 5221
+    },
+    {
+      "epoch": 0.4016614106607184,
+      "learning_rate": 0.0019560731871738333,
+      "loss": 1.7767,
+      "step": 5222
+    },
+    {
+      "epoch": 0.4017383278209368,
+      "learning_rate": 0.0019557278705922853,
+      "loss": 1.0942,
+      "step": 5223
+    },
+    {
+      "epoch": 0.4018152449811553,
+      "learning_rate": 0.0019553825274003125,
+      "loss": 1.2732,
+      "step": 5224
+    },
+    {
+      "epoch": 0.40189216214137374,
+      "learning_rate": 0.001955037157618079,
+      "loss": 1.2963,
+      "step": 5225
+    },
+    {
+      "epoch": 0.4019690793015922,
+      "learning_rate": 0.001954691761265752,
+      "loss": 1.0306,
+      "step": 5226
+    },
+    {
+      "epoch": 0.4020459964618106,
+      "learning_rate": 0.001954346338363499,
+      "loss": 1.1381,
+      "step": 5227
+    },
+    {
+      "epoch": 0.40212291362202907,
+      "learning_rate": 0.0019540008889314896,
+      "loss": 1.1768,
+      "step": 5228
+    },
+    {
+      "epoch": 0.4021998307822475,
+      "learning_rate": 0.001953655412989895,
+      "loss": 0.9864,
+      "step": 5229
+    },
+    {
+      "epoch": 0.402276747942466,
+      "learning_rate": 0.001953309910558888,
+      "loss": 1.0862,
+      "step": 5230
+    },
+    {
+      "epoch": 0.4023536651026844,
+      "learning_rate": 0.0019529643816586437,
+      "loss": 1.0612,
+      "step": 5231
+    },
+    {
+      "epoch": 0.40243058226290285,
+      "learning_rate": 0.0019526188263093366,
+      "loss": 1.0074,
+      "step": 5232
+    },
+    {
+      "epoch": 0.4025074994231213,
+      "learning_rate": 0.0019522732445311445,
+      "loss": 1.1565,
+      "step": 5233
+    },
+    {
+      "epoch": 0.40258441658333977,
+      "learning_rate": 0.0019519276363442464,
+      "loss": 0.9626,
+      "step": 5234
+    },
+    {
+      "epoch": 0.40266133374355817,
+      "learning_rate": 0.0019515820017688226,
+      "loss": 1.3013,
+      "step": 5235
+    },
+    {
+      "epoch": 0.40273825090377663,
+      "learning_rate": 0.001951236340825055,
+      "loss": 1.0675,
+      "step": 5236
+    },
+    {
+      "epoch": 0.4028151680639951,
+      "learning_rate": 0.0019508906535331272,
+      "loss": 0.8322,
+      "step": 5237
+    },
+    {
+      "epoch": 0.40289208522421355,
+      "learning_rate": 0.0019505449399132245,
+      "loss": 1.354,
+      "step": 5238
+    },
+    {
+      "epoch": 0.40296900238443195,
+      "learning_rate": 0.001950199199985533,
+      "loss": 0.9497,
+      "step": 5239
+    },
+    {
+      "epoch": 0.4030459195446504,
+      "learning_rate": 0.001949853433770241,
+      "loss": 1.0087,
+      "step": 5240
+    },
+    {
+      "epoch": 0.40312283670486887,
+      "learning_rate": 0.001949507641287538,
+      "loss": 1.4448,
+      "step": 5241
+    },
+    {
+      "epoch": 0.4031997538650873,
+      "learning_rate": 0.0019491618225576158,
+      "loss": 1.2243,
+      "step": 5242
+    },
+    {
+      "epoch": 0.40327667102530573,
+      "learning_rate": 0.0019488159776006664,
+      "loss": 1.516,
+      "step": 5243
+    },
+    {
+      "epoch": 0.4033535881855242,
+      "learning_rate": 0.0019484701064368842,
+      "loss": 1.4581,
+      "step": 5244
+    },
+    {
+      "epoch": 0.40343050534574265,
+      "learning_rate": 0.0019481242090864655,
+      "loss": 1.1713,
+      "step": 5245
+    },
+    {
+      "epoch": 0.40350742250596106,
+      "learning_rate": 0.0019477782855696069,
+      "loss": 1.663,
+      "step": 5246
+    },
+    {
+      "epoch": 0.4035843396661795,
+      "learning_rate": 0.0019474323359065077,
+      "loss": 1.2683,
+      "step": 5247
+    },
+    {
+      "epoch": 0.403661256826398,
+      "learning_rate": 0.0019470863601173678,
+      "loss": 1.7624,
+      "step": 5248
+    },
+    {
+      "epoch": 0.40373817398661643,
+      "learning_rate": 0.0019467403582223894,
+      "loss": 0.7888,
+      "step": 5249
+    },
+    {
+      "epoch": 0.40381509114683484,
+      "learning_rate": 0.0019463943302417766,
+      "loss": 1.2357,
+      "step": 5250
+    },
+    {
+      "epoch": 0.4038920083070533,
+      "learning_rate": 0.0019460482761957333,
+      "loss": 1.1859,
+      "step": 5251
+    },
+    {
+      "epoch": 0.40396892546727176,
+      "learning_rate": 0.0019457021961044659,
+      "loss": 0.9087,
+      "step": 5252
+    },
+    {
+      "epoch": 0.4040458426274902,
+      "learning_rate": 0.001945356089988183,
+      "loss": 0.9273,
+      "step": 5253
+    },
+    {
+      "epoch": 0.4041227597877086,
+      "learning_rate": 0.001945009957867094,
+      "loss": 1.2712,
+      "step": 5254
+    },
+    {
+      "epoch": 0.4041996769479271,
+      "learning_rate": 0.0019446637997614093,
+      "loss": 1.2907,
+      "step": 5255
+    },
+    {
+      "epoch": 0.40427659410814554,
+      "learning_rate": 0.0019443176156913424,
+      "loss": 1.1121,
+      "step": 5256
+    },
+    {
+      "epoch": 0.404353511268364,
+      "learning_rate": 0.0019439714056771063,
+      "loss": 1.2598,
+      "step": 5257
+    },
+    {
+      "epoch": 0.4044304284285824,
+      "learning_rate": 0.0019436251697389183,
+      "loss": 1.3612,
+      "step": 5258
+    },
+    {
+      "epoch": 0.40450734558880086,
+      "learning_rate": 0.001943278907896993,
+      "loss": 1.1818,
+      "step": 5259
+    },
+    {
+      "epoch": 0.4045842627490193,
+      "learning_rate": 0.0019429326201715511,
+      "loss": 0.9665,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4046611799092378,
+      "learning_rate": 0.0019425863065828115,
+      "loss": 1.2597,
+      "step": 5261
+    },
+    {
+      "epoch": 0.4047380970694562,
+      "learning_rate": 0.0019422399671509968,
+      "loss": 1.4157,
+      "step": 5262
+    },
+    {
+      "epoch": 0.40481501422967464,
+      "learning_rate": 0.0019418936018963289,
+      "loss": 1.0955,
+      "step": 5263
+    },
+    {
+      "epoch": 0.4048919313898931,
+      "learning_rate": 0.0019415472108390331,
+      "loss": 1.1623,
+      "step": 5264
+    },
+    {
+      "epoch": 0.4049688485501115,
+      "learning_rate": 0.0019412007939993356,
+      "loss": 1.0951,
+      "step": 5265
+    },
+    {
+      "epoch": 0.40504576571032996,
+      "learning_rate": 0.0019408543513974637,
+      "loss": 0.9792,
+      "step": 5266
+    },
+    {
+      "epoch": 0.4051226828705484,
+      "learning_rate": 0.0019405078830536473,
+      "loss": 1.039,
+      "step": 5267
+    },
+    {
+      "epoch": 0.4051996000307669,
+      "learning_rate": 0.0019401613889881162,
+      "loss": 1.0961,
+      "step": 5268
+    },
+    {
+      "epoch": 0.4052765171909853,
+      "learning_rate": 0.0019398148692211033,
+      "loss": 1.0801,
+      "step": 5269
+    },
+    {
+      "epoch": 0.40535343435120375,
+      "learning_rate": 0.001939468323772841,
+      "loss": 0.7251,
+      "step": 5270
+    },
+    {
+      "epoch": 0.4054303515114222,
+      "learning_rate": 0.001939121752663566,
+      "loss": 1.0864,
+      "step": 5271
+    },
+    {
+      "epoch": 0.40550726867164066,
+      "learning_rate": 0.0019387751559135137,
+      "loss": 1.3339,
+      "step": 5272
+    },
+    {
+      "epoch": 0.40558418583185907,
+      "learning_rate": 0.0019384285335429228,
+      "loss": 0.8261,
+      "step": 5273
+    },
+    {
+      "epoch": 0.4056611029920775,
+      "learning_rate": 0.0019380818855720334,
+      "loss": 0.9781,
+      "step": 5274
+    },
+    {
+      "epoch": 0.405738020152296,
+      "learning_rate": 0.0019377352120210861,
+      "loss": 1.5219,
+      "step": 5275
+    },
+    {
+      "epoch": 0.40581493731251445,
+      "learning_rate": 0.0019373885129103231,
+      "loss": 1.1983,
+      "step": 5276
+    },
+    {
+      "epoch": 0.40589185447273285,
+      "learning_rate": 0.0019370417882599892,
+      "loss": 1.3282,
+      "step": 5277
+    },
+    {
+      "epoch": 0.4059687716329513,
+      "learning_rate": 0.0019366950380903297,
+      "loss": 1.3461,
+      "step": 5278
+    },
+    {
+      "epoch": 0.40604568879316977,
+      "learning_rate": 0.0019363482624215917,
+      "loss": 1.0908,
+      "step": 5279
+    },
+    {
+      "epoch": 0.4061226059533882,
+      "learning_rate": 0.0019360014612740245,
+      "loss": 1.3082,
+      "step": 5280
+    },
+    {
+      "epoch": 0.40619952311360663,
+      "learning_rate": 0.0019356546346678773,
+      "loss": 1.2681,
+      "step": 5281
+    },
+    {
+      "epoch": 0.4062764402738251,
+      "learning_rate": 0.001935307782623402,
+      "loss": 0.9953,
+      "step": 5282
+    },
+    {
+      "epoch": 0.40635335743404355,
+      "learning_rate": 0.0019349609051608511,
+      "loss": 1.1353,
+      "step": 5283
+    },
+    {
+      "epoch": 0.40643027459426195,
+      "learning_rate": 0.0019346140023004802,
+      "loss": 1.6298,
+      "step": 5284
+    },
+    {
+      "epoch": 0.4065071917544804,
+      "learning_rate": 0.0019342670740625447,
+      "loss": 1.1867,
+      "step": 5285
+    },
+    {
+      "epoch": 0.40658410891469887,
+      "learning_rate": 0.0019339201204673017,
+      "loss": 0.8309,
+      "step": 5286
+    },
+    {
+      "epoch": 0.40666102607491733,
+      "learning_rate": 0.0019335731415350112,
+      "loss": 1.4024,
+      "step": 5287
+    },
+    {
+      "epoch": 0.40673794323513573,
+      "learning_rate": 0.0019332261372859332,
+      "loss": 1.2121,
+      "step": 5288
+    },
+    {
+      "epoch": 0.4068148603953542,
+      "learning_rate": 0.0019328791077403294,
+      "loss": 1.2537,
+      "step": 5289
+    },
+    {
+      "epoch": 0.40689177755557265,
+      "learning_rate": 0.0019325320529184628,
+      "loss": 1.2006,
+      "step": 5290
+    },
+    {
+      "epoch": 0.4069686947157911,
+      "learning_rate": 0.0019321849728405995,
+      "loss": 1.4749,
+      "step": 5291
+    },
+    {
+      "epoch": 0.4070456118760095,
+      "learning_rate": 0.0019318378675270051,
+      "loss": 1.3887,
+      "step": 5292
+    },
+    {
+      "epoch": 0.407122529036228,
+      "learning_rate": 0.0019314907369979474,
+      "loss": 0.9718,
+      "step": 5293
+    },
+    {
+      "epoch": 0.40719944619644644,
+      "learning_rate": 0.0019311435812736958,
+      "loss": 1.0322,
+      "step": 5294
+    },
+    {
+      "epoch": 0.4072763633566649,
+      "learning_rate": 0.0019307964003745215,
+      "loss": 1.1825,
+      "step": 5295
+    },
+    {
+      "epoch": 0.4073532805168833,
+      "learning_rate": 0.0019304491943206963,
+      "loss": 0.9931,
+      "step": 5296
+    },
+    {
+      "epoch": 0.40743019767710176,
+      "learning_rate": 0.001930101963132494,
+      "loss": 1.2207,
+      "step": 5297
+    },
+    {
+      "epoch": 0.4075071148373202,
+      "learning_rate": 0.0019297547068301895,
+      "loss": 1.8569,
+      "step": 5298
+    },
+    {
+      "epoch": 0.4075840319975387,
+      "learning_rate": 0.00192940742543406,
+      "loss": 1.2515,
+      "step": 5299
+    },
+    {
+      "epoch": 0.4076609491577571,
+      "learning_rate": 0.0019290601189643837,
+      "loss": 1.0231,
+      "step": 5300
+    },
+    {
+      "epoch": 0.40773786631797554,
+      "learning_rate": 0.0019287127874414395,
+      "loss": 1.1391,
+      "step": 5301
+    },
+    {
+      "epoch": 0.407814783478194,
+      "learning_rate": 0.0019283654308855092,
+      "loss": 1.1344,
+      "step": 5302
+    },
+    {
+      "epoch": 0.4078917006384124,
+      "learning_rate": 0.001928018049316875,
+      "loss": 1.3173,
+      "step": 5303
+    },
+    {
+      "epoch": 0.40796861779863086,
+      "learning_rate": 0.0019276706427558206,
+      "loss": 1.2478,
+      "step": 5304
+    },
+    {
+      "epoch": 0.4080455349588493,
+      "learning_rate": 0.0019273232112226315,
+      "loss": 0.9447,
+      "step": 5305
+    },
+    {
+      "epoch": 0.4081224521190678,
+      "learning_rate": 0.0019269757547375952,
+      "loss": 1.4481,
+      "step": 5306
+    },
+    {
+      "epoch": 0.4081993692792862,
+      "learning_rate": 0.0019266282733209994,
+      "loss": 1.2329,
+      "step": 5307
+    },
+    {
+      "epoch": 0.40827628643950464,
+      "learning_rate": 0.001926280766993134,
+      "loss": 1.2065,
+      "step": 5308
+    },
+    {
+      "epoch": 0.4083532035997231,
+      "learning_rate": 0.0019259332357742902,
+      "loss": 1.0585,
+      "step": 5309
+    },
+    {
+      "epoch": 0.40843012075994156,
+      "learning_rate": 0.0019255856796847608,
+      "loss": 0.7993,
+      "step": 5310
+    },
+    {
+      "epoch": 0.40850703792015997,
+      "learning_rate": 0.0019252380987448408,
+      "loss": 1.4017,
+      "step": 5311
+    },
+    {
+      "epoch": 0.4085839550803784,
+      "learning_rate": 0.001924890492974824,
+      "loss": 1.3804,
+      "step": 5312
+    },
+    {
+      "epoch": 0.4086608722405969,
+      "learning_rate": 0.0019245428623950091,
+      "loss": 0.8605,
+      "step": 5313
+    },
+    {
+      "epoch": 0.40873778940081534,
+      "learning_rate": 0.0019241952070256938,
+      "loss": 1.347,
+      "step": 5314
+    },
+    {
+      "epoch": 0.40881470656103375,
+      "learning_rate": 0.001923847526887179,
+      "loss": 1.0736,
+      "step": 5315
+    },
+    {
+      "epoch": 0.4088916237212522,
+      "learning_rate": 0.0019234998219997649,
+      "loss": 1.2824,
+      "step": 5316
+    },
+    {
+      "epoch": 0.40896854088147067,
+      "learning_rate": 0.0019231520923837547,
+      "loss": 0.9933,
+      "step": 5317
+    },
+    {
+      "epoch": 0.4090454580416891,
+      "learning_rate": 0.0019228043380594524,
+      "loss": 1.2015,
+      "step": 5318
+    },
+    {
+      "epoch": 0.40912237520190753,
+      "learning_rate": 0.0019224565590471648,
+      "loss": 0.7587,
+      "step": 5319
+    },
+    {
+      "epoch": 0.409199292362126,
+      "learning_rate": 0.0019221087553671983,
+      "loss": 1.2312,
+      "step": 5320
+    },
+    {
+      "epoch": 0.40927620952234445,
+      "learning_rate": 0.0019217609270398615,
+      "loss": 1.2008,
+      "step": 5321
+    },
+    {
+      "epoch": 0.4093531266825629,
+      "learning_rate": 0.0019214130740854648,
+      "loss": 1.3925,
+      "step": 5322
+    },
+    {
+      "epoch": 0.4094300438427813,
+      "learning_rate": 0.0019210651965243196,
+      "loss": 1.0074,
+      "step": 5323
+    },
+    {
+      "epoch": 0.40950696100299977,
+      "learning_rate": 0.001920717294376739,
+      "loss": 1.5791,
+      "step": 5324
+    },
+    {
+      "epoch": 0.40958387816321823,
+      "learning_rate": 0.0019203693676630364,
+      "loss": 1.162,
+      "step": 5325
+    },
+    {
+      "epoch": 0.40966079532343663,
+      "learning_rate": 0.0019200214164035291,
+      "loss": 1.3833,
+      "step": 5326
+    },
+    {
+      "epoch": 0.4097377124836551,
+      "learning_rate": 0.001919673440618533,
+      "loss": 1.287,
+      "step": 5327
+    },
+    {
+      "epoch": 0.40981462964387355,
+      "learning_rate": 0.0019193254403283675,
+      "loss": 1.3652,
+      "step": 5328
+    },
+    {
+      "epoch": 0.409891546804092,
+      "learning_rate": 0.0019189774155533525,
+      "loss": 0.8447,
+      "step": 5329
+    },
+    {
+      "epoch": 0.4099684639643104,
+      "learning_rate": 0.0019186293663138092,
+      "loss": 1.4372,
+      "step": 5330
+    },
+    {
+      "epoch": 0.4100453811245289,
+      "learning_rate": 0.0019182812926300614,
+      "loss": 1.3139,
+      "step": 5331
+    },
+    {
+      "epoch": 0.41012229828474733,
+      "learning_rate": 0.0019179331945224328,
+      "loss": 0.9102,
+      "step": 5332
+    },
+    {
+      "epoch": 0.4101992154449658,
+      "learning_rate": 0.0019175850720112495,
+      "loss": 1.0867,
+      "step": 5333
+    },
+    {
+      "epoch": 0.4102761326051842,
+      "learning_rate": 0.001917236925116839,
+      "loss": 1.0796,
+      "step": 5334
+    },
+    {
+      "epoch": 0.41035304976540266,
+      "learning_rate": 0.0019168887538595291,
+      "loss": 0.8754,
+      "step": 5335
+    },
+    {
+      "epoch": 0.4104299669256211,
+      "learning_rate": 0.0019165405582596503,
+      "loss": 1.4631,
+      "step": 5336
+    },
+    {
+      "epoch": 0.4105068840858396,
+      "learning_rate": 0.0019161923383375344,
+      "loss": 1.0335,
+      "step": 5337
+    },
+    {
+      "epoch": 0.410583801246058,
+      "learning_rate": 0.0019158440941135144,
+      "loss": 1.2179,
+      "step": 5338
+    },
+    {
+      "epoch": 0.41066071840627644,
+      "learning_rate": 0.0019154958256079242,
+      "loss": 1.4298,
+      "step": 5339
+    },
+    {
+      "epoch": 0.4107376355664949,
+      "learning_rate": 0.0019151475328410994,
+      "loss": 1.6193,
+      "step": 5340
+    },
+    {
+      "epoch": 0.41081455272671336,
+      "learning_rate": 0.001914799215833378,
+      "loss": 1.0342,
+      "step": 5341
+    },
+    {
+      "epoch": 0.41089146988693176,
+      "learning_rate": 0.0019144508746050976,
+      "loss": 1.6808,
+      "step": 5342
+    },
+    {
+      "epoch": 0.4109683870471502,
+      "learning_rate": 0.0019141025091765989,
+      "loss": 1.2041,
+      "step": 5343
+    },
+    {
+      "epoch": 0.4110453042073687,
+      "learning_rate": 0.0019137541195682233,
+      "loss": 0.9751,
+      "step": 5344
+    },
+    {
+      "epoch": 0.4111222213675871,
+      "learning_rate": 0.0019134057058003133,
+      "loss": 1.1871,
+      "step": 5345
+    },
+    {
+      "epoch": 0.41119913852780554,
+      "learning_rate": 0.0019130572678932134,
+      "loss": 0.9748,
+      "step": 5346
+    },
+    {
+      "epoch": 0.411276055688024,
+      "learning_rate": 0.001912708805867269,
+      "loss": 1.3121,
+      "step": 5347
+    },
+    {
+      "epoch": 0.41135297284824246,
+      "learning_rate": 0.0019123603197428272,
+      "loss": 1.0555,
+      "step": 5348
+    },
+    {
+      "epoch": 0.41142989000846086,
+      "learning_rate": 0.001912011809540237,
+      "loss": 1.4181,
+      "step": 5349
+    },
+    {
+      "epoch": 0.4115068071686793,
+      "learning_rate": 0.0019116632752798474,
+      "loss": 0.8591,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4115837243288978,
+      "learning_rate": 0.0019113147169820103,
+      "loss": 1.5652,
+      "step": 5351
+    },
+    {
+      "epoch": 0.41166064148911624,
+      "learning_rate": 0.0019109661346670785,
+      "loss": 1.2278,
+      "step": 5352
+    },
+    {
+      "epoch": 0.41173755864933465,
+      "learning_rate": 0.0019106175283554055,
+      "loss": 1.597,
+      "step": 5353
+    },
+    {
+      "epoch": 0.4118144758095531,
+      "learning_rate": 0.0019102688980673467,
+      "loss": 1.3299,
+      "step": 5354
+    },
+    {
+      "epoch": 0.41189139296977156,
+      "learning_rate": 0.0019099202438232598,
+      "loss": 1.1454,
+      "step": 5355
+    },
+    {
+      "epoch": 0.41196831012999,
+      "learning_rate": 0.0019095715656435028,
+      "loss": 1.018,
+      "step": 5356
+    },
+    {
+      "epoch": 0.4120452272902084,
+      "learning_rate": 0.0019092228635484354,
+      "loss": 1.1463,
+      "step": 5357
+    },
+    {
+      "epoch": 0.4121221444504269,
+      "learning_rate": 0.001908874137558418,
+      "loss": 1.1272,
+      "step": 5358
+    },
+    {
+      "epoch": 0.41219906161064535,
+      "learning_rate": 0.0019085253876938138,
+      "loss": 1.3921,
+      "step": 5359
+    },
+    {
+      "epoch": 0.4122759787708638,
+      "learning_rate": 0.0019081766139749862,
+      "loss": 1.11,
+      "step": 5360
+    },
+    {
+      "epoch": 0.4123528959310822,
+      "learning_rate": 0.0019078278164223013,
+      "loss": 1.4963,
+      "step": 5361
+    },
+    {
+      "epoch": 0.41242981309130067,
+      "learning_rate": 0.001907478995056125,
+      "loss": 1.184,
+      "step": 5362
+    },
+    {
+      "epoch": 0.4125067302515191,
+      "learning_rate": 0.001907130149896825,
+      "loss": 1.1806,
+      "step": 5363
+    },
+    {
+      "epoch": 0.4125836474117376,
+      "learning_rate": 0.0019067812809647716,
+      "loss": 1.0272,
+      "step": 5364
+    },
+    {
+      "epoch": 0.412660564571956,
+      "learning_rate": 0.0019064323882803352,
+      "loss": 1.2647,
+      "step": 5365
+    },
+    {
+      "epoch": 0.41273748173217445,
+      "learning_rate": 0.0019060834718638883,
+      "loss": 1.3375,
+      "step": 5366
+    },
+    {
+      "epoch": 0.4128143988923929,
+      "learning_rate": 0.001905734531735804,
+      "loss": 1.114,
+      "step": 5367
+    },
+    {
+      "epoch": 0.4128913160526113,
+      "learning_rate": 0.001905385567916458,
+      "loss": 1.443,
+      "step": 5368
+    },
+    {
+      "epoch": 0.41296823321282977,
+      "learning_rate": 0.0019050365804262255,
+      "loss": 1.3102,
+      "step": 5369
+    },
+    {
+      "epoch": 0.41304515037304823,
+      "learning_rate": 0.0019046875692854857,
+      "loss": 1.1855,
+      "step": 5370
+    },
+    {
+      "epoch": 0.4131220675332667,
+      "learning_rate": 0.001904338534514616,
+      "loss": 0.974,
+      "step": 5371
+    },
+    {
+      "epoch": 0.4131989846934851,
+      "learning_rate": 0.0019039894761339988,
+      "loss": 0.8566,
+      "step": 5372
+    },
+    {
+      "epoch": 0.41327590185370355,
+      "learning_rate": 0.0019036403941640146,
+      "loss": 0.8169,
+      "step": 5373
+    },
+    {
+      "epoch": 0.413352819013922,
+      "learning_rate": 0.0019032912886250471,
+      "loss": 1.17,
+      "step": 5374
+    },
+    {
+      "epoch": 0.41342973617414047,
+      "learning_rate": 0.0019029421595374814,
+      "loss": 0.9946,
+      "step": 5375
+    },
+    {
+      "epoch": 0.4135066533343589,
+      "learning_rate": 0.0019025930069217025,
+      "loss": 1.0334,
+      "step": 5376
+    },
+    {
+      "epoch": 0.41358357049457734,
+      "learning_rate": 0.001902243830798099,
+      "loss": 1.1297,
+      "step": 5377
+    },
+    {
+      "epoch": 0.4136604876547958,
+      "learning_rate": 0.0019018946311870582,
+      "loss": 0.894,
+      "step": 5378
+    },
+    {
+      "epoch": 0.41373740481501425,
+      "learning_rate": 0.0019015454081089717,
+      "loss": 0.9276,
+      "step": 5379
+    },
+    {
+      "epoch": 0.41381432197523266,
+      "learning_rate": 0.0019011961615842303,
+      "loss": 1.2543,
+      "step": 5380
+    },
+    {
+      "epoch": 0.4138912391354511,
+      "learning_rate": 0.001900846891633227,
+      "loss": 0.8844,
+      "step": 5381
+    },
+    {
+      "epoch": 0.4139681562956696,
+      "learning_rate": 0.0019004975982763552,
+      "loss": 1.4388,
+      "step": 5382
+    },
+    {
+      "epoch": 0.41404507345588804,
+      "learning_rate": 0.0019001482815340118,
+      "loss": 1.2338,
+      "step": 5383
+    },
+    {
+      "epoch": 0.41412199061610644,
+      "learning_rate": 0.001899798941426593,
+      "loss": 1.1151,
+      "step": 5384
+    },
+    {
+      "epoch": 0.4141989077763249,
+      "learning_rate": 0.0018994495779744976,
+      "loss": 1.1242,
+      "step": 5385
+    },
+    {
+      "epoch": 0.41427582493654336,
+      "learning_rate": 0.0018991001911981247,
+      "loss": 1.0195,
+      "step": 5386
+    },
+    {
+      "epoch": 0.41435274209676176,
+      "learning_rate": 0.0018987507811178762,
+      "loss": 1.0197,
+      "step": 5387
+    },
+    {
+      "epoch": 0.4144296592569802,
+      "learning_rate": 0.0018984013477541537,
+      "loss": 1.2167,
+      "step": 5388
+    },
+    {
+      "epoch": 0.4145065764171987,
+      "learning_rate": 0.0018980518911273611,
+      "loss": 0.9649,
+      "step": 5389
+    },
+    {
+      "epoch": 0.41458349357741714,
+      "learning_rate": 0.0018977024112579043,
+      "loss": 0.9731,
+      "step": 5390
+    },
+    {
+      "epoch": 0.41466041073763554,
+      "learning_rate": 0.001897352908166189,
+      "loss": 1.101,
+      "step": 5391
+    },
+    {
+      "epoch": 0.414737327897854,
+      "learning_rate": 0.0018970033818726231,
+      "loss": 1.3306,
+      "step": 5392
+    },
+    {
+      "epoch": 0.41481424505807246,
+      "learning_rate": 0.001896653832397616,
+      "loss": 1.1095,
+      "step": 5393
+    },
+    {
+      "epoch": 0.4148911622182909,
+      "learning_rate": 0.0018963042597615785,
+      "loss": 1.1044,
+      "step": 5394
+    },
+    {
+      "epoch": 0.4149680793785093,
+      "learning_rate": 0.0018959546639849224,
+      "loss": 0.8383,
+      "step": 5395
+    },
+    {
+      "epoch": 0.4150449965387278,
+      "learning_rate": 0.00189560504508806,
+      "loss": 1.2859,
+      "step": 5396
+    },
+    {
+      "epoch": 0.41512191369894624,
+      "learning_rate": 0.0018952554030914075,
+      "loss": 1.0922,
+      "step": 5397
+    },
+    {
+      "epoch": 0.4151988308591647,
+      "learning_rate": 0.0018949057380153794,
+      "loss": 1.3724,
+      "step": 5398
+    },
+    {
+      "epoch": 0.4152757480193831,
+      "learning_rate": 0.0018945560498803945,
+      "loss": 0.913,
+      "step": 5399
+    },
+    {
+      "epoch": 0.41535266517960157,
+      "learning_rate": 0.00189420633870687,
+      "loss": 1.3185,
+      "step": 5400
+    },
+    {
+      "epoch": 0.41542958233982,
+      "learning_rate": 0.001893856604515227,
+      "loss": 1.3982,
+      "step": 5401
+    },
+    {
+      "epoch": 0.4155064995000385,
+      "learning_rate": 0.001893506847325886,
+      "loss": 1.3029,
+      "step": 5402
+    },
+    {
+      "epoch": 0.4155834166602569,
+      "learning_rate": 0.001893157067159271,
+      "loss": 1.2488,
+      "step": 5403
+    },
+    {
+      "epoch": 0.41566033382047535,
+      "learning_rate": 0.001892807264035804,
+      "loss": 0.968,
+      "step": 5404
+    },
+    {
+      "epoch": 0.4157372509806938,
+      "learning_rate": 0.0018924574379759116,
+      "loss": 0.9755,
+      "step": 5405
+    },
+    {
+      "epoch": 0.4158141681409122,
+      "learning_rate": 0.001892107589000021,
+      "loss": 1.2083,
+      "step": 5406
+    },
+    {
+      "epoch": 0.41589108530113067,
+      "learning_rate": 0.001891757717128559,
+      "loss": 1.126,
+      "step": 5407
+    },
+    {
+      "epoch": 0.41596800246134913,
+      "learning_rate": 0.001891407822381956,
+      "loss": 1.2708,
+      "step": 5408
+    },
+    {
+      "epoch": 0.4160449196215676,
+      "learning_rate": 0.001891057904780642,
+      "loss": 1.1473,
+      "step": 5409
+    },
+    {
+      "epoch": 0.416121836781786,
+      "learning_rate": 0.0018907079643450501,
+      "loss": 1.0215,
+      "step": 5410
+    },
+    {
+      "epoch": 0.41619875394200445,
+      "learning_rate": 0.0018903580010956125,
+      "loss": 1.1171,
+      "step": 5411
+    },
+    {
+      "epoch": 0.4162756711022229,
+      "learning_rate": 0.0018900080150527645,
+      "loss": 1.1427,
+      "step": 5412
+    },
+    {
+      "epoch": 0.41635258826244137,
+      "learning_rate": 0.0018896580062369415,
+      "loss": 1.4094,
+      "step": 5413
+    },
+    {
+      "epoch": 0.4164295054226598,
+      "learning_rate": 0.0018893079746685822,
+      "loss": 1.1212,
+      "step": 5414
+    },
+    {
+      "epoch": 0.41650642258287823,
+      "learning_rate": 0.0018889579203681243,
+      "loss": 1.337,
+      "step": 5415
+    },
+    {
+      "epoch": 0.4165833397430967,
+      "learning_rate": 0.001888607843356008,
+      "loss": 1.1354,
+      "step": 5416
+    },
+    {
+      "epoch": 0.41666025690331515,
+      "learning_rate": 0.0018882577436526743,
+      "loss": 1.4084,
+      "step": 5417
+    },
+    {
+      "epoch": 0.41673717406353356,
+      "learning_rate": 0.0018879076212785669,
+      "loss": 0.8814,
+      "step": 5418
+    },
+    {
+      "epoch": 0.416814091223752,
+      "learning_rate": 0.001887557476254129,
+      "loss": 0.5657,
+      "step": 5419
+    },
+    {
+      "epoch": 0.4168910083839705,
+      "learning_rate": 0.001887207308599806,
+      "loss": 1.0559,
+      "step": 5420
+    },
+    {
+      "epoch": 0.41696792554418893,
+      "learning_rate": 0.001886857118336045,
+      "loss": 0.9819,
+      "step": 5421
+    },
+    {
+      "epoch": 0.41704484270440734,
+      "learning_rate": 0.001886506905483294,
+      "loss": 1.298,
+      "step": 5422
+    },
+    {
+      "epoch": 0.4171217598646258,
+      "learning_rate": 0.0018861566700620015,
+      "loss": 1.2829,
+      "step": 5423
+    },
+    {
+      "epoch": 0.41719867702484426,
+      "learning_rate": 0.0018858064120926183,
+      "loss": 0.927,
+      "step": 5424
+    },
+    {
+      "epoch": 0.4172755941850627,
+      "learning_rate": 0.0018854561315955975,
+      "loss": 0.9657,
+      "step": 5425
+    },
+    {
+      "epoch": 0.4173525113452811,
+      "learning_rate": 0.0018851058285913912,
+      "loss": 0.9917,
+      "step": 5426
+    },
+    {
+      "epoch": 0.4174294285054996,
+      "learning_rate": 0.001884755503100454,
+      "loss": 1.196,
+      "step": 5427
+    },
+    {
+      "epoch": 0.41750634566571804,
+      "learning_rate": 0.0018844051551432416,
+      "loss": 1.2799,
+      "step": 5428
+    },
+    {
+      "epoch": 0.41758326282593644,
+      "learning_rate": 0.001884054784740212,
+      "loss": 0.8154,
+      "step": 5429
+    },
+    {
+      "epoch": 0.4176601799861549,
+      "learning_rate": 0.0018837043919118232,
+      "loss": 1.239,
+      "step": 5430
+    },
+    {
+      "epoch": 0.41773709714637336,
+      "learning_rate": 0.0018833539766785349,
+      "loss": 0.971,
+      "step": 5431
+    },
+    {
+      "epoch": 0.4178140143065918,
+      "learning_rate": 0.0018830035390608089,
+      "loss": 0.7035,
+      "step": 5432
+    },
+    {
+      "epoch": 0.4178909314668102,
+      "learning_rate": 0.0018826530790791065,
+      "loss": 1.2166,
+      "step": 5433
+    },
+    {
+      "epoch": 0.4179678486270287,
+      "learning_rate": 0.0018823025967538926,
+      "loss": 1.1455,
+      "step": 5434
+    },
+    {
+      "epoch": 0.41804476578724714,
+      "learning_rate": 0.001881952092105631,
+      "loss": 1.0731,
+      "step": 5435
+    },
+    {
+      "epoch": 0.4181216829474656,
+      "learning_rate": 0.0018816015651547892,
+      "loss": 0.9191,
+      "step": 5436
+    },
+    {
+      "epoch": 0.418198600107684,
+      "learning_rate": 0.0018812510159218344,
+      "loss": 1.2045,
+      "step": 5437
+    },
+    {
+      "epoch": 0.41827551726790246,
+      "learning_rate": 0.0018809004444272348,
+      "loss": 1.2937,
+      "step": 5438
+    },
+    {
+      "epoch": 0.4183524344281209,
+      "learning_rate": 0.0018805498506914615,
+      "loss": 1.1727,
+      "step": 5439
+    },
+    {
+      "epoch": 0.4184293515883394,
+      "learning_rate": 0.0018801992347349862,
+      "loss": 0.9741,
+      "step": 5440
+    },
+    {
+      "epoch": 0.4185062687485578,
+      "learning_rate": 0.0018798485965782813,
+      "loss": 1.2351,
+      "step": 5441
+    },
+    {
+      "epoch": 0.41858318590877625,
+      "learning_rate": 0.0018794979362418206,
+      "loss": 0.9889,
+      "step": 5442
+    },
+    {
+      "epoch": 0.4186601030689947,
+      "learning_rate": 0.0018791472537460802,
+      "loss": 1.1544,
+      "step": 5443
+    },
+    {
+      "epoch": 0.41873702022921316,
+      "learning_rate": 0.0018787965491115363,
+      "loss": 1.1413,
+      "step": 5444
+    },
+    {
+      "epoch": 0.41881393738943157,
+      "learning_rate": 0.0018784458223586678,
+      "loss": 1.21,
+      "step": 5445
+    },
+    {
+      "epoch": 0.41889085454965,
+      "learning_rate": 0.0018780950735079523,
+      "loss": 0.8953,
+      "step": 5446
+    },
+    {
+      "epoch": 0.4189677717098685,
+      "learning_rate": 0.001877744302579872,
+      "loss": 0.7567,
+      "step": 5447
+    },
+    {
+      "epoch": 0.4190446888700869,
+      "learning_rate": 0.0018773935095949084,
+      "loss": 1.1003,
+      "step": 5448
+    },
+    {
+      "epoch": 0.41912160603030535,
+      "learning_rate": 0.0018770426945735436,
+      "loss": 0.9116,
+      "step": 5449
+    },
+    {
+      "epoch": 0.4191985231905238,
+      "learning_rate": 0.001876691857536264,
+      "loss": 1.0287,
+      "step": 5450
+    },
+    {
+      "epoch": 0.41927544035074227,
+      "learning_rate": 0.0018763409985035533,
+      "loss": 1.4729,
+      "step": 5451
+    },
+    {
+      "epoch": 0.41935235751096067,
+      "learning_rate": 0.0018759901174959003,
+      "loss": 1.2076,
+      "step": 5452
+    },
+    {
+      "epoch": 0.41942927467117913,
+      "learning_rate": 0.001875639214533792,
+      "loss": 1.1054,
+      "step": 5453
+    },
+    {
+      "epoch": 0.4195061918313976,
+      "learning_rate": 0.0018752882896377189,
+      "loss": 1.0048,
+      "step": 5454
+    },
+    {
+      "epoch": 0.41958310899161605,
+      "learning_rate": 0.0018749373428281707,
+      "loss": 1.3677,
+      "step": 5455
+    },
+    {
+      "epoch": 0.41966002615183445,
+      "learning_rate": 0.0018745863741256412,
+      "loss": 1.2008,
+      "step": 5456
+    },
+    {
+      "epoch": 0.4197369433120529,
+      "learning_rate": 0.0018742353835506224,
+      "loss": 0.9679,
+      "step": 5457
+    },
+    {
+      "epoch": 0.41981386047227137,
+      "learning_rate": 0.0018738843711236094,
+      "loss": 1.0629,
+      "step": 5458
+    },
+    {
+      "epoch": 0.41989077763248983,
+      "learning_rate": 0.0018735333368650982,
+      "loss": 1.288,
+      "step": 5459
+    },
+    {
+      "epoch": 0.41996769479270823,
+      "learning_rate": 0.0018731822807955868,
+      "loss": 1.1632,
+      "step": 5460
+    },
+    {
+      "epoch": 0.4200446119529267,
+      "learning_rate": 0.001872831202935573,
+      "loss": 1.4204,
+      "step": 5461
+    },
+    {
+      "epoch": 0.42012152911314515,
+      "learning_rate": 0.0018724801033055563,
+      "loss": 1.1945,
+      "step": 5462
+    },
+    {
+      "epoch": 0.4201984462733636,
+      "learning_rate": 0.0018721289819260383,
+      "loss": 1.0537,
+      "step": 5463
+    },
+    {
+      "epoch": 0.420275363433582,
+      "learning_rate": 0.0018717778388175213,
+      "loss": 0.9874,
+      "step": 5464
+    },
+    {
+      "epoch": 0.4203522805938005,
+      "learning_rate": 0.001871426674000509,
+      "loss": 0.6534,
+      "step": 5465
+    },
+    {
+      "epoch": 0.42042919775401894,
+      "learning_rate": 0.001871075487495506,
+      "loss": 1.3472,
+      "step": 5466
+    },
+    {
+      "epoch": 0.42050611491423734,
+      "learning_rate": 0.0018707242793230185,
+      "loss": 0.9003,
+      "step": 5467
+    },
+    {
+      "epoch": 0.4205830320744558,
+      "learning_rate": 0.001870373049503554,
+      "loss": 1.5472,
+      "step": 5468
+    },
+    {
+      "epoch": 0.42065994923467426,
+      "learning_rate": 0.0018700217980576212,
+      "loss": 1.2338,
+      "step": 5469
+    },
+    {
+      "epoch": 0.4207368663948927,
+      "learning_rate": 0.0018696705250057296,
+      "loss": 1.6406,
+      "step": 5470
+    },
+    {
+      "epoch": 0.4208137835551111,
+      "learning_rate": 0.0018693192303683914,
+      "loss": 0.986,
+      "step": 5471
+    },
+    {
+      "epoch": 0.4208907007153296,
+      "learning_rate": 0.0018689679141661183,
+      "loss": 1.5575,
+      "step": 5472
+    },
+    {
+      "epoch": 0.42096761787554804,
+      "learning_rate": 0.001868616576419424,
+      "loss": 1.0817,
+      "step": 5473
+    },
+    {
+      "epoch": 0.4210445350357665,
+      "learning_rate": 0.0018682652171488235,
+      "loss": 1.5008,
+      "step": 5474
+    },
+    {
+      "epoch": 0.4211214521959849,
+      "learning_rate": 0.0018679138363748334,
+      "loss": 1.1259,
+      "step": 5475
+    },
+    {
+      "epoch": 0.42119836935620336,
+      "learning_rate": 0.001867562434117971,
+      "loss": 1.2873,
+      "step": 5476
+    },
+    {
+      "epoch": 0.4212752865164218,
+      "learning_rate": 0.0018672110103987547,
+      "loss": 1.1144,
+      "step": 5477
+    },
+    {
+      "epoch": 0.4213522036766403,
+      "learning_rate": 0.0018668595652377052,
+      "loss": 1.278,
+      "step": 5478
+    },
+    {
+      "epoch": 0.4214291208368587,
+      "learning_rate": 0.001866508098655343,
+      "loss": 0.724,
+      "step": 5479
+    },
+    {
+      "epoch": 0.42150603799707714,
+      "learning_rate": 0.0018661566106721913,
+      "loss": 1.1591,
+      "step": 5480
+    },
+    {
+      "epoch": 0.4215829551572956,
+      "learning_rate": 0.0018658051013087731,
+      "loss": 1.2845,
+      "step": 5481
+    },
+    {
+      "epoch": 0.42165987231751406,
+      "learning_rate": 0.001865453570585614,
+      "loss": 0.8868,
+      "step": 5482
+    },
+    {
+      "epoch": 0.42173678947773247,
+      "learning_rate": 0.0018651020185232404,
+      "loss": 1.2903,
+      "step": 5483
+    },
+    {
+      "epoch": 0.4218137066379509,
+      "learning_rate": 0.0018647504451421786,
+      "loss": 0.896,
+      "step": 5484
+    },
+    {
+      "epoch": 0.4218906237981694,
+      "learning_rate": 0.001864398850462959,
+      "loss": 0.9226,
+      "step": 5485
+    },
+    {
+      "epoch": 0.42196754095838784,
+      "learning_rate": 0.00186404723450611,
+      "loss": 1.099,
+      "step": 5486
+    },
+    {
+      "epoch": 0.42204445811860625,
+      "learning_rate": 0.0018636955972921644,
+      "loss": 1.0727,
+      "step": 5487
+    },
+    {
+      "epoch": 0.4221213752788247,
+      "learning_rate": 0.001863343938841653,
+      "loss": 1.1478,
+      "step": 5488
+    },
+    {
+      "epoch": 0.42219829243904317,
+      "learning_rate": 0.0018629922591751112,
+      "loss": 1.372,
+      "step": 5489
+    },
+    {
+      "epoch": 0.42227520959926157,
+      "learning_rate": 0.0018626405583130727,
+      "loss": 1.057,
+      "step": 5490
+    },
+    {
+      "epoch": 0.42235212675948003,
+      "learning_rate": 0.0018622888362760741,
+      "loss": 1.0178,
+      "step": 5491
+    },
+    {
+      "epoch": 0.4224290439196985,
+      "learning_rate": 0.0018619370930846525,
+      "loss": 1.108,
+      "step": 5492
+    },
+    {
+      "epoch": 0.42250596107991695,
+      "learning_rate": 0.0018615853287593472,
+      "loss": 1.1004,
+      "step": 5493
+    },
+    {
+      "epoch": 0.42258287824013535,
+      "learning_rate": 0.0018612335433206976,
+      "loss": 1.305,
+      "step": 5494
+    },
+    {
+      "epoch": 0.4226597954003538,
+      "learning_rate": 0.0018608817367892446,
+      "loss": 1.0859,
+      "step": 5495
+    },
+    {
+      "epoch": 0.42273671256057227,
+      "learning_rate": 0.0018605299091855316,
+      "loss": 1.0684,
+      "step": 5496
+    },
+    {
+      "epoch": 0.42281362972079073,
+      "learning_rate": 0.0018601780605301006,
+      "loss": 1.174,
+      "step": 5497
+    },
+    {
+      "epoch": 0.42289054688100913,
+      "learning_rate": 0.0018598261908434983,
+      "loss": 0.8731,
+      "step": 5498
+    },
+    {
+      "epoch": 0.4229674640412276,
+      "learning_rate": 0.001859474300146269,
+      "loss": 1.2816,
+      "step": 5499
+    },
+    {
+      "epoch": 0.42304438120144605,
+      "learning_rate": 0.001859122388458961,
+      "loss": 1.2088,
+      "step": 5500
+    },
+    {
+      "epoch": 0.4231212983616645,
+      "learning_rate": 0.0018587704558021218,
+      "loss": 1.0698,
+      "step": 5501
+    },
+    {
+      "epoch": 0.4231982155218829,
+      "learning_rate": 0.0018584185021963023,
+      "loss": 0.8157,
+      "step": 5502
+    },
+    {
+      "epoch": 0.4232751326821014,
+      "learning_rate": 0.0018580665276620532,
+      "loss": 0.8452,
+      "step": 5503
+    },
+    {
+      "epoch": 0.42335204984231983,
+      "learning_rate": 0.001857714532219926,
+      "loss": 1.0249,
+      "step": 5504
+    },
+    {
+      "epoch": 0.4234289670025383,
+      "learning_rate": 0.0018573625158904744,
+      "loss": 1.104,
+      "step": 5505
+    },
+    {
+      "epoch": 0.4235058841627567,
+      "learning_rate": 0.0018570104786942534,
+      "loss": 1.1015,
+      "step": 5506
+    },
+    {
+      "epoch": 0.42358280132297516,
+      "learning_rate": 0.0018566584206518182,
+      "loss": 1.105,
+      "step": 5507
+    },
+    {
+      "epoch": 0.4236597184831936,
+      "learning_rate": 0.0018563063417837262,
+      "loss": 1.0311,
+      "step": 5508
+    },
+    {
+      "epoch": 0.423736635643412,
+      "learning_rate": 0.001855954242110536,
+      "loss": 1.395,
+      "step": 5509
+    },
+    {
+      "epoch": 0.4238135528036305,
+      "learning_rate": 0.001855602121652806,
+      "loss": 1.0516,
+      "step": 5510
+    },
+    {
+      "epoch": 0.42389046996384894,
+      "learning_rate": 0.0018552499804310976,
+      "loss": 1.8601,
+      "step": 5511
+    },
+    {
+      "epoch": 0.4239673871240674,
+      "learning_rate": 0.0018548978184659728,
+      "loss": 1.1041,
+      "step": 5512
+    },
+    {
+      "epoch": 0.4240443042842858,
+      "learning_rate": 0.0018545456357779945,
+      "loss": 1.058,
+      "step": 5513
+    },
+    {
+      "epoch": 0.42412122144450426,
+      "learning_rate": 0.001854193432387727,
+      "loss": 0.8967,
+      "step": 5514
+    },
+    {
+      "epoch": 0.4241981386047227,
+      "learning_rate": 0.0018538412083157355,
+      "loss": 1.004,
+      "step": 5515
+    },
+    {
+      "epoch": 0.4242750557649412,
+      "learning_rate": 0.0018534889635825878,
+      "loss": 1.0781,
+      "step": 5516
+    },
+    {
+      "epoch": 0.4243519729251596,
+      "learning_rate": 0.0018531366982088506,
+      "loss": 0.9593,
+      "step": 5517
+    },
+    {
+      "epoch": 0.42442889008537804,
+      "learning_rate": 0.001852784412215094,
+      "loss": 1.1843,
+      "step": 5518
+    },
+    {
+      "epoch": 0.4245058072455965,
+      "learning_rate": 0.0018524321056218873,
+      "loss": 1.1723,
+      "step": 5519
+    },
+    {
+      "epoch": 0.42458272440581496,
+      "learning_rate": 0.0018520797784498028,
+      "loss": 1.0247,
+      "step": 5520
+    },
+    {
+      "epoch": 0.42465964156603336,
+      "learning_rate": 0.0018517274307194135,
+      "loss": 1.1903,
+      "step": 5521
+    },
+    {
+      "epoch": 0.4247365587262518,
+      "learning_rate": 0.001851375062451293,
+      "loss": 1.2003,
+      "step": 5522
+    },
+    {
+      "epoch": 0.4248134758864703,
+      "learning_rate": 0.0018510226736660157,
+      "loss": 1.0411,
+      "step": 5523
+    },
+    {
+      "epoch": 0.42489039304668874,
+      "learning_rate": 0.0018506702643841593,
+      "loss": 1.2441,
+      "step": 5524
+    },
+    {
+      "epoch": 0.42496731020690715,
+      "learning_rate": 0.0018503178346263006,
+      "loss": 1.2971,
+      "step": 5525
+    },
+    {
+      "epoch": 0.4250442273671256,
+      "learning_rate": 0.001849965384413018,
+      "loss": 0.9883,
+      "step": 5526
+    },
+    {
+      "epoch": 0.42512114452734406,
+      "learning_rate": 0.0018496129137648926,
+      "loss": 0.9447,
+      "step": 5527
+    },
+    {
+      "epoch": 0.42519806168756247,
+      "learning_rate": 0.0018492604227025042,
+      "loss": 1.2593,
+      "step": 5528
+    },
+    {
+      "epoch": 0.4252749788477809,
+      "learning_rate": 0.0018489079112464363,
+      "loss": 1.2315,
+      "step": 5529
+    },
+    {
+      "epoch": 0.4253518960079994,
+      "learning_rate": 0.0018485553794172712,
+      "loss": 1.2355,
+      "step": 5530
+    },
+    {
+      "epoch": 0.42542881316821785,
+      "learning_rate": 0.0018482028272355947,
+      "loss": 1.0028,
+      "step": 5531
+    },
+    {
+      "epoch": 0.42550573032843625,
+      "learning_rate": 0.0018478502547219923,
+      "loss": 1.4217,
+      "step": 5532
+    },
+    {
+      "epoch": 0.4255826474886547,
+      "learning_rate": 0.0018474976618970509,
+      "loss": 1.092,
+      "step": 5533
+    },
+    {
+      "epoch": 0.42565956464887317,
+      "learning_rate": 0.0018471450487813584,
+      "loss": 1.2022,
+      "step": 5534
+    },
+    {
+      "epoch": 0.4257364818090916,
+      "learning_rate": 0.001846792415395505,
+      "loss": 1.1645,
+      "step": 5535
+    },
+    {
+      "epoch": 0.42581339896931003,
+      "learning_rate": 0.0018464397617600813,
+      "loss": 1.3899,
+      "step": 5536
+    },
+    {
+      "epoch": 0.4258903161295285,
+      "learning_rate": 0.0018460870878956785,
+      "loss": 1.3774,
+      "step": 5537
+    },
+    {
+      "epoch": 0.42596723328974695,
+      "learning_rate": 0.0018457343938228902,
+      "loss": 1.3572,
+      "step": 5538
+    },
+    {
+      "epoch": 0.4260441504499654,
+      "learning_rate": 0.0018453816795623102,
+      "loss": 1.3027,
+      "step": 5539
+    },
+    {
+      "epoch": 0.4261210676101838,
+      "learning_rate": 0.0018450289451345343,
+      "loss": 1.1324,
+      "step": 5540
+    },
+    {
+      "epoch": 0.42619798477040227,
+      "learning_rate": 0.001844676190560158,
+      "loss": 1.3033,
+      "step": 5541
+    },
+    {
+      "epoch": 0.42627490193062073,
+      "learning_rate": 0.0018443234158597806,
+      "loss": 1.0375,
+      "step": 5542
+    },
+    {
+      "epoch": 0.4263518190908392,
+      "learning_rate": 0.0018439706210539991,
+      "loss": 1.0948,
+      "step": 5543
+    },
+    {
+      "epoch": 0.4264287362510576,
+      "learning_rate": 0.0018436178061634156,
+      "loss": 1.4625,
+      "step": 5544
+    },
+    {
+      "epoch": 0.42650565341127605,
+      "learning_rate": 0.0018432649712086296,
+      "loss": 0.951,
+      "step": 5545
+    },
+    {
+      "epoch": 0.4265825705714945,
+      "learning_rate": 0.0018429121162102447,
+      "loss": 1.0098,
+      "step": 5546
+    },
+    {
+      "epoch": 0.42665948773171297,
+      "learning_rate": 0.0018425592411888635,
+      "loss": 0.9701,
+      "step": 5547
+    },
+    {
+      "epoch": 0.4267364048919314,
+      "learning_rate": 0.0018422063461650915,
+      "loss": 1.4429,
+      "step": 5548
+    },
+    {
+      "epoch": 0.42681332205214983,
+      "learning_rate": 0.0018418534311595341,
+      "loss": 1.1941,
+      "step": 5549
+    },
+    {
+      "epoch": 0.4268902392123683,
+      "learning_rate": 0.0018415004961927983,
+      "loss": 1.0569,
+      "step": 5550
+    },
+    {
+      "epoch": 0.4269671563725867,
+      "learning_rate": 0.0018411475412854934,
+      "loss": 1.1631,
+      "step": 5551
+    },
+    {
+      "epoch": 0.42704407353280516,
+      "learning_rate": 0.0018407945664582273,
+      "loss": 1.0821,
+      "step": 5552
+    },
+    {
+      "epoch": 0.4271209906930236,
+      "learning_rate": 0.0018404415717316119,
+      "loss": 1.229,
+      "step": 5553
+    },
+    {
+      "epoch": 0.4271979078532421,
+      "learning_rate": 0.0018400885571262576,
+      "loss": 1.1604,
+      "step": 5554
+    },
+    {
+      "epoch": 0.4272748250134605,
+      "learning_rate": 0.0018397355226627793,
+      "loss": 1.2475,
+      "step": 5555
+    },
+    {
+      "epoch": 0.42735174217367894,
+      "learning_rate": 0.0018393824683617885,
+      "loss": 0.8418,
+      "step": 5556
+    },
+    {
+      "epoch": 0.4274286593338974,
+      "learning_rate": 0.0018390293942439023,
+      "loss": 0.9409,
+      "step": 5557
+    },
+    {
+      "epoch": 0.42750557649411586,
+      "learning_rate": 0.0018386763003297357,
+      "loss": 0.9576,
+      "step": 5558
+    },
+    {
+      "epoch": 0.42758249365433426,
+      "learning_rate": 0.0018383231866399077,
+      "loss": 1.0819,
+      "step": 5559
+    },
+    {
+      "epoch": 0.4276594108145527,
+      "learning_rate": 0.0018379700531950362,
+      "loss": 1.0788,
+      "step": 5560
+    },
+    {
+      "epoch": 0.4277363279747712,
+      "learning_rate": 0.0018376169000157406,
+      "loss": 1.0406,
+      "step": 5561
+    },
+    {
+      "epoch": 0.42781324513498964,
+      "learning_rate": 0.0018372637271226427,
+      "loss": 1.2434,
+      "step": 5562
+    },
+    {
+      "epoch": 0.42789016229520804,
+      "learning_rate": 0.0018369105345363646,
+      "loss": 1.2091,
+      "step": 5563
+    },
+    {
+      "epoch": 0.4279670794554265,
+      "learning_rate": 0.0018365573222775288,
+      "loss": 1.5499,
+      "step": 5564
+    },
+    {
+      "epoch": 0.42804399661564496,
+      "learning_rate": 0.0018362040903667602,
+      "loss": 1.2836,
+      "step": 5565
+    },
+    {
+      "epoch": 0.4281209137758634,
+      "learning_rate": 0.0018358508388246844,
+      "loss": 1.3395,
+      "step": 5566
+    },
+    {
+      "epoch": 0.4281978309360818,
+      "learning_rate": 0.0018354975676719285,
+      "loss": 1.2794,
+      "step": 5567
+    },
+    {
+      "epoch": 0.4282747480963003,
+      "learning_rate": 0.0018351442769291195,
+      "loss": 0.9135,
+      "step": 5568
+    },
+    {
+      "epoch": 0.42835166525651874,
+      "learning_rate": 0.001834790966616887,
+      "loss": 0.9336,
+      "step": 5569
+    },
+    {
+      "epoch": 0.42842858241673715,
+      "learning_rate": 0.0018344376367558613,
+      "loss": 1.334,
+      "step": 5570
+    },
+    {
+      "epoch": 0.4285054995769556,
+      "learning_rate": 0.0018340842873666734,
+      "loss": 0.8759,
+      "step": 5571
+    },
+    {
+      "epoch": 0.42858241673717407,
+      "learning_rate": 0.0018337309184699555,
+      "loss": 1.2565,
+      "step": 5572
+    },
+    {
+      "epoch": 0.4286593338973925,
+      "learning_rate": 0.0018333775300863415,
+      "loss": 1.0318,
+      "step": 5573
+    },
+    {
+      "epoch": 0.42873625105761093,
+      "learning_rate": 0.0018330241222364663,
+      "loss": 1.0225,
+      "step": 5574
+    },
+    {
+      "epoch": 0.4288131682178294,
+      "learning_rate": 0.0018326706949409657,
+      "loss": 1.1291,
+      "step": 5575
+    },
+    {
+      "epoch": 0.42889008537804785,
+      "learning_rate": 0.001832317248220476,
+      "loss": 1.0597,
+      "step": 5576
+    },
+    {
+      "epoch": 0.4289670025382663,
+      "learning_rate": 0.0018319637820956361,
+      "loss": 1.4029,
+      "step": 5577
+    },
+    {
+      "epoch": 0.4290439196984847,
+      "learning_rate": 0.0018316102965870852,
+      "loss": 1.3606,
+      "step": 5578
+    },
+    {
+      "epoch": 0.42912083685870317,
+      "learning_rate": 0.0018312567917154632,
+      "loss": 0.8228,
+      "step": 5579
+    },
+    {
+      "epoch": 0.42919775401892163,
+      "learning_rate": 0.001830903267501412,
+      "loss": 0.9363,
+      "step": 5580
+    },
+    {
+      "epoch": 0.4292746711791401,
+      "learning_rate": 0.001830549723965574,
+      "loss": 1.0098,
+      "step": 5581
+    },
+    {
+      "epoch": 0.4293515883393585,
+      "learning_rate": 0.0018301961611285936,
+      "loss": 1.3514,
+      "step": 5582
+    },
+    {
+      "epoch": 0.42942850549957695,
+      "learning_rate": 0.0018298425790111146,
+      "loss": 1.3572,
+      "step": 5583
+    },
+    {
+      "epoch": 0.4295054226597954,
+      "learning_rate": 0.001829488977633784,
+      "loss": 0.7179,
+      "step": 5584
+    },
+    {
+      "epoch": 0.42958233982001387,
+      "learning_rate": 0.0018291353570172484,
+      "loss": 1.2272,
+      "step": 5585
+    },
+    {
+      "epoch": 0.4296592569802323,
+      "learning_rate": 0.0018287817171821568,
+      "loss": 1.2496,
+      "step": 5586
+    },
+    {
+      "epoch": 0.42973617414045073,
+      "learning_rate": 0.0018284280581491576,
+      "loss": 1.0349,
+      "step": 5587
+    },
+    {
+      "epoch": 0.4298130913006692,
+      "learning_rate": 0.001828074379938902,
+      "loss": 1.124,
+      "step": 5588
+    },
+    {
+      "epoch": 0.42989000846088765,
+      "learning_rate": 0.001827720682572041,
+      "loss": 0.9621,
+      "step": 5589
+    },
+    {
+      "epoch": 0.42996692562110606,
+      "learning_rate": 0.0018273669660692285,
+      "loss": 0.7076,
+      "step": 5590
+    },
+    {
+      "epoch": 0.4300438427813245,
+      "learning_rate": 0.0018270132304511173,
+      "loss": 1.2028,
+      "step": 5591
+    },
+    {
+      "epoch": 0.430120759941543,
+      "learning_rate": 0.0018266594757383626,
+      "loss": 1.197,
+      "step": 5592
+    },
+    {
+      "epoch": 0.4301976771017614,
+      "learning_rate": 0.0018263057019516213,
+      "loss": 1.3546,
+      "step": 5593
+    },
+    {
+      "epoch": 0.43027459426197984,
+      "learning_rate": 0.0018259519091115492,
+      "loss": 1.001,
+      "step": 5594
+    },
+    {
+      "epoch": 0.4303515114221983,
+      "learning_rate": 0.0018255980972388063,
+      "loss": 0.8959,
+      "step": 5595
+    },
+    {
+      "epoch": 0.43042842858241676,
+      "learning_rate": 0.0018252442663540505,
+      "loss": 1.3395,
+      "step": 5596
+    },
+    {
+      "epoch": 0.43050534574263516,
+      "learning_rate": 0.0018248904164779437,
+      "loss": 1.051,
+      "step": 5597
+    },
+    {
+      "epoch": 0.4305822629028536,
+      "learning_rate": 0.001824536547631146,
+      "loss": 1.0275,
+      "step": 5598
+    },
+    {
+      "epoch": 0.4306591800630721,
+      "learning_rate": 0.0018241826598343217,
+      "loss": 1.1403,
+      "step": 5599
+    },
+    {
+      "epoch": 0.43073609722329054,
+      "learning_rate": 0.0018238287531081334,
+      "loss": 1.1042,
+      "step": 5600
+    },
+    {
+      "epoch": 0.43081301438350894,
+      "learning_rate": 0.0018234748274732473,
+      "loss": 1.3162,
+      "step": 5601
+    },
+    {
+      "epoch": 0.4308899315437274,
+      "learning_rate": 0.0018231208829503286,
+      "loss": 1.0446,
+      "step": 5602
+    },
+    {
+      "epoch": 0.43096684870394586,
+      "learning_rate": 0.0018227669195600448,
+      "loss": 1.2103,
+      "step": 5603
+    },
+    {
+      "epoch": 0.4310437658641643,
+      "learning_rate": 0.0018224129373230647,
+      "loss": 1.2208,
+      "step": 5604
+    },
+    {
+      "epoch": 0.4311206830243827,
+      "learning_rate": 0.0018220589362600568,
+      "loss": 1.1432,
+      "step": 5605
+    },
+    {
+      "epoch": 0.4311976001846012,
+      "learning_rate": 0.001821704916391692,
+      "loss": 1.3191,
+      "step": 5606
+    },
+    {
+      "epoch": 0.43127451734481964,
+      "learning_rate": 0.0018213508777386418,
+      "loss": 1.0648,
+      "step": 5607
+    },
+    {
+      "epoch": 0.4313514345050381,
+      "learning_rate": 0.0018209968203215788,
+      "loss": 1.197,
+      "step": 5608
+    },
+    {
+      "epoch": 0.4314283516652565,
+      "learning_rate": 0.0018206427441611773,
+      "loss": 1.1027,
+      "step": 5609
+    },
+    {
+      "epoch": 0.43150526882547496,
+      "learning_rate": 0.0018202886492781117,
+      "loss": 1.0762,
+      "step": 5610
+    },
+    {
+      "epoch": 0.4315821859856934,
+      "learning_rate": 0.0018199345356930579,
+      "loss": 1.2543,
+      "step": 5611
+    },
+    {
+      "epoch": 0.4316591031459118,
+      "learning_rate": 0.001819580403426693,
+      "loss": 1.2031,
+      "step": 5612
+    },
+    {
+      "epoch": 0.4317360203061303,
+      "learning_rate": 0.0018192262524996956,
+      "loss": 1.2362,
+      "step": 5613
+    },
+    {
+      "epoch": 0.43181293746634875,
+      "learning_rate": 0.0018188720829327447,
+      "loss": 1.3068,
+      "step": 5614
+    },
+    {
+      "epoch": 0.4318898546265672,
+      "learning_rate": 0.0018185178947465201,
+      "loss": 1.1597,
+      "step": 5615
+    },
+    {
+      "epoch": 0.4319667717867856,
+      "learning_rate": 0.0018181636879617041,
+      "loss": 1.1482,
+      "step": 5616
+    },
+    {
+      "epoch": 0.43204368894700407,
+      "learning_rate": 0.0018178094625989787,
+      "loss": 1.3016,
+      "step": 5617
+    },
+    {
+      "epoch": 0.4321206061072225,
+      "learning_rate": 0.0018174552186790273,
+      "loss": 1.0175,
+      "step": 5618
+    },
+    {
+      "epoch": 0.432197523267441,
+      "learning_rate": 0.0018171009562225353,
+      "loss": 1.2541,
+      "step": 5619
+    },
+    {
+      "epoch": 0.4322744404276594,
+      "learning_rate": 0.0018167466752501877,
+      "loss": 1.147,
+      "step": 5620
+    },
+    {
+      "epoch": 0.43235135758787785,
+      "learning_rate": 0.0018163923757826717,
+      "loss": 0.9687,
+      "step": 5621
+    },
+    {
+      "epoch": 0.4324282747480963,
+      "learning_rate": 0.001816038057840675,
+      "loss": 1.0988,
+      "step": 5622
+    },
+    {
+      "epoch": 0.43250519190831477,
+      "learning_rate": 0.0018156837214448871,
+      "loss": 1.2579,
+      "step": 5623
+    },
+    {
+      "epoch": 0.43258210906853317,
+      "learning_rate": 0.0018153293666159976,
+      "loss": 1.4583,
+      "step": 5624
+    },
+    {
+      "epoch": 0.43265902622875163,
+      "learning_rate": 0.0018149749933746972,
+      "loss": 1.2231,
+      "step": 5625
+    },
+    {
+      "epoch": 0.4327359433889701,
+      "learning_rate": 0.0018146206017416793,
+      "loss": 1.2069,
+      "step": 5626
+    },
+    {
+      "epoch": 0.43281286054918855,
+      "learning_rate": 0.0018142661917376361,
+      "loss": 1.0469,
+      "step": 5627
+    },
+    {
+      "epoch": 0.43288977770940695,
+      "learning_rate": 0.001813911763383263,
+      "loss": 1.0712,
+      "step": 5628
+    },
+    {
+      "epoch": 0.4329666948696254,
+      "learning_rate": 0.0018135573166992542,
+      "loss": 0.8022,
+      "step": 5629
+    },
+    {
+      "epoch": 0.43304361202984387,
+      "learning_rate": 0.0018132028517063076,
+      "loss": 1.0543,
+      "step": 5630
+    },
+    {
+      "epoch": 0.4331205291900623,
+      "learning_rate": 0.0018128483684251198,
+      "loss": 1.0692,
+      "step": 5631
+    },
+    {
+      "epoch": 0.43319744635028073,
+      "learning_rate": 0.0018124938668763896,
+      "loss": 1.1745,
+      "step": 5632
+    },
+    {
+      "epoch": 0.4332743635104992,
+      "learning_rate": 0.001812139347080817,
+      "loss": 1.2819,
+      "step": 5633
+    },
+    {
+      "epoch": 0.43335128067071765,
+      "learning_rate": 0.0018117848090591026,
+      "loss": 1.1315,
+      "step": 5634
+    },
+    {
+      "epoch": 0.43342819783093606,
+      "learning_rate": 0.001811430252831948,
+      "loss": 0.8297,
+      "step": 5635
+    },
+    {
+      "epoch": 0.4335051149911545,
+      "learning_rate": 0.0018110756784200563,
+      "loss": 1.1009,
+      "step": 5636
+    },
+    {
+      "epoch": 0.433582032151373,
+      "learning_rate": 0.001810721085844132,
+      "loss": 1.7179,
+      "step": 5637
+    },
+    {
+      "epoch": 0.43365894931159144,
+      "learning_rate": 0.0018103664751248792,
+      "loss": 1.2815,
+      "step": 5638
+    },
+    {
+      "epoch": 0.43373586647180984,
+      "learning_rate": 0.0018100118462830055,
+      "loss": 0.7268,
+      "step": 5639
+    },
+    {
+      "epoch": 0.4338127836320283,
+      "learning_rate": 0.0018096571993392156,
+      "loss": 1.2658,
+      "step": 5640
+    },
+    {
+      "epoch": 0.43388970079224676,
+      "learning_rate": 0.00180930253431422,
+      "loss": 1.2305,
+      "step": 5641
+    },
+    {
+      "epoch": 0.4339666179524652,
+      "learning_rate": 0.0018089478512287268,
+      "loss": 1.2178,
+      "step": 5642
+    },
+    {
+      "epoch": 0.4340435351126836,
+      "learning_rate": 0.001808593150103447,
+      "loss": 1.0977,
+      "step": 5643
+    },
+    {
+      "epoch": 0.4341204522729021,
+      "learning_rate": 0.0018082384309590909,
+      "loss": 0.9142,
+      "step": 5644
+    },
+    {
+      "epoch": 0.43419736943312054,
+      "learning_rate": 0.0018078836938163721,
+      "loss": 1.1176,
+      "step": 5645
+    },
+    {
+      "epoch": 0.434274286593339,
+      "learning_rate": 0.001807528938696003,
+      "loss": 1.1686,
+      "step": 5646
+    },
+    {
+      "epoch": 0.4343512037535574,
+      "learning_rate": 0.001807174165618699,
+      "loss": 1.4266,
+      "step": 5647
+    },
+    {
+      "epoch": 0.43442812091377586,
+      "learning_rate": 0.0018068193746051755,
+      "loss": 0.9128,
+      "step": 5648
+    },
+    {
+      "epoch": 0.4345050380739943,
+      "learning_rate": 0.0018064645656761485,
+      "loss": 1.3591,
+      "step": 5649
+    },
+    {
+      "epoch": 0.4345819552342128,
+      "learning_rate": 0.0018061097388523365,
+      "loss": 0.9289,
+      "step": 5650
+    },
+    {
+      "epoch": 0.4346588723944312,
+      "learning_rate": 0.0018057548941544576,
+      "loss": 1.333,
+      "step": 5651
+    },
+    {
+      "epoch": 0.43473578955464964,
+      "learning_rate": 0.0018054000316032324,
+      "loss": 1.2954,
+      "step": 5652
+    },
+    {
+      "epoch": 0.4348127067148681,
+      "learning_rate": 0.0018050451512193805,
+      "loss": 1.2906,
+      "step": 5653
+    },
+    {
+      "epoch": 0.4348896238750865,
+      "learning_rate": 0.0018046902530236245,
+      "loss": 1.2782,
+      "step": 5654
+    },
+    {
+      "epoch": 0.43496654103530497,
+      "learning_rate": 0.001804335337036687,
+      "loss": 1.3585,
+      "step": 5655
+    },
+    {
+      "epoch": 0.4350434581955234,
+      "learning_rate": 0.001803980403279292,
+      "loss": 1.176,
+      "step": 5656
+    },
+    {
+      "epoch": 0.4351203753557419,
+      "learning_rate": 0.001803625451772164,
+      "loss": 1.3991,
+      "step": 5657
+    },
+    {
+      "epoch": 0.4351972925159603,
+      "learning_rate": 0.0018032704825360303,
+      "loss": 1.5181,
+      "step": 5658
+    },
+    {
+      "epoch": 0.43527420967617875,
+      "learning_rate": 0.0018029154955916166,
+      "loss": 1.3791,
+      "step": 5659
+    },
+    {
+      "epoch": 0.4353511268363972,
+      "learning_rate": 0.0018025604909596513,
+      "loss": 1.3091,
+      "step": 5660
+    },
+    {
+      "epoch": 0.43542804399661567,
+      "learning_rate": 0.0018022054686608639,
+      "loss": 0.866,
+      "step": 5661
+    },
+    {
+      "epoch": 0.43550496115683407,
+      "learning_rate": 0.0018018504287159842,
+      "loss": 1.2028,
+      "step": 5662
+    },
+    {
+      "epoch": 0.43558187831705253,
+      "learning_rate": 0.0018014953711457432,
+      "loss": 1.0731,
+      "step": 5663
+    },
+    {
+      "epoch": 0.435658795477271,
+      "learning_rate": 0.0018011402959708734,
+      "loss": 1.0611,
+      "step": 5664
+    },
+    {
+      "epoch": 0.43573571263748945,
+      "learning_rate": 0.0018007852032121076,
+      "loss": 1.2305,
+      "step": 5665
+    },
+    {
+      "epoch": 0.43581262979770785,
+      "learning_rate": 0.0018004300928901806,
+      "loss": 0.7995,
+      "step": 5666
+    },
+    {
+      "epoch": 0.4358895469579263,
+      "learning_rate": 0.0018000749650258273,
+      "loss": 1.0112,
+      "step": 5667
+    },
+    {
+      "epoch": 0.43596646411814477,
+      "learning_rate": 0.001799719819639784,
+      "loss": 1.1825,
+      "step": 5668
+    },
+    {
+      "epoch": 0.43604338127836323,
+      "learning_rate": 0.0017993646567527886,
+      "loss": 1.0716,
+      "step": 5669
+    },
+    {
+      "epoch": 0.43612029843858163,
+      "learning_rate": 0.0017990094763855784,
+      "loss": 1.0991,
+      "step": 5670
+    },
+    {
+      "epoch": 0.4361972155988001,
+      "learning_rate": 0.001798654278558893,
+      "loss": 1.1066,
+      "step": 5671
+    },
+    {
+      "epoch": 0.43627413275901855,
+      "learning_rate": 0.0017982990632934735,
+      "loss": 1.0699,
+      "step": 5672
+    },
+    {
+      "epoch": 0.43635104991923696,
+      "learning_rate": 0.0017979438306100605,
+      "loss": 1.3438,
+      "step": 5673
+    },
+    {
+      "epoch": 0.4364279670794554,
+      "learning_rate": 0.001797588580529397,
+      "loss": 1.0997,
+      "step": 5674
+    },
+    {
+      "epoch": 0.4365048842396739,
+      "learning_rate": 0.001797233313072226,
+      "loss": 1.1237,
+      "step": 5675
+    },
+    {
+      "epoch": 0.43658180139989233,
+      "learning_rate": 0.001796878028259292,
+      "loss": 1.2894,
+      "step": 5676
+    },
+    {
+      "epoch": 0.43665871856011074,
+      "learning_rate": 0.0017965227261113408,
+      "loss": 1.348,
+      "step": 5677
+    },
+    {
+      "epoch": 0.4367356357203292,
+      "learning_rate": 0.0017961674066491184,
+      "loss": 1.0409,
+      "step": 5678
+    },
+    {
+      "epoch": 0.43681255288054766,
+      "learning_rate": 0.0017958120698933725,
+      "loss": 1.4251,
+      "step": 5679
+    },
+    {
+      "epoch": 0.4368894700407661,
+      "learning_rate": 0.0017954567158648513,
+      "loss": 1.2653,
+      "step": 5680
+    },
+    {
+      "epoch": 0.4369663872009845,
+      "learning_rate": 0.0017951013445843054,
+      "loss": 0.9678,
+      "step": 5681
+    },
+    {
+      "epoch": 0.437043304361203,
+      "learning_rate": 0.0017947459560724833,
+      "loss": 1.2467,
+      "step": 5682
+    },
+    {
+      "epoch": 0.43712022152142144,
+      "learning_rate": 0.0017943905503501384,
+      "loss": 1.444,
+      "step": 5683
+    },
+    {
+      "epoch": 0.4371971386816399,
+      "learning_rate": 0.001794035127438022,
+      "loss": 1.026,
+      "step": 5684
+    },
+    {
+      "epoch": 0.4372740558418583,
+      "learning_rate": 0.0017936796873568887,
+      "loss": 1.3701,
+      "step": 5685
+    },
+    {
+      "epoch": 0.43735097300207676,
+      "learning_rate": 0.0017933242301274917,
+      "loss": 1.5726,
+      "step": 5686
+    },
+    {
+      "epoch": 0.4374278901622952,
+      "learning_rate": 0.0017929687557705876,
+      "loss": 1.0995,
+      "step": 5687
+    },
+    {
+      "epoch": 0.4375048073225137,
+      "learning_rate": 0.0017926132643069322,
+      "loss": 1.2614,
+      "step": 5688
+    },
+    {
+      "epoch": 0.4375817244827321,
+      "learning_rate": 0.0017922577557572838,
+      "loss": 1.1429,
+      "step": 5689
+    },
+    {
+      "epoch": 0.43765864164295054,
+      "learning_rate": 0.0017919022301424003,
+      "loss": 0.8692,
+      "step": 5690
+    },
+    {
+      "epoch": 0.437735558803169,
+      "learning_rate": 0.0017915466874830412,
+      "loss": 1.3014,
+      "step": 5691
+    },
+    {
+      "epoch": 0.4378124759633874,
+      "learning_rate": 0.0017911911277999676,
+      "loss": 1.3442,
+      "step": 5692
+    },
+    {
+      "epoch": 0.43788939312360586,
+      "learning_rate": 0.00179083555111394,
+      "loss": 1.2153,
+      "step": 5693
+    },
+    {
+      "epoch": 0.4379663102838243,
+      "learning_rate": 0.0017904799574457219,
+      "loss": 0.8997,
+      "step": 5694
+    },
+    {
+      "epoch": 0.4380432274440428,
+      "learning_rate": 0.001790124346816076,
+      "loss": 1.1885,
+      "step": 5695
+    },
+    {
+      "epoch": 0.4381201446042612,
+      "learning_rate": 0.0017897687192457675,
+      "loss": 1.3203,
+      "step": 5696
+    },
+    {
+      "epoch": 0.43819706176447965,
+      "learning_rate": 0.001789413074755561,
+      "loss": 1.3783,
+      "step": 5697
+    },
+    {
+      "epoch": 0.4382739789246981,
+      "learning_rate": 0.001789057413366224,
+      "loss": 1.1416,
+      "step": 5698
+    },
+    {
+      "epoch": 0.43835089608491656,
+      "learning_rate": 0.001788701735098523,
+      "loss": 1.0234,
+      "step": 5699
+    },
+    {
+      "epoch": 0.43842781324513497,
+      "learning_rate": 0.0017883460399732267,
+      "loss": 1.4129,
+      "step": 5700
+    },
+    {
+      "epoch": 0.4385047304053534,
+      "learning_rate": 0.0017879903280111052,
+      "loss": 0.9845,
+      "step": 5701
+    },
+    {
+      "epoch": 0.4385816475655719,
+      "learning_rate": 0.0017876345992329278,
+      "loss": 1.2713,
+      "step": 5702
+    },
+    {
+      "epoch": 0.43865856472579035,
+      "learning_rate": 0.0017872788536594665,
+      "loss": 1.2226,
+      "step": 5703
+    },
+    {
+      "epoch": 0.43873548188600875,
+      "learning_rate": 0.0017869230913114937,
+      "loss": 1.0973,
+      "step": 5704
+    },
+    {
+      "epoch": 0.4388123990462272,
+      "learning_rate": 0.0017865673122097825,
+      "loss": 0.9595,
+      "step": 5705
+    },
+    {
+      "epoch": 0.43888931620644567,
+      "learning_rate": 0.001786211516375107,
+      "loss": 1.3162,
+      "step": 5706
+    },
+    {
+      "epoch": 0.4389662333666641,
+      "learning_rate": 0.0017858557038282433,
+      "loss": 1.8682,
+      "step": 5707
+    },
+    {
+      "epoch": 0.43904315052688253,
+      "learning_rate": 0.001785499874589967,
+      "loss": 1.0754,
+      "step": 5708
+    },
+    {
+      "epoch": 0.439120067687101,
+      "learning_rate": 0.0017851440286810554,
+      "loss": 0.9753,
+      "step": 5709
+    },
+    {
+      "epoch": 0.43919698484731945,
+      "learning_rate": 0.0017847881661222865,
+      "loss": 0.9731,
+      "step": 5710
+    },
+    {
+      "epoch": 0.4392739020075379,
+      "learning_rate": 0.00178443228693444,
+      "loss": 1.3326,
+      "step": 5711
+    },
+    {
+      "epoch": 0.4393508191677563,
+      "learning_rate": 0.0017840763911382961,
+      "loss": 1.1624,
+      "step": 5712
+    },
+    {
+      "epoch": 0.43942773632797477,
+      "learning_rate": 0.0017837204787546353,
+      "loss": 1.0467,
+      "step": 5713
+    },
+    {
+      "epoch": 0.43950465348819323,
+      "learning_rate": 0.0017833645498042403,
+      "loss": 1.3491,
+      "step": 5714
+    },
+    {
+      "epoch": 0.43958157064841163,
+      "learning_rate": 0.0017830086043078936,
+      "loss": 1.1612,
+      "step": 5715
+    },
+    {
+      "epoch": 0.4396584878086301,
+      "learning_rate": 0.0017826526422863801,
+      "loss": 1.4372,
+      "step": 5716
+    },
+    {
+      "epoch": 0.43973540496884855,
+      "learning_rate": 0.0017822966637604837,
+      "loss": 1.0515,
+      "step": 5717
+    },
+    {
+      "epoch": 0.439812322129067,
+      "learning_rate": 0.001781940668750991,
+      "loss": 1.3669,
+      "step": 5718
+    },
+    {
+      "epoch": 0.4398892392892854,
+      "learning_rate": 0.0017815846572786893,
+      "loss": 0.7621,
+      "step": 5719
+    },
+    {
+      "epoch": 0.4399661564495039,
+      "learning_rate": 0.0017812286293643657,
+      "loss": 1.3997,
+      "step": 5720
+    },
+    {
+      "epoch": 0.44004307360972233,
+      "learning_rate": 0.0017808725850288088,
+      "loss": 1.1387,
+      "step": 5721
+    },
+    {
+      "epoch": 0.4401199907699408,
+      "learning_rate": 0.0017805165242928095,
+      "loss": 0.9814,
+      "step": 5722
+    },
+    {
+      "epoch": 0.4401969079301592,
+      "learning_rate": 0.0017801604471771582,
+      "loss": 1.1478,
+      "step": 5723
+    },
+    {
+      "epoch": 0.44027382509037766,
+      "learning_rate": 0.001779804353702646,
+      "loss": 1.3558,
+      "step": 5724
+    },
+    {
+      "epoch": 0.4403507422505961,
+      "learning_rate": 0.001779448243890066,
+      "loss": 1.2087,
+      "step": 5725
+    },
+    {
+      "epoch": 0.4404276594108146,
+      "learning_rate": 0.0017790921177602117,
+      "loss": 0.8501,
+      "step": 5726
+    },
+    {
+      "epoch": 0.440504576571033,
+      "learning_rate": 0.0017787359753338786,
+      "loss": 1.232,
+      "step": 5727
+    },
+    {
+      "epoch": 0.44058149373125144,
+      "learning_rate": 0.0017783798166318605,
+      "loss": 0.9937,
+      "step": 5728
+    },
+    {
+      "epoch": 0.4406584108914699,
+      "learning_rate": 0.001778023641674955,
+      "loss": 1.2035,
+      "step": 5729
+    },
+    {
+      "epoch": 0.44073532805168836,
+      "learning_rate": 0.0017776674504839593,
+      "loss": 0.9493,
+      "step": 5730
+    },
+    {
+      "epoch": 0.44081224521190676,
+      "learning_rate": 0.001777311243079672,
+      "loss": 1.2342,
+      "step": 5731
+    },
+    {
+      "epoch": 0.4408891623721252,
+      "learning_rate": 0.0017769550194828923,
+      "loss": 1.0352,
+      "step": 5732
+    },
+    {
+      "epoch": 0.4409660795323437,
+      "learning_rate": 0.0017765987797144203,
+      "loss": 0.9839,
+      "step": 5733
+    },
+    {
+      "epoch": 0.4410429966925621,
+      "learning_rate": 0.0017762425237950572,
+      "loss": 1.2379,
+      "step": 5734
+    },
+    {
+      "epoch": 0.44111991385278054,
+      "learning_rate": 0.0017758862517456048,
+      "loss": 1.3134,
+      "step": 5735
+    },
+    {
+      "epoch": 0.441196831012999,
+      "learning_rate": 0.0017755299635868673,
+      "loss": 1.2512,
+      "step": 5736
+    },
+    {
+      "epoch": 0.44127374817321746,
+      "learning_rate": 0.0017751736593396474,
+      "loss": 1.0343,
+      "step": 5737
+    },
+    {
+      "epoch": 0.44135066533343587,
+      "learning_rate": 0.0017748173390247518,
+      "loss": 1.2389,
+      "step": 5738
+    },
+    {
+      "epoch": 0.4414275824936543,
+      "learning_rate": 0.0017744610026629846,
+      "loss": 1.0723,
+      "step": 5739
+    },
+    {
+      "epoch": 0.4415044996538728,
+      "learning_rate": 0.0017741046502751535,
+      "loss": 0.9816,
+      "step": 5740
+    },
+    {
+      "epoch": 0.44158141681409124,
+      "learning_rate": 0.001773748281882066,
+      "loss": 1.0717,
+      "step": 5741
+    },
+    {
+      "epoch": 0.44165833397430965,
+      "learning_rate": 0.0017733918975045314,
+      "loss": 1.069,
+      "step": 5742
+    },
+    {
+      "epoch": 0.4417352511345281,
+      "learning_rate": 0.001773035497163359,
+      "loss": 0.9084,
+      "step": 5743
+    },
+    {
+      "epoch": 0.44181216829474657,
+      "learning_rate": 0.0017726790808793589,
+      "loss": 1.2703,
+      "step": 5744
+    },
+    {
+      "epoch": 0.441889085454965,
+      "learning_rate": 0.0017723226486733436,
+      "loss": 1.0079,
+      "step": 5745
+    },
+    {
+      "epoch": 0.44196600261518343,
+      "learning_rate": 0.001771966200566125,
+      "loss": 1.2455,
+      "step": 5746
+    },
+    {
+      "epoch": 0.4420429197754019,
+      "learning_rate": 0.0017716097365785164,
+      "loss": 1.0852,
+      "step": 5747
+    },
+    {
+      "epoch": 0.44211983693562035,
+      "learning_rate": 0.0017712532567313321,
+      "loss": 1.359,
+      "step": 5748
+    },
+    {
+      "epoch": 0.4421967540958388,
+      "learning_rate": 0.0017708967610453876,
+      "loss": 0.7557,
+      "step": 5749
+    },
+    {
+      "epoch": 0.4422736712560572,
+      "learning_rate": 0.001770540249541499,
+      "loss": 1.3303,
+      "step": 5750
+    },
+    {
+      "epoch": 0.44235058841627567,
+      "learning_rate": 0.0017701837222404834,
+      "loss": 1.401,
+      "step": 5751
+    },
+    {
+      "epoch": 0.44242750557649413,
+      "learning_rate": 0.0017698271791631577,
+      "loss": 1.2947,
+      "step": 5752
+    },
+    {
+      "epoch": 0.4425044227367126,
+      "learning_rate": 0.0017694706203303428,
+      "loss": 0.7679,
+      "step": 5753
+    },
+    {
+      "epoch": 0.442581339896931,
+      "learning_rate": 0.0017691140457628573,
+      "loss": 1.3605,
+      "step": 5754
+    },
+    {
+      "epoch": 0.44265825705714945,
+      "learning_rate": 0.001768757455481522,
+      "loss": 1.3312,
+      "step": 5755
+    },
+    {
+      "epoch": 0.4427351742173679,
+      "learning_rate": 0.0017684008495071591,
+      "loss": 1.252,
+      "step": 5756
+    },
+    {
+      "epoch": 0.4428120913775863,
+      "learning_rate": 0.0017680442278605907,
+      "loss": 1.215,
+      "step": 5757
+    },
+    {
+      "epoch": 0.4428890085378048,
+      "learning_rate": 0.0017676875905626404,
+      "loss": 0.9961,
+      "step": 5758
+    },
+    {
+      "epoch": 0.44296592569802323,
+      "learning_rate": 0.0017673309376341328,
+      "loss": 0.9695,
+      "step": 5759
+    },
+    {
+      "epoch": 0.4430428428582417,
+      "learning_rate": 0.0017669742690958933,
+      "loss": 1.3749,
+      "step": 5760
+    },
+    {
+      "epoch": 0.4431197600184601,
+      "learning_rate": 0.0017666175849687478,
+      "loss": 1.0788,
+      "step": 5761
+    },
+    {
+      "epoch": 0.44319667717867856,
+      "learning_rate": 0.001766260885273524,
+      "loss": 1.4633,
+      "step": 5762
+    },
+    {
+      "epoch": 0.443273594338897,
+      "learning_rate": 0.001765904170031049,
+      "loss": 1.0799,
+      "step": 5763
+    },
+    {
+      "epoch": 0.4433505114991155,
+      "learning_rate": 0.0017655474392621526,
+      "loss": 1.1455,
+      "step": 5764
+    },
+    {
+      "epoch": 0.4434274286593339,
+      "learning_rate": 0.0017651906929876647,
+      "loss": 1.2592,
+      "step": 5765
+    },
+    {
+      "epoch": 0.44350434581955234,
+      "learning_rate": 0.0017648339312284155,
+      "loss": 1.2114,
+      "step": 5766
+    },
+    {
+      "epoch": 0.4435812629797708,
+      "learning_rate": 0.0017644771540052376,
+      "loss": 1.0834,
+      "step": 5767
+    },
+    {
+      "epoch": 0.44365818013998926,
+      "learning_rate": 0.0017641203613389625,
+      "loss": 1.2354,
+      "step": 5768
+    },
+    {
+      "epoch": 0.44373509730020766,
+      "learning_rate": 0.0017637635532504249,
+      "loss": 1.2626,
+      "step": 5769
+    },
+    {
+      "epoch": 0.4438120144604261,
+      "learning_rate": 0.001763406729760458,
+      "loss": 1.2834,
+      "step": 5770
+    },
+    {
+      "epoch": 0.4438889316206446,
+      "learning_rate": 0.0017630498908898977,
+      "loss": 1.4358,
+      "step": 5771
+    },
+    {
+      "epoch": 0.44396584878086304,
+      "learning_rate": 0.0017626930366595797,
+      "loss": 1.4712,
+      "step": 5772
+    },
+    {
+      "epoch": 0.44404276594108144,
+      "learning_rate": 0.0017623361670903426,
+      "loss": 1.3903,
+      "step": 5773
+    },
+    {
+      "epoch": 0.4441196831012999,
+      "learning_rate": 0.0017619792822030226,
+      "loss": 1.1489,
+      "step": 5774
+    },
+    {
+      "epoch": 0.44419660026151836,
+      "learning_rate": 0.0017616223820184594,
+      "loss": 1.3875,
+      "step": 5775
+    },
+    {
+      "epoch": 0.44427351742173676,
+      "learning_rate": 0.0017612654665574926,
+      "loss": 0.8662,
+      "step": 5776
+    },
+    {
+      "epoch": 0.4443504345819552,
+      "learning_rate": 0.001760908535840963,
+      "loss": 1.3263,
+      "step": 5777
+    },
+    {
+      "epoch": 0.4444273517421737,
+      "learning_rate": 0.0017605515898897121,
+      "loss": 1.2111,
+      "step": 5778
+    },
+    {
+      "epoch": 0.44450426890239214,
+      "learning_rate": 0.0017601946287245818,
+      "loss": 1.7218,
+      "step": 5779
+    },
+    {
+      "epoch": 0.44458118606261054,
+      "learning_rate": 0.0017598376523664172,
+      "loss": 1.4569,
+      "step": 5780
+    },
+    {
+      "epoch": 0.444658103222829,
+      "learning_rate": 0.0017594806608360603,
+      "loss": 1.1493,
+      "step": 5781
+    },
+    {
+      "epoch": 0.44473502038304746,
+      "learning_rate": 0.0017591236541543574,
+      "loss": 1.0143,
+      "step": 5782
+    },
+    {
+      "epoch": 0.4448119375432659,
+      "learning_rate": 0.001758766632342154,
+      "loss": 1.2816,
+      "step": 5783
+    },
+    {
+      "epoch": 0.4448888547034843,
+      "learning_rate": 0.001758409595420298,
+      "loss": 1.386,
+      "step": 5784
+    },
+    {
+      "epoch": 0.4449657718637028,
+      "learning_rate": 0.0017580525434096354,
+      "loss": 1.3543,
+      "step": 5785
+    },
+    {
+      "epoch": 0.44504268902392125,
+      "learning_rate": 0.0017576954763310168,
+      "loss": 1.3586,
+      "step": 5786
+    },
+    {
+      "epoch": 0.4451196061841397,
+      "learning_rate": 0.0017573383942052902,
+      "loss": 1.0857,
+      "step": 5787
+    },
+    {
+      "epoch": 0.4451965233443581,
+      "learning_rate": 0.0017569812970533068,
+      "loss": 1.2279,
+      "step": 5788
+    },
+    {
+      "epoch": 0.44527344050457657,
+      "learning_rate": 0.0017566241848959176,
+      "loss": 1.2447,
+      "step": 5789
+    },
+    {
+      "epoch": 0.445350357664795,
+      "learning_rate": 0.0017562670577539747,
+      "loss": 0.8943,
+      "step": 5790
+    },
+    {
+      "epoch": 0.4454272748250135,
+      "learning_rate": 0.0017559099156483312,
+      "loss": 1.0126,
+      "step": 5791
+    },
+    {
+      "epoch": 0.4455041919852319,
+      "learning_rate": 0.0017555527585998414,
+      "loss": 1.0517,
+      "step": 5792
+    },
+    {
+      "epoch": 0.44558110914545035,
+      "learning_rate": 0.0017551955866293596,
+      "loss": 1.0309,
+      "step": 5793
+    },
+    {
+      "epoch": 0.4456580263056688,
+      "learning_rate": 0.0017548383997577411,
+      "loss": 0.8542,
+      "step": 5794
+    },
+    {
+      "epoch": 0.4457349434658872,
+      "learning_rate": 0.0017544811980058437,
+      "loss": 1.2182,
+      "step": 5795
+    },
+    {
+      "epoch": 0.44581186062610567,
+      "learning_rate": 0.001754123981394523,
+      "loss": 1.0964,
+      "step": 5796
+    },
+    {
+      "epoch": 0.44588877778632413,
+      "learning_rate": 0.0017537667499446386,
+      "loss": 1.03,
+      "step": 5797
+    },
+    {
+      "epoch": 0.4459656949465426,
+      "learning_rate": 0.0017534095036770491,
+      "loss": 1.2094,
+      "step": 5798
+    },
+    {
+      "epoch": 0.446042612106761,
+      "learning_rate": 0.0017530522426126148,
+      "loss": 1.1494,
+      "step": 5799
+    },
+    {
+      "epoch": 0.44611952926697945,
+      "learning_rate": 0.0017526949667721959,
+      "loss": 1.0827,
+      "step": 5800
+    },
+    {
+      "epoch": 0.4461964464271979,
+      "learning_rate": 0.0017523376761766547,
+      "loss": 1.0839,
+      "step": 5801
+    },
+    {
+      "epoch": 0.44627336358741637,
+      "learning_rate": 0.0017519803708468536,
+      "loss": 1.3063,
+      "step": 5802
+    },
+    {
+      "epoch": 0.4463502807476348,
+      "learning_rate": 0.0017516230508036565,
+      "loss": 1.3156,
+      "step": 5803
+    },
+    {
+      "epoch": 0.44642719790785323,
+      "learning_rate": 0.001751265716067927,
+      "loss": 1.3592,
+      "step": 5804
+    },
+    {
+      "epoch": 0.4465041150680717,
+      "learning_rate": 0.00175090836666053,
+      "loss": 0.8599,
+      "step": 5805
+    },
+    {
+      "epoch": 0.44658103222829015,
+      "learning_rate": 0.0017505510026023326,
+      "loss": 0.965,
+      "step": 5806
+    },
+    {
+      "epoch": 0.44665794938850856,
+      "learning_rate": 0.0017501936239142008,
+      "loss": 1.2277,
+      "step": 5807
+    },
+    {
+      "epoch": 0.446734866548727,
+      "learning_rate": 0.0017498362306170028,
+      "loss": 0.9069,
+      "step": 5808
+    },
+    {
+      "epoch": 0.4468117837089455,
+      "learning_rate": 0.0017494788227316062,
+      "loss": 1.3905,
+      "step": 5809
+    },
+    {
+      "epoch": 0.44688870086916394,
+      "learning_rate": 0.0017491214002788816,
+      "loss": 1.2333,
+      "step": 5810
+    },
+    {
+      "epoch": 0.44696561802938234,
+      "learning_rate": 0.001748763963279699,
+      "loss": 1.1398,
+      "step": 5811
+    },
+    {
+      "epoch": 0.4470425351896008,
+      "learning_rate": 0.0017484065117549289,
+      "loss": 1.2049,
+      "step": 5812
+    },
+    {
+      "epoch": 0.44711945234981926,
+      "learning_rate": 0.001748049045725444,
+      "loss": 1.0972,
+      "step": 5813
+    },
+    {
+      "epoch": 0.4471963695100377,
+      "learning_rate": 0.0017476915652121165,
+      "loss": 0.8411,
+      "step": 5814
+    },
+    {
+      "epoch": 0.4472732866702561,
+      "learning_rate": 0.001747334070235821,
+      "loss": 1.3048,
+      "step": 5815
+    },
+    {
+      "epoch": 0.4473502038304746,
+      "learning_rate": 0.0017469765608174306,
+      "loss": 0.9927,
+      "step": 5816
+    },
+    {
+      "epoch": 0.44742712099069304,
+      "learning_rate": 0.0017466190369778219,
+      "loss": 1.1908,
+      "step": 5817
+    },
+    {
+      "epoch": 0.44750403815091144,
+      "learning_rate": 0.0017462614987378707,
+      "loss": 1.0945,
+      "step": 5818
+    },
+    {
+      "epoch": 0.4475809553111299,
+      "learning_rate": 0.0017459039461184536,
+      "loss": 1.183,
+      "step": 5819
+    },
+    {
+      "epoch": 0.44765787247134836,
+      "learning_rate": 0.001745546379140449,
+      "loss": 0.9948,
+      "step": 5820
+    },
+    {
+      "epoch": 0.4477347896315668,
+      "learning_rate": 0.0017451887978247353,
+      "loss": 1.2216,
+      "step": 5821
+    },
+    {
+      "epoch": 0.4478117067917852,
+      "learning_rate": 0.0017448312021921923,
+      "loss": 1.1367,
+      "step": 5822
+    },
+    {
+      "epoch": 0.4478886239520037,
+      "learning_rate": 0.0017444735922637,
+      "loss": 0.8306,
+      "step": 5823
+    },
+    {
+      "epoch": 0.44796554111222214,
+      "learning_rate": 0.0017441159680601404,
+      "loss": 1.0624,
+      "step": 5824
+    },
+    {
+      "epoch": 0.4480424582724406,
+      "learning_rate": 0.0017437583296023944,
+      "loss": 1.1724,
+      "step": 5825
+    },
+    {
+      "epoch": 0.448119375432659,
+      "learning_rate": 0.0017434006769113462,
+      "loss": 1.2228,
+      "step": 5826
+    },
+    {
+      "epoch": 0.44819629259287747,
+      "learning_rate": 0.001743043010007878,
+      "loss": 1.3137,
+      "step": 5827
+    },
+    {
+      "epoch": 0.4482732097530959,
+      "learning_rate": 0.0017426853289128758,
+      "loss": 1.1691,
+      "step": 5828
+    },
+    {
+      "epoch": 0.4483501269133144,
+      "learning_rate": 0.0017423276336472238,
+      "loss": 1.2528,
+      "step": 5829
+    },
+    {
+      "epoch": 0.4484270440735328,
+      "learning_rate": 0.001741969924231809,
+      "loss": 1.4442,
+      "step": 5830
+    },
+    {
+      "epoch": 0.44850396123375125,
+      "learning_rate": 0.0017416122006875183,
+      "loss": 1.3941,
+      "step": 5831
+    },
+    {
+      "epoch": 0.4485808783939697,
+      "learning_rate": 0.0017412544630352389,
+      "loss": 1.0074,
+      "step": 5832
+    },
+    {
+      "epoch": 0.44865779555418817,
+      "learning_rate": 0.0017408967112958606,
+      "loss": 0.8635,
+      "step": 5833
+    },
+    {
+      "epoch": 0.44873471271440657,
+      "learning_rate": 0.0017405389454902718,
+      "loss": 1.213,
+      "step": 5834
+    },
+    {
+      "epoch": 0.44881162987462503,
+      "learning_rate": 0.0017401811656393637,
+      "loss": 1.4124,
+      "step": 5835
+    },
+    {
+      "epoch": 0.4488885470348435,
+      "learning_rate": 0.0017398233717640264,
+      "loss": 1.3143,
+      "step": 5836
+    },
+    {
+      "epoch": 0.4489654641950619,
+      "learning_rate": 0.0017394655638851533,
+      "loss": 0.8716,
+      "step": 5837
+    },
+    {
+      "epoch": 0.44904238135528035,
+      "learning_rate": 0.0017391077420236358,
+      "loss": 1.4332,
+      "step": 5838
+    },
+    {
+      "epoch": 0.4491192985154988,
+      "learning_rate": 0.0017387499062003684,
+      "loss": 1.2571,
+      "step": 5839
+    },
+    {
+      "epoch": 0.44919621567571727,
+      "learning_rate": 0.0017383920564362446,
+      "loss": 1.1221,
+      "step": 5840
+    },
+    {
+      "epoch": 0.4492731328359357,
+      "learning_rate": 0.0017380341927521604,
+      "loss": 0.9311,
+      "step": 5841
+    },
+    {
+      "epoch": 0.44935004999615413,
+      "learning_rate": 0.001737676315169012,
+      "loss": 1.265,
+      "step": 5842
+    },
+    {
+      "epoch": 0.4494269671563726,
+      "learning_rate": 0.0017373184237076953,
+      "loss": 1.0609,
+      "step": 5843
+    },
+    {
+      "epoch": 0.44950388431659105,
+      "learning_rate": 0.001736960518389109,
+      "loss": 1.4313,
+      "step": 5844
+    },
+    {
+      "epoch": 0.44958080147680946,
+      "learning_rate": 0.0017366025992341509,
+      "loss": 1.187,
+      "step": 5845
+    },
+    {
+      "epoch": 0.4496577186370279,
+      "learning_rate": 0.0017362446662637203,
+      "loss": 0.9655,
+      "step": 5846
+    },
+    {
+      "epoch": 0.4497346357972464,
+      "learning_rate": 0.0017358867194987176,
+      "loss": 1.354,
+      "step": 5847
+    },
+    {
+      "epoch": 0.44981155295746483,
+      "learning_rate": 0.0017355287589600435,
+      "loss": 1.2368,
+      "step": 5848
+    },
+    {
+      "epoch": 0.44988847011768324,
+      "learning_rate": 0.0017351707846685997,
+      "loss": 1.2649,
+      "step": 5849
+    },
+    {
+      "epoch": 0.4499653872779017,
+      "learning_rate": 0.0017348127966452889,
+      "loss": 0.961,
+      "step": 5850
+    },
+    {
+      "epoch": 0.45004230443812016,
+      "learning_rate": 0.0017344547949110138,
+      "loss": 1.3079,
+      "step": 5851
+    },
+    {
+      "epoch": 0.4501192215983386,
+      "learning_rate": 0.0017340967794866788,
+      "loss": 0.9241,
+      "step": 5852
+    },
+    {
+      "epoch": 0.450196138758557,
+      "learning_rate": 0.0017337387503931895,
+      "loss": 1.0215,
+      "step": 5853
+    },
+    {
+      "epoch": 0.4502730559187755,
+      "learning_rate": 0.0017333807076514503,
+      "loss": 1.0799,
+      "step": 5854
+    },
+    {
+      "epoch": 0.45034997307899394,
+      "learning_rate": 0.0017330226512823688,
+      "loss": 1.0383,
+      "step": 5855
+    },
+    {
+      "epoch": 0.45042689023921234,
+      "learning_rate": 0.0017326645813068517,
+      "loss": 1.2903,
+      "step": 5856
+    },
+    {
+      "epoch": 0.4505038073994308,
+      "learning_rate": 0.001732306497745807,
+      "loss": 1.0836,
+      "step": 5857
+    },
+    {
+      "epoch": 0.45058072455964926,
+      "learning_rate": 0.001731948400620144,
+      "loss": 0.9747,
+      "step": 5858
+    },
+    {
+      "epoch": 0.4506576417198677,
+      "learning_rate": 0.0017315902899507722,
+      "loss": 1.3402,
+      "step": 5859
+    },
+    {
+      "epoch": 0.4507345588800861,
+      "learning_rate": 0.0017312321657586018,
+      "loss": 1.0289,
+      "step": 5860
+    },
+    {
+      "epoch": 0.4508114760403046,
+      "learning_rate": 0.0017308740280645447,
+      "loss": 1.0714,
+      "step": 5861
+    },
+    {
+      "epoch": 0.45088839320052304,
+      "learning_rate": 0.0017305158768895121,
+      "loss": 1.3094,
+      "step": 5862
+    },
+    {
+      "epoch": 0.4509653103607415,
+      "learning_rate": 0.0017301577122544175,
+      "loss": 1.3609,
+      "step": 5863
+    },
+    {
+      "epoch": 0.4510422275209599,
+      "learning_rate": 0.001729799534180174,
+      "loss": 1.097,
+      "step": 5864
+    },
+    {
+      "epoch": 0.45111914468117836,
+      "learning_rate": 0.0017294413426876961,
+      "loss": 1.1695,
+      "step": 5865
+    },
+    {
+      "epoch": 0.4511960618413968,
+      "learning_rate": 0.0017290831377978991,
+      "loss": 1.1512,
+      "step": 5866
+    },
+    {
+      "epoch": 0.4512729790016153,
+      "learning_rate": 0.0017287249195316989,
+      "loss": 1.2946,
+      "step": 5867
+    },
+    {
+      "epoch": 0.4513498961618337,
+      "learning_rate": 0.0017283666879100127,
+      "loss": 1.1683,
+      "step": 5868
+    },
+    {
+      "epoch": 0.45142681332205215,
+      "learning_rate": 0.0017280084429537573,
+      "loss": 1.0327,
+      "step": 5869
+    },
+    {
+      "epoch": 0.4515037304822706,
+      "learning_rate": 0.0017276501846838514,
+      "loss": 1.0762,
+      "step": 5870
+    },
+    {
+      "epoch": 0.45158064764248906,
+      "learning_rate": 0.0017272919131212138,
+      "loss": 1.0974,
+      "step": 5871
+    },
+    {
+      "epoch": 0.45165756480270747,
+      "learning_rate": 0.001726933628286765,
+      "loss": 1.1356,
+      "step": 5872
+    },
+    {
+      "epoch": 0.4517344819629259,
+      "learning_rate": 0.0017265753302014245,
+      "loss": 0.9996,
+      "step": 5873
+    },
+    {
+      "epoch": 0.4518113991231444,
+      "learning_rate": 0.0017262170188861144,
+      "loss": 1.367,
+      "step": 5874
+    },
+    {
+      "epoch": 0.45188831628336285,
+      "learning_rate": 0.001725858694361757,
+      "loss": 1.3565,
+      "step": 5875
+    },
+    {
+      "epoch": 0.45196523344358125,
+      "learning_rate": 0.001725500356649275,
+      "loss": 1.0324,
+      "step": 5876
+    },
+    {
+      "epoch": 0.4520421506037997,
+      "learning_rate": 0.0017251420057695923,
+      "loss": 1.1557,
+      "step": 5877
+    },
+    {
+      "epoch": 0.45211906776401817,
+      "learning_rate": 0.0017247836417436332,
+      "loss": 1.046,
+      "step": 5878
+    },
+    {
+      "epoch": 0.45219598492423657,
+      "learning_rate": 0.0017244252645923232,
+      "loss": 1.2456,
+      "step": 5879
+    },
+    {
+      "epoch": 0.45227290208445503,
+      "learning_rate": 0.0017240668743365877,
+      "loss": 0.9816,
+      "step": 5880
+    },
+    {
+      "epoch": 0.4523498192446735,
+      "learning_rate": 0.001723708470997354,
+      "loss": 0.8587,
+      "step": 5881
+    },
+    {
+      "epoch": 0.45242673640489195,
+      "learning_rate": 0.0017233500545955492,
+      "loss": 1.2587,
+      "step": 5882
+    },
+    {
+      "epoch": 0.45250365356511035,
+      "learning_rate": 0.0017229916251521027,
+      "loss": 1.4925,
+      "step": 5883
+    },
+    {
+      "epoch": 0.4525805707253288,
+      "learning_rate": 0.0017226331826879421,
+      "loss": 0.8827,
+      "step": 5884
+    },
+    {
+      "epoch": 0.45265748788554727,
+      "learning_rate": 0.0017222747272239987,
+      "loss": 1.0074,
+      "step": 5885
+    },
+    {
+      "epoch": 0.45273440504576573,
+      "learning_rate": 0.0017219162587812015,
+      "loss": 0.9539,
+      "step": 5886
+    },
+    {
+      "epoch": 0.45281132220598413,
+      "learning_rate": 0.0017215577773804832,
+      "loss": 1.323,
+      "step": 5887
+    },
+    {
+      "epoch": 0.4528882393662026,
+      "learning_rate": 0.0017211992830427755,
+      "loss": 1.3439,
+      "step": 5888
+    },
+    {
+      "epoch": 0.45296515652642105,
+      "learning_rate": 0.0017208407757890107,
+      "loss": 1.5041,
+      "step": 5889
+    },
+    {
+      "epoch": 0.4530420736866395,
+      "learning_rate": 0.001720482255640123,
+      "loss": 0.9133,
+      "step": 5890
+    },
+    {
+      "epoch": 0.4531189908468579,
+      "learning_rate": 0.0017201237226170468,
+      "loss": 1.3905,
+      "step": 5891
+    },
+    {
+      "epoch": 0.4531959080070764,
+      "learning_rate": 0.001719765176740717,
+      "loss": 1.0082,
+      "step": 5892
+    },
+    {
+      "epoch": 0.45327282516729483,
+      "learning_rate": 0.0017194066180320692,
+      "loss": 0.8672,
+      "step": 5893
+    },
+    {
+      "epoch": 0.4533497423275133,
+      "learning_rate": 0.0017190480465120405,
+      "loss": 1.1731,
+      "step": 5894
+    },
+    {
+      "epoch": 0.4534266594877317,
+      "learning_rate": 0.001718689462201568,
+      "loss": 1.1501,
+      "step": 5895
+    },
+    {
+      "epoch": 0.45350357664795016,
+      "learning_rate": 0.00171833086512159,
+      "loss": 1.1021,
+      "step": 5896
+    },
+    {
+      "epoch": 0.4535804938081686,
+      "learning_rate": 0.001717972255293045,
+      "loss": 1.3561,
+      "step": 5897
+    },
+    {
+      "epoch": 0.453657410968387,
+      "learning_rate": 0.0017176136327368735,
+      "loss": 1.106,
+      "step": 5898
+    },
+    {
+      "epoch": 0.4537343281286055,
+      "learning_rate": 0.0017172549974740147,
+      "loss": 1.1652,
+      "step": 5899
+    },
+    {
+      "epoch": 0.45381124528882394,
+      "learning_rate": 0.0017168963495254098,
+      "loss": 0.9884,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4538881624490424,
+      "learning_rate": 0.0017165376889120018,
+      "loss": 1.6408,
+      "step": 5901
+    },
+    {
+      "epoch": 0.4539650796092608,
+      "learning_rate": 0.001716179015654732,
+      "loss": 1.4123,
+      "step": 5902
+    },
+    {
+      "epoch": 0.45404199676947926,
+      "learning_rate": 0.0017158203297745443,
+      "loss": 1.0873,
+      "step": 5903
+    },
+    {
+      "epoch": 0.4541189139296977,
+      "learning_rate": 0.0017154616312923825,
+      "loss": 0.9099,
+      "step": 5904
+    },
+    {
+      "epoch": 0.4541958310899162,
+      "learning_rate": 0.0017151029202291917,
+      "loss": 0.9838,
+      "step": 5905
+    },
+    {
+      "epoch": 0.4542727482501346,
+      "learning_rate": 0.0017147441966059173,
+      "loss": 0.8756,
+      "step": 5906
+    },
+    {
+      "epoch": 0.45434966541035304,
+      "learning_rate": 0.001714385460443505,
+      "loss": 0.9054,
+      "step": 5907
+    },
+    {
+      "epoch": 0.4544265825705715,
+      "learning_rate": 0.0017140267117629027,
+      "loss": 1.375,
+      "step": 5908
+    },
+    {
+      "epoch": 0.45450349973078996,
+      "learning_rate": 0.0017136679505850572,
+      "loss": 1.2602,
+      "step": 5909
+    },
+    {
+      "epoch": 0.45458041689100837,
+      "learning_rate": 0.001713309176930918,
+      "loss": 1.0451,
+      "step": 5910
+    },
+    {
+      "epoch": 0.4546573340512268,
+      "learning_rate": 0.0017129503908214331,
+      "loss": 1.2153,
+      "step": 5911
+    },
+    {
+      "epoch": 0.4547342512114453,
+      "learning_rate": 0.001712591592277553,
+      "loss": 1.3271,
+      "step": 5912
+    },
+    {
+      "epoch": 0.45481116837166374,
+      "learning_rate": 0.0017122327813202283,
+      "loss": 1.3862,
+      "step": 5913
+    },
+    {
+      "epoch": 0.45488808553188215,
+      "learning_rate": 0.001711873957970411,
+      "loss": 1.17,
+      "step": 5914
+    },
+    {
+      "epoch": 0.4549650026921006,
+      "learning_rate": 0.0017115151222490516,
+      "loss": 1.4342,
+      "step": 5915
+    },
+    {
+      "epoch": 0.45504191985231907,
+      "learning_rate": 0.0017111562741771043,
+      "loss": 1.3069,
+      "step": 5916
+    },
+    {
+      "epoch": 0.45511883701253747,
+      "learning_rate": 0.0017107974137755219,
+      "loss": 1.0684,
+      "step": 5917
+    },
+    {
+      "epoch": 0.45519575417275593,
+      "learning_rate": 0.0017104385410652585,
+      "loss": 1.1121,
+      "step": 5918
+    },
+    {
+      "epoch": 0.4552726713329744,
+      "learning_rate": 0.0017100796560672697,
+      "loss": 1.2236,
+      "step": 5919
+    },
+    {
+      "epoch": 0.45534958849319285,
+      "learning_rate": 0.0017097207588025105,
+      "loss": 1.1302,
+      "step": 5920
+    },
+    {
+      "epoch": 0.45542650565341125,
+      "learning_rate": 0.0017093618492919386,
+      "loss": 1.1827,
+      "step": 5921
+    },
+    {
+      "epoch": 0.4555034228136297,
+      "learning_rate": 0.0017090029275565093,
+      "loss": 1.0719,
+      "step": 5922
+    },
+    {
+      "epoch": 0.45558033997384817,
+      "learning_rate": 0.0017086439936171815,
+      "loss": 1.2071,
+      "step": 5923
+    },
+    {
+      "epoch": 0.45565725713406663,
+      "learning_rate": 0.0017082850474949129,
+      "loss": 1.2975,
+      "step": 5924
+    },
+    {
+      "epoch": 0.45573417429428503,
+      "learning_rate": 0.0017079260892106643,
+      "loss": 1.4163,
+      "step": 5925
+    },
+    {
+      "epoch": 0.4558110914545035,
+      "learning_rate": 0.001707567118785394,
+      "loss": 1.0864,
+      "step": 5926
+    },
+    {
+      "epoch": 0.45588800861472195,
+      "learning_rate": 0.0017072081362400635,
+      "loss": 1.3652,
+      "step": 5927
+    },
+    {
+      "epoch": 0.4559649257749404,
+      "learning_rate": 0.001706849141595634,
+      "loss": 1.2553,
+      "step": 5928
+    },
+    {
+      "epoch": 0.4560418429351588,
+      "learning_rate": 0.0017064901348730673,
+      "loss": 1.3937,
+      "step": 5929
+    },
+    {
+      "epoch": 0.4561187600953773,
+      "learning_rate": 0.0017061311160933268,
+      "loss": 0.9029,
+      "step": 5930
+    },
+    {
+      "epoch": 0.45619567725559573,
+      "learning_rate": 0.0017057720852773754,
+      "loss": 1.3697,
+      "step": 5931
+    },
+    {
+      "epoch": 0.4562725944158142,
+      "learning_rate": 0.0017054130424461775,
+      "loss": 0.9505,
+      "step": 5932
+    },
+    {
+      "epoch": 0.4563495115760326,
+      "learning_rate": 0.001705053987620698,
+      "loss": 0.9906,
+      "step": 5933
+    },
+    {
+      "epoch": 0.45642642873625106,
+      "learning_rate": 0.0017046949208219028,
+      "loss": 1.4199,
+      "step": 5934
+    },
+    {
+      "epoch": 0.4565033458964695,
+      "learning_rate": 0.0017043358420707568,
+      "loss": 1.0986,
+      "step": 5935
+    },
+    {
+      "epoch": 0.456580263056688,
+      "learning_rate": 0.0017039767513882293,
+      "loss": 1.0399,
+      "step": 5936
+    },
+    {
+      "epoch": 0.4566571802169064,
+      "learning_rate": 0.0017036176487952859,
+      "loss": 1.2513,
+      "step": 5937
+    },
+    {
+      "epoch": 0.45673409737712484,
+      "learning_rate": 0.001703258534312896,
+      "loss": 0.9261,
+      "step": 5938
+    },
+    {
+      "epoch": 0.4568110145373433,
+      "learning_rate": 0.0017028994079620284,
+      "loss": 1.0229,
+      "step": 5939
+    },
+    {
+      "epoch": 0.4568879316975617,
+      "learning_rate": 0.001702540269763653,
+      "loss": 0.8337,
+      "step": 5940
+    },
+    {
+      "epoch": 0.45696484885778016,
+      "learning_rate": 0.0017021811197387403,
+      "loss": 1.044,
+      "step": 5941
+    },
+    {
+      "epoch": 0.4570417660179986,
+      "learning_rate": 0.0017018219579082608,
+      "loss": 1.3212,
+      "step": 5942
+    },
+    {
+      "epoch": 0.4571186831782171,
+      "learning_rate": 0.0017014627842931872,
+      "loss": 1.3879,
+      "step": 5943
+    },
+    {
+      "epoch": 0.4571956003384355,
+      "learning_rate": 0.001701103598914492,
+      "loss": 0.8131,
+      "step": 5944
+    },
+    {
+      "epoch": 0.45727251749865394,
+      "learning_rate": 0.001700744401793148,
+      "loss": 1.0003,
+      "step": 5945
+    },
+    {
+      "epoch": 0.4573494346588724,
+      "learning_rate": 0.0017003851929501283,
+      "loss": 0.9824,
+      "step": 5946
+    },
+    {
+      "epoch": 0.45742635181909086,
+      "learning_rate": 0.0017000259724064093,
+      "loss": 1.1699,
+      "step": 5947
+    },
+    {
+      "epoch": 0.45750326897930926,
+      "learning_rate": 0.001699666740182965,
+      "loss": 1.2166,
+      "step": 5948
+    },
+    {
+      "epoch": 0.4575801861395277,
+      "learning_rate": 0.0016993074963007715,
+      "loss": 1.1913,
+      "step": 5949
+    },
+    {
+      "epoch": 0.4576571032997462,
+      "learning_rate": 0.0016989482407808055,
+      "loss": 1.3081,
+      "step": 5950
+    },
+    {
+      "epoch": 0.45773402045996464,
+      "learning_rate": 0.001698588973644045,
+      "loss": 1.4225,
+      "step": 5951
+    },
+    {
+      "epoch": 0.45781093762018304,
+      "learning_rate": 0.0016982296949114666,
+      "loss": 0.9996,
+      "step": 5952
+    },
+    {
+      "epoch": 0.4578878547804015,
+      "learning_rate": 0.00169787040460405,
+      "loss": 1.4416,
+      "step": 5953
+    },
+    {
+      "epoch": 0.45796477194061996,
+      "learning_rate": 0.0016975111027427744,
+      "loss": 1.1461,
+      "step": 5954
+    },
+    {
+      "epoch": 0.4580416891008384,
+      "learning_rate": 0.0016971517893486193,
+      "loss": 1.0895,
+      "step": 5955
+    },
+    {
+      "epoch": 0.4581186062610568,
+      "learning_rate": 0.0016967924644425666,
+      "loss": 1.5107,
+      "step": 5956
+    },
+    {
+      "epoch": 0.4581955234212753,
+      "learning_rate": 0.001696433128045596,
+      "loss": 1.0817,
+      "step": 5957
+    },
+    {
+      "epoch": 0.45827244058149375,
+      "learning_rate": 0.0016960737801786907,
+      "loss": 1.1311,
+      "step": 5958
+    },
+    {
+      "epoch": 0.45834935774171215,
+      "learning_rate": 0.001695714420862833,
+      "loss": 1.1138,
+      "step": 5959
+    },
+    {
+      "epoch": 0.4584262749019306,
+      "learning_rate": 0.0016953550501190066,
+      "loss": 1.1194,
+      "step": 5960
+    },
+    {
+      "epoch": 0.45850319206214907,
+      "learning_rate": 0.0016949956679681944,
+      "loss": 1.5385,
+      "step": 5961
+    },
+    {
+      "epoch": 0.4585801092223675,
+      "learning_rate": 0.0016946362744313821,
+      "loss": 1.1169,
+      "step": 5962
+    },
+    {
+      "epoch": 0.45865702638258593,
+      "learning_rate": 0.0016942768695295555,
+      "loss": 0.9635,
+      "step": 5963
+    },
+    {
+      "epoch": 0.4587339435428044,
+      "learning_rate": 0.0016939174532836997,
+      "loss": 0.9801,
+      "step": 5964
+    },
+    {
+      "epoch": 0.45881086070302285,
+      "learning_rate": 0.0016935580257148013,
+      "loss": 1.3425,
+      "step": 5965
+    },
+    {
+      "epoch": 0.4588877778632413,
+      "learning_rate": 0.0016931985868438486,
+      "loss": 1.2531,
+      "step": 5966
+    },
+    {
+      "epoch": 0.4589646950234597,
+      "learning_rate": 0.001692839136691829,
+      "loss": 0.8637,
+      "step": 5967
+    },
+    {
+      "epoch": 0.45904161218367817,
+      "learning_rate": 0.001692479675279731,
+      "loss": 1.1803,
+      "step": 5968
+    },
+    {
+      "epoch": 0.45911852934389663,
+      "learning_rate": 0.0016921202026285447,
+      "loss": 1.106,
+      "step": 5969
+    },
+    {
+      "epoch": 0.4591954465041151,
+      "learning_rate": 0.0016917607187592588,
+      "loss": 1.7642,
+      "step": 5970
+    },
+    {
+      "epoch": 0.4592723636643335,
+      "learning_rate": 0.001691401223692865,
+      "loss": 0.9494,
+      "step": 5971
+    },
+    {
+      "epoch": 0.45934928082455195,
+      "learning_rate": 0.0016910417174503544,
+      "loss": 1.2692,
+      "step": 5972
+    },
+    {
+      "epoch": 0.4594261979847704,
+      "learning_rate": 0.0016906822000527185,
+      "loss": 1.1,
+      "step": 5973
+    },
+    {
+      "epoch": 0.45950311514498887,
+      "learning_rate": 0.0016903226715209506,
+      "loss": 1.1767,
+      "step": 5974
+    },
+    {
+      "epoch": 0.4595800323052073,
+      "learning_rate": 0.0016899631318760436,
+      "loss": 1.1886,
+      "step": 5975
+    },
+    {
+      "epoch": 0.45965694946542573,
+      "learning_rate": 0.0016896035811389911,
+      "loss": 1.0692,
+      "step": 5976
+    },
+    {
+      "epoch": 0.4597338666256442,
+      "learning_rate": 0.001689244019330788,
+      "loss": 1.1346,
+      "step": 5977
+    },
+    {
+      "epoch": 0.45981078378586265,
+      "learning_rate": 0.0016888844464724294,
+      "loss": 1.1183,
+      "step": 5978
+    },
+    {
+      "epoch": 0.45988770094608106,
+      "learning_rate": 0.0016885248625849111,
+      "loss": 1.3074,
+      "step": 5979
+    },
+    {
+      "epoch": 0.4599646181062995,
+      "learning_rate": 0.0016881652676892294,
+      "loss": 1.518,
+      "step": 5980
+    },
+    {
+      "epoch": 0.460041535266518,
+      "learning_rate": 0.0016878056618063817,
+      "loss": 1.3334,
+      "step": 5981
+    },
+    {
+      "epoch": 0.4601184524267364,
+      "learning_rate": 0.001687446044957366,
+      "loss": 1.162,
+      "step": 5982
+    },
+    {
+      "epoch": 0.46019536958695484,
+      "learning_rate": 0.00168708641716318,
+      "loss": 1.3031,
+      "step": 5983
+    },
+    {
+      "epoch": 0.4602722867471733,
+      "learning_rate": 0.0016867267784448233,
+      "loss": 1.6319,
+      "step": 5984
+    },
+    {
+      "epoch": 0.46034920390739176,
+      "learning_rate": 0.0016863671288232956,
+      "loss": 1.1516,
+      "step": 5985
+    },
+    {
+      "epoch": 0.46042612106761016,
+      "learning_rate": 0.0016860074683195966,
+      "loss": 1.4387,
+      "step": 5986
+    },
+    {
+      "epoch": 0.4605030382278286,
+      "learning_rate": 0.001685647796954728,
+      "loss": 0.7825,
+      "step": 5987
+    },
+    {
+      "epoch": 0.4605799553880471,
+      "learning_rate": 0.0016852881147496907,
+      "loss": 1.2067,
+      "step": 5988
+    },
+    {
+      "epoch": 0.46065687254826554,
+      "learning_rate": 0.0016849284217254877,
+      "loss": 1.3889,
+      "step": 5989
+    },
+    {
+      "epoch": 0.46073378970848394,
+      "learning_rate": 0.0016845687179031215,
+      "loss": 1.0775,
+      "step": 5990
+    },
+    {
+      "epoch": 0.4608107068687024,
+      "learning_rate": 0.0016842090033035951,
+      "loss": 1.0139,
+      "step": 5991
+    },
+    {
+      "epoch": 0.46088762402892086,
+      "learning_rate": 0.001683849277947913,
+      "loss": 1.4182,
+      "step": 5992
+    },
+    {
+      "epoch": 0.4609645411891393,
+      "learning_rate": 0.00168348954185708,
+      "loss": 1.1822,
+      "step": 5993
+    },
+    {
+      "epoch": 0.4610414583493577,
+      "learning_rate": 0.0016831297950521016,
+      "loss": 1.0535,
+      "step": 5994
+    },
+    {
+      "epoch": 0.4611183755095762,
+      "learning_rate": 0.0016827700375539837,
+      "loss": 1.0001,
+      "step": 5995
+    },
+    {
+      "epoch": 0.46119529266979464,
+      "learning_rate": 0.0016824102693837326,
+      "loss": 1.5293,
+      "step": 5996
+    },
+    {
+      "epoch": 0.4612722098300131,
+      "learning_rate": 0.0016820504905623557,
+      "loss": 1.057,
+      "step": 5997
+    },
+    {
+      "epoch": 0.4613491269902315,
+      "learning_rate": 0.0016816907011108612,
+      "loss": 0.9821,
+      "step": 5998
+    },
+    {
+      "epoch": 0.46142604415044997,
+      "learning_rate": 0.0016813309010502574,
+      "loss": 1.3119,
+      "step": 5999
+    },
+    {
+      "epoch": 0.4615029613106684,
+      "learning_rate": 0.0016809710904015529,
+      "loss": 1.0554,
+      "step": 6000
+    },
+    {
+      "epoch": 0.46157987847088683,
+      "learning_rate": 0.0016806112691857584,
+      "loss": 1.0643,
+      "step": 6001
+    },
+    {
+      "epoch": 0.4616567956311053,
+      "learning_rate": 0.0016802514374238831,
+      "loss": 0.9915,
+      "step": 6002
+    },
+    {
+      "epoch": 0.46173371279132375,
+      "learning_rate": 0.0016798915951369386,
+      "loss": 1.2561,
+      "step": 6003
+    },
+    {
+      "epoch": 0.4618106299515422,
+      "learning_rate": 0.0016795317423459366,
+      "loss": 1.2189,
+      "step": 6004
+    },
+    {
+      "epoch": 0.4618875471117606,
+      "learning_rate": 0.0016791718790718892,
+      "loss": 1.2343,
+      "step": 6005
+    },
+    {
+      "epoch": 0.46196446427197907,
+      "learning_rate": 0.0016788120053358084,
+      "loss": 1.0122,
+      "step": 6006
+    },
+    {
+      "epoch": 0.46204138143219753,
+      "learning_rate": 0.0016784521211587087,
+      "loss": 1.3605,
+      "step": 6007
+    },
+    {
+      "epoch": 0.462118298592416,
+      "learning_rate": 0.0016780922265616034,
+      "loss": 1.2304,
+      "step": 6008
+    },
+    {
+      "epoch": 0.4621952157526344,
+      "learning_rate": 0.001677732321565508,
+      "loss": 0.8988,
+      "step": 6009
+    },
+    {
+      "epoch": 0.46227213291285285,
+      "learning_rate": 0.0016773724061914367,
+      "loss": 1.105,
+      "step": 6010
+    },
+    {
+      "epoch": 0.4623490500730713,
+      "learning_rate": 0.0016770124804604057,
+      "loss": 1.3922,
+      "step": 6011
+    },
+    {
+      "epoch": 0.46242596723328977,
+      "learning_rate": 0.0016766525443934314,
+      "loss": 1.0368,
+      "step": 6012
+    },
+    {
+      "epoch": 0.4625028843935082,
+      "learning_rate": 0.0016762925980115314,
+      "loss": 0.9141,
+      "step": 6013
+    },
+    {
+      "epoch": 0.46257980155372663,
+      "learning_rate": 0.0016759326413357224,
+      "loss": 1.1835,
+      "step": 6014
+    },
+    {
+      "epoch": 0.4626567187139451,
+      "learning_rate": 0.0016755726743870234,
+      "loss": 1.0215,
+      "step": 6015
+    },
+    {
+      "epoch": 0.46273363587416355,
+      "learning_rate": 0.0016752126971864526,
+      "loss": 1.1779,
+      "step": 6016
+    },
+    {
+      "epoch": 0.46281055303438196,
+      "learning_rate": 0.00167485270975503,
+      "loss": 1.2531,
+      "step": 6017
+    },
+    {
+      "epoch": 0.4628874701946004,
+      "learning_rate": 0.0016744927121137756,
+      "loss": 1.2628,
+      "step": 6018
+    },
+    {
+      "epoch": 0.4629643873548189,
+      "learning_rate": 0.0016741327042837097,
+      "loss": 1.3413,
+      "step": 6019
+    },
+    {
+      "epoch": 0.4630413045150373,
+      "learning_rate": 0.0016737726862858543,
+      "loss": 1.256,
+      "step": 6020
+    },
+    {
+      "epoch": 0.46311822167525574,
+      "learning_rate": 0.00167341265814123,
+      "loss": 1.0495,
+      "step": 6021
+    },
+    {
+      "epoch": 0.4631951388354742,
+      "learning_rate": 0.0016730526198708601,
+      "loss": 1.2592,
+      "step": 6022
+    },
+    {
+      "epoch": 0.46327205599569266,
+      "learning_rate": 0.0016726925714957669,
+      "loss": 0.8827,
+      "step": 6023
+    },
+    {
+      "epoch": 0.46334897315591106,
+      "learning_rate": 0.0016723325130369754,
+      "loss": 0.9721,
+      "step": 6024
+    },
+    {
+      "epoch": 0.4634258903161295,
+      "learning_rate": 0.0016719724445155083,
+      "loss": 1.4127,
+      "step": 6025
+    },
+    {
+      "epoch": 0.463502807476348,
+      "learning_rate": 0.0016716123659523908,
+      "loss": 0.8854,
+      "step": 6026
+    },
+    {
+      "epoch": 0.46357972463656644,
+      "learning_rate": 0.0016712522773686485,
+      "loss": 1.5001,
+      "step": 6027
+    },
+    {
+      "epoch": 0.46365664179678484,
+      "learning_rate": 0.0016708921787853073,
+      "loss": 0.7913,
+      "step": 6028
+    },
+    {
+      "epoch": 0.4637335589570033,
+      "learning_rate": 0.001670532070223394,
+      "loss": 1.0606,
+      "step": 6029
+    },
+    {
+      "epoch": 0.46381047611722176,
+      "learning_rate": 0.0016701719517039349,
+      "loss": 1.1232,
+      "step": 6030
+    },
+    {
+      "epoch": 0.4638873932774402,
+      "learning_rate": 0.0016698118232479583,
+      "loss": 1.4346,
+      "step": 6031
+    },
+    {
+      "epoch": 0.4639643104376586,
+      "learning_rate": 0.0016694516848764927,
+      "loss": 1.0104,
+      "step": 6032
+    },
+    {
+      "epoch": 0.4640412275978771,
+      "learning_rate": 0.0016690915366105665,
+      "loss": 1.0202,
+      "step": 6033
+    },
+    {
+      "epoch": 0.46411814475809554,
+      "learning_rate": 0.001668731378471209,
+      "loss": 0.9859,
+      "step": 6034
+    },
+    {
+      "epoch": 0.464195061918314,
+      "learning_rate": 0.0016683712104794506,
+      "loss": 1.0598,
+      "step": 6035
+    },
+    {
+      "epoch": 0.4642719790785324,
+      "learning_rate": 0.0016680110326563217,
+      "loss": 1.3897,
+      "step": 6036
+    },
+    {
+      "epoch": 0.46434889623875086,
+      "learning_rate": 0.001667650845022854,
+      "loss": 1.1436,
+      "step": 6037
+    },
+    {
+      "epoch": 0.4644258133989693,
+      "learning_rate": 0.0016672906476000778,
+      "loss": 1.3955,
+      "step": 6038
+    },
+    {
+      "epoch": 0.4645027305591878,
+      "learning_rate": 0.001666930440409027,
+      "loss": 1.0686,
+      "step": 6039
+    },
+    {
+      "epoch": 0.4645796477194062,
+      "learning_rate": 0.0016665702234707337,
+      "loss": 0.8183,
+      "step": 6040
+    },
+    {
+      "epoch": 0.46465656487962465,
+      "learning_rate": 0.0016662099968062312,
+      "loss": 1.3284,
+      "step": 6041
+    },
+    {
+      "epoch": 0.4647334820398431,
+      "learning_rate": 0.001665849760436554,
+      "loss": 1.5238,
+      "step": 6042
+    },
+    {
+      "epoch": 0.4648103992000615,
+      "learning_rate": 0.0016654895143827367,
+      "loss": 1.0193,
+      "step": 6043
+    },
+    {
+      "epoch": 0.46488731636027997,
+      "learning_rate": 0.001665129258665814,
+      "loss": 1.0677,
+      "step": 6044
+    },
+    {
+      "epoch": 0.4649642335204984,
+      "learning_rate": 0.0016647689933068216,
+      "loss": 1.2368,
+      "step": 6045
+    },
+    {
+      "epoch": 0.4650411506807169,
+      "learning_rate": 0.001664408718326796,
+      "loss": 0.9672,
+      "step": 6046
+    },
+    {
+      "epoch": 0.4651180678409353,
+      "learning_rate": 0.0016640484337467742,
+      "loss": 1.1939,
+      "step": 6047
+    },
+    {
+      "epoch": 0.46519498500115375,
+      "learning_rate": 0.001663688139587793,
+      "loss": 0.9955,
+      "step": 6048
+    },
+    {
+      "epoch": 0.4652719021613722,
+      "learning_rate": 0.001663327835870891,
+      "loss": 0.8805,
+      "step": 6049
+    },
+    {
+      "epoch": 0.46534881932159067,
+      "learning_rate": 0.001662967522617106,
+      "loss": 1.0091,
+      "step": 6050
+    },
+    {
+      "epoch": 0.46542573648180907,
+      "learning_rate": 0.0016626071998474786,
+      "loss": 0.996,
+      "step": 6051
+    },
+    {
+      "epoch": 0.46550265364202753,
+      "learning_rate": 0.0016622468675830462,
+      "loss": 1.1809,
+      "step": 6052
+    },
+    {
+      "epoch": 0.465579570802246,
+      "learning_rate": 0.001661886525844851,
+      "loss": 1.4364,
+      "step": 6053
+    },
+    {
+      "epoch": 0.46565648796246445,
+      "learning_rate": 0.0016615261746539321,
+      "loss": 1.4154,
+      "step": 6054
+    },
+    {
+      "epoch": 0.46573340512268285,
+      "learning_rate": 0.0016611658140313324,
+      "loss": 1.246,
+      "step": 6055
+    },
+    {
+      "epoch": 0.4658103222829013,
+      "learning_rate": 0.001660805443998092,
+      "loss": 0.8958,
+      "step": 6056
+    },
+    {
+      "epoch": 0.46588723944311977,
+      "learning_rate": 0.0016604450645752544,
+      "loss": 1.1856,
+      "step": 6057
+    },
+    {
+      "epoch": 0.46596415660333823,
+      "learning_rate": 0.001660084675783862,
+      "loss": 1.2212,
+      "step": 6058
+    },
+    {
+      "epoch": 0.46604107376355663,
+      "learning_rate": 0.0016597242776449591,
+      "loss": 1.1431,
+      "step": 6059
+    },
+    {
+      "epoch": 0.4661179909237751,
+      "learning_rate": 0.001659363870179589,
+      "loss": 1.0644,
+      "step": 6060
+    },
+    {
+      "epoch": 0.46619490808399355,
+      "learning_rate": 0.001659003453408796,
+      "loss": 1.1554,
+      "step": 6061
+    },
+    {
+      "epoch": 0.46627182524421196,
+      "learning_rate": 0.0016586430273536265,
+      "loss": 0.9181,
+      "step": 6062
+    },
+    {
+      "epoch": 0.4663487424044304,
+      "learning_rate": 0.0016582825920351246,
+      "loss": 0.976,
+      "step": 6063
+    },
+    {
+      "epoch": 0.4664256595646489,
+      "learning_rate": 0.0016579221474743375,
+      "loss": 1.2245,
+      "step": 6064
+    },
+    {
+      "epoch": 0.46650257672486733,
+      "learning_rate": 0.0016575616936923114,
+      "loss": 1.2207,
+      "step": 6065
+    },
+    {
+      "epoch": 0.46657949388508574,
+      "learning_rate": 0.0016572012307100941,
+      "loss": 1.3792,
+      "step": 6066
+    },
+    {
+      "epoch": 0.4666564110453042,
+      "learning_rate": 0.0016568407585487329,
+      "loss": 1.3489,
+      "step": 6067
+    },
+    {
+      "epoch": 0.46673332820552266,
+      "learning_rate": 0.0016564802772292765,
+      "loss": 1.539,
+      "step": 6068
+    },
+    {
+      "epoch": 0.4668102453657411,
+      "learning_rate": 0.0016561197867727734,
+      "loss": 1.4299,
+      "step": 6069
+    },
+    {
+      "epoch": 0.4668871625259595,
+      "learning_rate": 0.0016557592872002732,
+      "loss": 0.9731,
+      "step": 6070
+    },
+    {
+      "epoch": 0.466964079686178,
+      "learning_rate": 0.001655398778532826,
+      "loss": 1.2043,
+      "step": 6071
+    },
+    {
+      "epoch": 0.46704099684639644,
+      "learning_rate": 0.0016550382607914823,
+      "loss": 0.9383,
+      "step": 6072
+    },
+    {
+      "epoch": 0.4671179140066149,
+      "learning_rate": 0.001654677733997293,
+      "loss": 1.1071,
+      "step": 6073
+    },
+    {
+      "epoch": 0.4671948311668333,
+      "learning_rate": 0.0016543171981713094,
+      "loss": 0.7593,
+      "step": 6074
+    },
+    {
+      "epoch": 0.46727174832705176,
+      "learning_rate": 0.001653956653334584,
+      "loss": 1.056,
+      "step": 6075
+    },
+    {
+      "epoch": 0.4673486654872702,
+      "learning_rate": 0.0016535960995081689,
+      "loss": 0.9245,
+      "step": 6076
+    },
+    {
+      "epoch": 0.4674255826474887,
+      "learning_rate": 0.0016532355367131176,
+      "loss": 1.3049,
+      "step": 6077
+    },
+    {
+      "epoch": 0.4675024998077071,
+      "learning_rate": 0.001652874964970484,
+      "loss": 1.1381,
+      "step": 6078
+    },
+    {
+      "epoch": 0.46757941696792554,
+      "learning_rate": 0.0016525143843013216,
+      "loss": 0.9822,
+      "step": 6079
+    },
+    {
+      "epoch": 0.467656334128144,
+      "learning_rate": 0.0016521537947266852,
+      "loss": 1.0136,
+      "step": 6080
+    },
+    {
+      "epoch": 0.4677332512883624,
+      "learning_rate": 0.0016517931962676301,
+      "loss": 1.2798,
+      "step": 6081
+    },
+    {
+      "epoch": 0.46781016844858087,
+      "learning_rate": 0.0016514325889452125,
+      "loss": 1.2401,
+      "step": 6082
+    },
+    {
+      "epoch": 0.4678870856087993,
+      "learning_rate": 0.0016510719727804875,
+      "loss": 1.2683,
+      "step": 6083
+    },
+    {
+      "epoch": 0.4679640027690178,
+      "learning_rate": 0.0016507113477945133,
+      "loss": 1.091,
+      "step": 6084
+    },
+    {
+      "epoch": 0.4680409199292362,
+      "learning_rate": 0.001650350714008346,
+      "loss": 0.9794,
+      "step": 6085
+    },
+    {
+      "epoch": 0.46811783708945465,
+      "learning_rate": 0.001649990071443044,
+      "loss": 1.0378,
+      "step": 6086
+    },
+    {
+      "epoch": 0.4681947542496731,
+      "learning_rate": 0.001649629420119665,
+      "loss": 1.0775,
+      "step": 6087
+    },
+    {
+      "epoch": 0.46827167140989157,
+      "learning_rate": 0.0016492687600592685,
+      "loss": 1.3136,
+      "step": 6088
+    },
+    {
+      "epoch": 0.46834858857010997,
+      "learning_rate": 0.0016489080912829139,
+      "loss": 1.2109,
+      "step": 6089
+    },
+    {
+      "epoch": 0.46842550573032843,
+      "learning_rate": 0.0016485474138116605,
+      "loss": 1.06,
+      "step": 6090
+    },
+    {
+      "epoch": 0.4685024228905469,
+      "learning_rate": 0.0016481867276665683,
+      "loss": 1.2542,
+      "step": 6091
+    },
+    {
+      "epoch": 0.46857934005076535,
+      "learning_rate": 0.001647826032868699,
+      "loss": 1.1768,
+      "step": 6092
+    },
+    {
+      "epoch": 0.46865625721098375,
+      "learning_rate": 0.0016474653294391139,
+      "loss": 1.1091,
+      "step": 6093
+    },
+    {
+      "epoch": 0.4687331743712022,
+      "learning_rate": 0.0016471046173988737,
+      "loss": 1.4344,
+      "step": 6094
+    },
+    {
+      "epoch": 0.46881009153142067,
+      "learning_rate": 0.0016467438967690423,
+      "loss": 0.8547,
+      "step": 6095
+    },
+    {
+      "epoch": 0.46888700869163913,
+      "learning_rate": 0.0016463831675706815,
+      "loss": 1.3919,
+      "step": 6096
+    },
+    {
+      "epoch": 0.46896392585185753,
+      "learning_rate": 0.0016460224298248554,
+      "loss": 1.4495,
+      "step": 6097
+    },
+    {
+      "epoch": 0.469040843012076,
+      "learning_rate": 0.001645661683552627,
+      "loss": 1.2432,
+      "step": 6098
+    },
+    {
+      "epoch": 0.46911776017229445,
+      "learning_rate": 0.0016453009287750611,
+      "loss": 0.7972,
+      "step": 6099
+    },
+    {
+      "epoch": 0.4691946773325129,
+      "learning_rate": 0.0016449401655132224,
+      "loss": 1.3865,
+      "step": 6100
+    },
+    {
+      "epoch": 0.4692715944927313,
+      "learning_rate": 0.0016445793937881772,
+      "loss": 1.1304,
+      "step": 6101
+    },
+    {
+      "epoch": 0.4693485116529498,
+      "learning_rate": 0.0016442186136209898,
+      "loss": 0.9214,
+      "step": 6102
+    },
+    {
+      "epoch": 0.46942542881316823,
+      "learning_rate": 0.0016438578250327276,
+      "loss": 1.3439,
+      "step": 6103
+    },
+    {
+      "epoch": 0.46950234597338664,
+      "learning_rate": 0.0016434970280444568,
+      "loss": 1.0002,
+      "step": 6104
+    },
+    {
+      "epoch": 0.4695792631336051,
+      "learning_rate": 0.001643136222677245,
+      "loss": 1.0149,
+      "step": 6105
+    },
+    {
+      "epoch": 0.46965618029382356,
+      "learning_rate": 0.0016427754089521603,
+      "loss": 1.0766,
+      "step": 6106
+    },
+    {
+      "epoch": 0.469733097454042,
+      "learning_rate": 0.00164241458689027,
+      "loss": 0.9845,
+      "step": 6107
+    },
+    {
+      "epoch": 0.4698100146142604,
+      "learning_rate": 0.0016420537565126443,
+      "loss": 1.3217,
+      "step": 6108
+    },
+    {
+      "epoch": 0.4698869317744789,
+      "learning_rate": 0.0016416929178403514,
+      "loss": 1.1593,
+      "step": 6109
+    },
+    {
+      "epoch": 0.46996384893469734,
+      "learning_rate": 0.0016413320708944612,
+      "loss": 1.4989,
+      "step": 6110
+    },
+    {
+      "epoch": 0.4700407660949158,
+      "learning_rate": 0.001640971215696044,
+      "loss": 0.9828,
+      "step": 6111
+    },
+    {
+      "epoch": 0.4701176832551342,
+      "learning_rate": 0.0016406103522661711,
+      "loss": 1.1914,
+      "step": 6112
+    },
+    {
+      "epoch": 0.47019460041535266,
+      "learning_rate": 0.001640249480625913,
+      "loss": 0.7579,
+      "step": 6113
+    },
+    {
+      "epoch": 0.4702715175755711,
+      "learning_rate": 0.0016398886007963416,
+      "loss": 0.8694,
+      "step": 6114
+    },
+    {
+      "epoch": 0.4703484347357896,
+      "learning_rate": 0.0016395277127985285,
+      "loss": 0.8679,
+      "step": 6115
+    },
+    {
+      "epoch": 0.470425351896008,
+      "learning_rate": 0.0016391668166535474,
+      "loss": 0.9447,
+      "step": 6116
+    },
+    {
+      "epoch": 0.47050226905622644,
+      "learning_rate": 0.001638805912382471,
+      "loss": 1.3569,
+      "step": 6117
+    },
+    {
+      "epoch": 0.4705791862164449,
+      "learning_rate": 0.001638445000006372,
+      "loss": 1.097,
+      "step": 6118
+    },
+    {
+      "epoch": 0.47065610337666336,
+      "learning_rate": 0.0016380840795463258,
+      "loss": 1.1325,
+      "step": 6119
+    },
+    {
+      "epoch": 0.47073302053688176,
+      "learning_rate": 0.0016377231510234062,
+      "loss": 1.335,
+      "step": 6120
+    },
+    {
+      "epoch": 0.4708099376971002,
+      "learning_rate": 0.001637362214458688,
+      "loss": 1.3611,
+      "step": 6121
+    },
+    {
+      "epoch": 0.4708868548573187,
+      "learning_rate": 0.001637001269873247,
+      "loss": 1.2211,
+      "step": 6122
+    },
+    {
+      "epoch": 0.4709637720175371,
+      "learning_rate": 0.0016366403172881593,
+      "loss": 1.0244,
+      "step": 6123
+    },
+    {
+      "epoch": 0.47104068917775554,
+      "learning_rate": 0.001636279356724501,
+      "loss": 1.1608,
+      "step": 6124
+    },
+    {
+      "epoch": 0.471117606337974,
+      "learning_rate": 0.0016359183882033487,
+      "loss": 1.3197,
+      "step": 6125
+    },
+    {
+      "epoch": 0.47119452349819246,
+      "learning_rate": 0.0016355574117457802,
+      "loss": 1.0408,
+      "step": 6126
+    },
+    {
+      "epoch": 0.47127144065841087,
+      "learning_rate": 0.0016351964273728736,
+      "loss": 0.9261,
+      "step": 6127
+    },
+    {
+      "epoch": 0.4713483578186293,
+      "learning_rate": 0.0016348354351057065,
+      "loss": 1.2108,
+      "step": 6128
+    },
+    {
+      "epoch": 0.4714252749788478,
+      "learning_rate": 0.0016344744349653573,
+      "loss": 1.2655,
+      "step": 6129
+    },
+    {
+      "epoch": 0.47150219213906625,
+      "learning_rate": 0.001634113426972906,
+      "loss": 0.8518,
+      "step": 6130
+    },
+    {
+      "epoch": 0.47157910929928465,
+      "learning_rate": 0.0016337524111494325,
+      "loss": 1.2985,
+      "step": 6131
+    },
+    {
+      "epoch": 0.4716560264595031,
+      "learning_rate": 0.0016333913875160157,
+      "loss": 1.1613,
+      "step": 6132
+    },
+    {
+      "epoch": 0.47173294361972157,
+      "learning_rate": 0.0016330303560937367,
+      "loss": 0.7984,
+      "step": 6133
+    },
+    {
+      "epoch": 0.47180986077994,
+      "learning_rate": 0.0016326693169036771,
+      "loss": 1.2623,
+      "step": 6134
+    },
+    {
+      "epoch": 0.47188677794015843,
+      "learning_rate": 0.0016323082699669174,
+      "loss": 0.7766,
+      "step": 6135
+    },
+    {
+      "epoch": 0.4719636951003769,
+      "learning_rate": 0.00163194721530454,
+      "loss": 1.1435,
+      "step": 6136
+    },
+    {
+      "epoch": 0.47204061226059535,
+      "learning_rate": 0.0016315861529376272,
+      "loss": 1.1985,
+      "step": 6137
+    },
+    {
+      "epoch": 0.4721175294208138,
+      "learning_rate": 0.0016312250828872615,
+      "loss": 1.299,
+      "step": 6138
+    },
+    {
+      "epoch": 0.4721944465810322,
+      "learning_rate": 0.0016308640051745274,
+      "loss": 1.5076,
+      "step": 6139
+    },
+    {
+      "epoch": 0.47227136374125067,
+      "learning_rate": 0.0016305029198205067,
+      "loss": 1.1281,
+      "step": 6140
+    },
+    {
+      "epoch": 0.47234828090146913,
+      "learning_rate": 0.001630141826846285,
+      "loss": 1.2295,
+      "step": 6141
+    },
+    {
+      "epoch": 0.4724251980616876,
+      "learning_rate": 0.0016297807262729458,
+      "loss": 1.2271,
+      "step": 6142
+    },
+    {
+      "epoch": 0.472502115221906,
+      "learning_rate": 0.0016294196181215756,
+      "loss": 1.1294,
+      "step": 6143
+    },
+    {
+      "epoch": 0.47257903238212445,
+      "learning_rate": 0.0016290585024132582,
+      "loss": 1.0136,
+      "step": 6144
+    },
+    {
+      "epoch": 0.4726559495423429,
+      "learning_rate": 0.0016286973791690808,
+      "loss": 1.2991,
+      "step": 6145
+    },
+    {
+      "epoch": 0.4727328667025613,
+      "learning_rate": 0.0016283362484101295,
+      "loss": 1.1902,
+      "step": 6146
+    },
+    {
+      "epoch": 0.4728097838627798,
+      "learning_rate": 0.0016279751101574902,
+      "loss": 1.0617,
+      "step": 6147
+    },
+    {
+      "epoch": 0.47288670102299823,
+      "learning_rate": 0.0016276139644322515,
+      "loss": 1.0008,
+      "step": 6148
+    },
+    {
+      "epoch": 0.4729636181832167,
+      "learning_rate": 0.0016272528112555,
+      "loss": 0.99,
+      "step": 6149
+    },
+    {
+      "epoch": 0.4730405353434351,
+      "learning_rate": 0.0016268916506483246,
+      "loss": 1.2492,
+      "step": 6150
+    },
+    {
+      "epoch": 0.47311745250365356,
+      "learning_rate": 0.0016265304826318134,
+      "loss": 1.0925,
+      "step": 6151
+    },
+    {
+      "epoch": 0.473194369663872,
+      "learning_rate": 0.0016261693072270552,
+      "loss": 1.3133,
+      "step": 6152
+    },
+    {
+      "epoch": 0.4732712868240905,
+      "learning_rate": 0.0016258081244551398,
+      "loss": 1.3241,
+      "step": 6153
+    },
+    {
+      "epoch": 0.4733482039843089,
+      "learning_rate": 0.0016254469343371573,
+      "loss": 1.0899,
+      "step": 6154
+    },
+    {
+      "epoch": 0.47342512114452734,
+      "learning_rate": 0.0016250857368941968,
+      "loss": 1.1011,
+      "step": 6155
+    },
+    {
+      "epoch": 0.4735020383047458,
+      "learning_rate": 0.0016247245321473504,
+      "loss": 1.2659,
+      "step": 6156
+    },
+    {
+      "epoch": 0.47357895546496426,
+      "learning_rate": 0.0016243633201177081,
+      "loss": 1.2514,
+      "step": 6157
+    },
+    {
+      "epoch": 0.47365587262518266,
+      "learning_rate": 0.0016240021008263623,
+      "loss": 1.1454,
+      "step": 6158
+    },
+    {
+      "epoch": 0.4737327897854011,
+      "learning_rate": 0.0016236408742944046,
+      "loss": 0.8226,
+      "step": 6159
+    },
+    {
+      "epoch": 0.4738097069456196,
+      "learning_rate": 0.001623279640542927,
+      "loss": 0.9157,
+      "step": 6160
+    },
+    {
+      "epoch": 0.47388662410583804,
+      "learning_rate": 0.001622918399593023,
+      "loss": 0.8018,
+      "step": 6161
+    },
+    {
+      "epoch": 0.47396354126605644,
+      "learning_rate": 0.0016225571514657854,
+      "loss": 0.9245,
+      "step": 6162
+    },
+    {
+      "epoch": 0.4740404584262749,
+      "learning_rate": 0.0016221958961823082,
+      "loss": 0.8069,
+      "step": 6163
+    },
+    {
+      "epoch": 0.47411737558649336,
+      "learning_rate": 0.001621834633763685,
+      "loss": 1.6641,
+      "step": 6164
+    },
+    {
+      "epoch": 0.47419429274671177,
+      "learning_rate": 0.0016214733642310112,
+      "loss": 1.115,
+      "step": 6165
+    },
+    {
+      "epoch": 0.4742712099069302,
+      "learning_rate": 0.0016211120876053803,
+      "loss": 0.8321,
+      "step": 6166
+    },
+    {
+      "epoch": 0.4743481270671487,
+      "learning_rate": 0.001620750803907889,
+      "loss": 1.2554,
+      "step": 6167
+    },
+    {
+      "epoch": 0.47442504422736714,
+      "learning_rate": 0.0016203895131596316,
+      "loss": 1.6233,
+      "step": 6168
+    },
+    {
+      "epoch": 0.47450196138758555,
+      "learning_rate": 0.0016200282153817059,
+      "loss": 1.4526,
+      "step": 6169
+    },
+    {
+      "epoch": 0.474578878547804,
+      "learning_rate": 0.0016196669105952074,
+      "loss": 1.1712,
+      "step": 6170
+    },
+    {
+      "epoch": 0.47465579570802247,
+      "learning_rate": 0.0016193055988212328,
+      "loss": 0.8857,
+      "step": 6171
+    },
+    {
+      "epoch": 0.4747327128682409,
+      "learning_rate": 0.0016189442800808806,
+      "loss": 1.0663,
+      "step": 6172
+    },
+    {
+      "epoch": 0.47480963002845933,
+      "learning_rate": 0.0016185829543952476,
+      "loss": 1.2616,
+      "step": 6173
+    },
+    {
+      "epoch": 0.4748865471886778,
+      "learning_rate": 0.0016182216217854324,
+      "loss": 0.761,
+      "step": 6174
+    },
+    {
+      "epoch": 0.47496346434889625,
+      "learning_rate": 0.001617860282272533,
+      "loss": 1.0008,
+      "step": 6175
+    },
+    {
+      "epoch": 0.4750403815091147,
+      "learning_rate": 0.0016174989358776494,
+      "loss": 1.1151,
+      "step": 6176
+    },
+    {
+      "epoch": 0.4751172986693331,
+      "learning_rate": 0.0016171375826218807,
+      "loss": 1.0568,
+      "step": 6177
+    },
+    {
+      "epoch": 0.47519421582955157,
+      "learning_rate": 0.0016167762225263262,
+      "loss": 1.0713,
+      "step": 6178
+    },
+    {
+      "epoch": 0.47527113298977003,
+      "learning_rate": 0.0016164148556120864,
+      "loss": 1.195,
+      "step": 6179
+    },
+    {
+      "epoch": 0.4753480501499885,
+      "learning_rate": 0.0016160534819002618,
+      "loss": 1.4284,
+      "step": 6180
+    },
+    {
+      "epoch": 0.4754249673102069,
+      "learning_rate": 0.0016156921014119532,
+      "loss": 1.3352,
+      "step": 6181
+    },
+    {
+      "epoch": 0.47550188447042535,
+      "learning_rate": 0.0016153307141682624,
+      "loss": 1.26,
+      "step": 6182
+    },
+    {
+      "epoch": 0.4755788016306438,
+      "learning_rate": 0.0016149693201902913,
+      "loss": 1.2073,
+      "step": 6183
+    },
+    {
+      "epoch": 0.4756557187908622,
+      "learning_rate": 0.0016146079194991416,
+      "loss": 1.2384,
+      "step": 6184
+    },
+    {
+      "epoch": 0.4757326359510807,
+      "learning_rate": 0.0016142465121159158,
+      "loss": 1.0126,
+      "step": 6185
+    },
+    {
+      "epoch": 0.47580955311129913,
+      "learning_rate": 0.0016138850980617172,
+      "loss": 1.0965,
+      "step": 6186
+    },
+    {
+      "epoch": 0.4758864702715176,
+      "learning_rate": 0.0016135236773576493,
+      "loss": 1.1451,
+      "step": 6187
+    },
+    {
+      "epoch": 0.475963387431736,
+      "learning_rate": 0.0016131622500248153,
+      "loss": 1.4201,
+      "step": 6188
+    },
+    {
+      "epoch": 0.47604030459195446,
+      "learning_rate": 0.00161280081608432,
+      "loss": 1.0825,
+      "step": 6189
+    },
+    {
+      "epoch": 0.4761172217521729,
+      "learning_rate": 0.0016124393755572667,
+      "loss": 1.3944,
+      "step": 6190
+    },
+    {
+      "epoch": 0.4761941389123914,
+      "learning_rate": 0.0016120779284647618,
+      "loss": 1.0913,
+      "step": 6191
+    },
+    {
+      "epoch": 0.4762710560726098,
+      "learning_rate": 0.0016117164748279096,
+      "loss": 1.2932,
+      "step": 6192
+    },
+    {
+      "epoch": 0.47634797323282824,
+      "learning_rate": 0.0016113550146678157,
+      "loss": 1.6216,
+      "step": 6193
+    },
+    {
+      "epoch": 0.4764248903930467,
+      "learning_rate": 0.001610993548005587,
+      "loss": 1.0621,
+      "step": 6194
+    },
+    {
+      "epoch": 0.47650180755326516,
+      "learning_rate": 0.0016106320748623287,
+      "loss": 1.107,
+      "step": 6195
+    },
+    {
+      "epoch": 0.47657872471348356,
+      "learning_rate": 0.0016102705952591493,
+      "loss": 1.1866,
+      "step": 6196
+    },
+    {
+      "epoch": 0.476655641873702,
+      "learning_rate": 0.0016099091092171538,
+      "loss": 1.0801,
+      "step": 6197
+    },
+    {
+      "epoch": 0.4767325590339205,
+      "learning_rate": 0.0016095476167574516,
+      "loss": 1.7075,
+      "step": 6198
+    },
+    {
+      "epoch": 0.47680947619413894,
+      "learning_rate": 0.0016091861179011492,
+      "loss": 1.1463,
+      "step": 6199
+    },
+    {
+      "epoch": 0.47688639335435734,
+      "learning_rate": 0.0016088246126693562,
+      "loss": 1.0707,
+      "step": 6200
+    },
+    {
+      "epoch": 0.4769633105145758,
+      "learning_rate": 0.0016084631010831808,
+      "loss": 1.2336,
+      "step": 6201
+    },
+    {
+      "epoch": 0.47704022767479426,
+      "learning_rate": 0.0016081015831637313,
+      "loss": 1.0221,
+      "step": 6202
+    },
+    {
+      "epoch": 0.4771171448350127,
+      "learning_rate": 0.001607740058932118,
+      "loss": 1.0629,
+      "step": 6203
+    },
+    {
+      "epoch": 0.4771940619952311,
+      "learning_rate": 0.0016073785284094503,
+      "loss": 1.1565,
+      "step": 6204
+    },
+    {
+      "epoch": 0.4772709791554496,
+      "learning_rate": 0.0016070169916168385,
+      "loss": 0.7952,
+      "step": 6205
+    },
+    {
+      "epoch": 0.47734789631566804,
+      "learning_rate": 0.0016066554485753928,
+      "loss": 1.104,
+      "step": 6206
+    },
+    {
+      "epoch": 0.47742481347588644,
+      "learning_rate": 0.001606293899306225,
+      "loss": 0.8004,
+      "step": 6207
+    },
+    {
+      "epoch": 0.4775017306361049,
+      "learning_rate": 0.001605932343830445,
+      "loss": 1.6401,
+      "step": 6208
+    },
+    {
+      "epoch": 0.47757864779632336,
+      "learning_rate": 0.001605570782169165,
+      "loss": 1.1741,
+      "step": 6209
+    },
+    {
+      "epoch": 0.4776555649565418,
+      "learning_rate": 0.001605209214343497,
+      "loss": 1.1576,
+      "step": 6210
+    },
+    {
+      "epoch": 0.4777324821167602,
+      "learning_rate": 0.0016048476403745537,
+      "loss": 0.8853,
+      "step": 6211
+    },
+    {
+      "epoch": 0.4778093992769787,
+      "learning_rate": 0.001604486060283447,
+      "loss": 1.0119,
+      "step": 6212
+    },
+    {
+      "epoch": 0.47788631643719715,
+      "learning_rate": 0.0016041244740912905,
+      "loss": 1.1708,
+      "step": 6213
+    },
+    {
+      "epoch": 0.4779632335974156,
+      "learning_rate": 0.0016037628818191976,
+      "loss": 1.12,
+      "step": 6214
+    },
+    {
+      "epoch": 0.478040150757634,
+      "learning_rate": 0.0016034012834882819,
+      "loss": 1.8501,
+      "step": 6215
+    },
+    {
+      "epoch": 0.47811706791785247,
+      "learning_rate": 0.0016030396791196573,
+      "loss": 0.99,
+      "step": 6216
+    },
+    {
+      "epoch": 0.4781939850780709,
+      "learning_rate": 0.0016026780687344385,
+      "loss": 1.0905,
+      "step": 6217
+    },
+    {
+      "epoch": 0.4782709022382894,
+      "learning_rate": 0.0016023164523537405,
+      "loss": 0.7269,
+      "step": 6218
+    },
+    {
+      "epoch": 0.4783478193985078,
+      "learning_rate": 0.0016019548299986781,
+      "loss": 1.0102,
+      "step": 6219
+    },
+    {
+      "epoch": 0.47842473655872625,
+      "learning_rate": 0.001601593201690367,
+      "loss": 1.033,
+      "step": 6220
+    },
+    {
+      "epoch": 0.4785016537189447,
+      "learning_rate": 0.0016012315674499222,
+      "loss": 1.0176,
+      "step": 6221
+    },
+    {
+      "epoch": 0.47857857087916317,
+      "learning_rate": 0.0016008699272984615,
+      "loss": 1.1314,
+      "step": 6222
+    },
+    {
+      "epoch": 0.47865548803938157,
+      "learning_rate": 0.0016005082812571008,
+      "loss": 1.1402,
+      "step": 6223
+    },
+    {
+      "epoch": 0.47873240519960003,
+      "learning_rate": 0.001600146629346956,
+      "loss": 1.4318,
+      "step": 6224
+    },
+    {
+      "epoch": 0.4788093223598185,
+      "learning_rate": 0.0015997849715891457,
+      "loss": 0.8946,
+      "step": 6225
+    },
+    {
+      "epoch": 0.4788862395200369,
+      "learning_rate": 0.0015994233080047866,
+      "loss": 1.1351,
+      "step": 6226
+    },
+    {
+      "epoch": 0.47896315668025535,
+      "learning_rate": 0.0015990616386149972,
+      "loss": 1.3198,
+      "step": 6227
+    },
+    {
+      "epoch": 0.4790400738404738,
+      "learning_rate": 0.0015986999634408955,
+      "loss": 1.1457,
+      "step": 6228
+    },
+    {
+      "epoch": 0.47911699100069227,
+      "learning_rate": 0.0015983382825035997,
+      "loss": 1.274,
+      "step": 6229
+    },
+    {
+      "epoch": 0.4791939081609107,
+      "learning_rate": 0.0015979765958242294,
+      "loss": 1.2721,
+      "step": 6230
+    },
+    {
+      "epoch": 0.47927082532112913,
+      "learning_rate": 0.0015976149034239039,
+      "loss": 1.0324,
+      "step": 6231
+    },
+    {
+      "epoch": 0.4793477424813476,
+      "learning_rate": 0.0015972532053237418,
+      "loss": 1.294,
+      "step": 6232
+    },
+    {
+      "epoch": 0.47942465964156605,
+      "learning_rate": 0.001596891501544864,
+      "loss": 1.0118,
+      "step": 6233
+    },
+    {
+      "epoch": 0.47950157680178446,
+      "learning_rate": 0.0015965297921083903,
+      "loss": 0.8374,
+      "step": 6234
+    },
+    {
+      "epoch": 0.4795784939620029,
+      "learning_rate": 0.0015961680770354416,
+      "loss": 1.227,
+      "step": 6235
+    },
+    {
+      "epoch": 0.4796554111222214,
+      "learning_rate": 0.0015958063563471384,
+      "loss": 1.3112,
+      "step": 6236
+    },
+    {
+      "epoch": 0.47973232828243983,
+      "learning_rate": 0.0015954446300646026,
+      "loss": 1.1344,
+      "step": 6237
+    },
+    {
+      "epoch": 0.47980924544265824,
+      "learning_rate": 0.0015950828982089552,
+      "loss": 0.9309,
+      "step": 6238
+    },
+    {
+      "epoch": 0.4798861626028767,
+      "learning_rate": 0.0015947211608013181,
+      "loss": 0.9173,
+      "step": 6239
+    },
+    {
+      "epoch": 0.47996307976309516,
+      "learning_rate": 0.0015943594178628143,
+      "loss": 1.324,
+      "step": 6240
+    },
+    {
+      "epoch": 0.4800399969233136,
+      "learning_rate": 0.0015939976694145648,
+      "loss": 1.3338,
+      "step": 6241
+    },
+    {
+      "epoch": 0.480116914083532,
+      "learning_rate": 0.0015936359154776942,
+      "loss": 1.3014,
+      "step": 6242
+    },
+    {
+      "epoch": 0.4801938312437505,
+      "learning_rate": 0.0015932741560733246,
+      "loss": 0.8919,
+      "step": 6243
+    },
+    {
+      "epoch": 0.48027074840396894,
+      "learning_rate": 0.00159291239122258,
+      "loss": 1.0611,
+      "step": 6244
+    },
+    {
+      "epoch": 0.48034766556418734,
+      "learning_rate": 0.001592550620946584,
+      "loss": 0.802,
+      "step": 6245
+    },
+    {
+      "epoch": 0.4804245827244058,
+      "learning_rate": 0.0015921888452664604,
+      "loss": 1.552,
+      "step": 6246
+    },
+    {
+      "epoch": 0.48050149988462426,
+      "learning_rate": 0.0015918270642033346,
+      "loss": 1.3838,
+      "step": 6247
+    },
+    {
+      "epoch": 0.4805784170448427,
+      "learning_rate": 0.0015914652777783303,
+      "loss": 1.3279,
+      "step": 6248
+    },
+    {
+      "epoch": 0.4806553342050611,
+      "learning_rate": 0.001591103486012574,
+      "loss": 0.9805,
+      "step": 6249
+    },
+    {
+      "epoch": 0.4807322513652796,
+      "learning_rate": 0.001590741688927189,
+      "loss": 1.1158,
+      "step": 6250
+    },
+    {
+      "epoch": 0.48080916852549804,
+      "learning_rate": 0.0015903798865433032,
+      "loss": 1.1117,
+      "step": 6251
+    },
+    {
+      "epoch": 0.4808860856857165,
+      "learning_rate": 0.0015900180788820408,
+      "loss": 1.1248,
+      "step": 6252
+    },
+    {
+      "epoch": 0.4809630028459349,
+      "learning_rate": 0.0015896562659645298,
+      "loss": 0.9676,
+      "step": 6253
+    },
+    {
+      "epoch": 0.48103992000615337,
+      "learning_rate": 0.0015892944478118956,
+      "loss": 1.0328,
+      "step": 6254
+    },
+    {
+      "epoch": 0.4811168371663718,
+      "learning_rate": 0.0015889326244452656,
+      "loss": 1.314,
+      "step": 6255
+    },
+    {
+      "epoch": 0.4811937543265903,
+      "learning_rate": 0.0015885707958857667,
+      "loss": 1.1711,
+      "step": 6256
+    },
+    {
+      "epoch": 0.4812706714868087,
+      "learning_rate": 0.0015882089621545269,
+      "loss": 1.5342,
+      "step": 6257
+    },
+    {
+      "epoch": 0.48134758864702715,
+      "learning_rate": 0.001587847123272674,
+      "loss": 1.2094,
+      "step": 6258
+    },
+    {
+      "epoch": 0.4814245058072456,
+      "learning_rate": 0.0015874852792613357,
+      "loss": 0.4793,
+      "step": 6259
+    },
+    {
+      "epoch": 0.48150142296746407,
+      "learning_rate": 0.001587123430141641,
+      "loss": 1.3869,
+      "step": 6260
+    },
+    {
+      "epoch": 0.48157834012768247,
+      "learning_rate": 0.0015867615759347187,
+      "loss": 1.2889,
+      "step": 6261
+    },
+    {
+      "epoch": 0.48165525728790093,
+      "learning_rate": 0.001586399716661697,
+      "loss": 1.0959,
+      "step": 6262
+    },
+    {
+      "epoch": 0.4817321744481194,
+      "learning_rate": 0.001586037852343706,
+      "loss": 1.1597,
+      "step": 6263
+    },
+    {
+      "epoch": 0.48180909160833785,
+      "learning_rate": 0.0015856759830018754,
+      "loss": 0.8337,
+      "step": 6264
+    },
+    {
+      "epoch": 0.48188600876855625,
+      "learning_rate": 0.0015853141086573346,
+      "loss": 1.0926,
+      "step": 6265
+    },
+    {
+      "epoch": 0.4819629259287747,
+      "learning_rate": 0.0015849522293312143,
+      "loss": 0.8937,
+      "step": 6266
+    },
+    {
+      "epoch": 0.48203984308899317,
+      "learning_rate": 0.0015845903450446442,
+      "loss": 1.1546,
+      "step": 6267
+    },
+    {
+      "epoch": 0.4821167602492116,
+      "learning_rate": 0.0015842284558187565,
+      "loss": 0.8617,
+      "step": 6268
+    },
+    {
+      "epoch": 0.48219367740943003,
+      "learning_rate": 0.0015838665616746811,
+      "loss": 1.2347,
+      "step": 6269
+    },
+    {
+      "epoch": 0.4822705945696485,
+      "learning_rate": 0.0015835046626335495,
+      "loss": 1.2136,
+      "step": 6270
+    },
+    {
+      "epoch": 0.48234751172986695,
+      "learning_rate": 0.001583142758716494,
+      "loss": 0.9565,
+      "step": 6271
+    },
+    {
+      "epoch": 0.48242442889008536,
+      "learning_rate": 0.0015827808499446461,
+      "loss": 1.4013,
+      "step": 6272
+    },
+    {
+      "epoch": 0.4825013460503038,
+      "learning_rate": 0.0015824189363391383,
+      "loss": 0.9568,
+      "step": 6273
+    },
+    {
+      "epoch": 0.4825782632105223,
+      "learning_rate": 0.0015820570179211023,
+      "loss": 0.9756,
+      "step": 6274
+    },
+    {
+      "epoch": 0.48265518037074073,
+      "learning_rate": 0.0015816950947116722,
+      "loss": 1.3315,
+      "step": 6275
+    },
+    {
+      "epoch": 0.48273209753095914,
+      "learning_rate": 0.0015813331667319799,
+      "loss": 1.2767,
+      "step": 6276
+    },
+    {
+      "epoch": 0.4828090146911776,
+      "learning_rate": 0.0015809712340031593,
+      "loss": 1.3024,
+      "step": 6277
+    },
+    {
+      "epoch": 0.48288593185139606,
+      "learning_rate": 0.0015806092965463443,
+      "loss": 1.3587,
+      "step": 6278
+    },
+    {
+      "epoch": 0.4829628490116145,
+      "learning_rate": 0.001580247354382668,
+      "loss": 1.3582,
+      "step": 6279
+    },
+    {
+      "epoch": 0.4830397661718329,
+      "learning_rate": 0.0015798854075332657,
+      "loss": 0.8317,
+      "step": 6280
+    },
+    {
+      "epoch": 0.4831166833320514,
+      "learning_rate": 0.00157952345601927,
+      "loss": 1.1435,
+      "step": 6281
+    },
+    {
+      "epoch": 0.48319360049226984,
+      "learning_rate": 0.0015791614998618179,
+      "loss": 1.0068,
+      "step": 6282
+    },
+    {
+      "epoch": 0.4832705176524883,
+      "learning_rate": 0.001578799539082043,
+      "loss": 1.4294,
+      "step": 6283
+    },
+    {
+      "epoch": 0.4833474348127067,
+      "learning_rate": 0.0015784375737010812,
+      "loss": 0.9777,
+      "step": 6284
+    },
+    {
+      "epoch": 0.48342435197292516,
+      "learning_rate": 0.0015780756037400672,
+      "loss": 1.2644,
+      "step": 6285
+    },
+    {
+      "epoch": 0.4835012691331436,
+      "learning_rate": 0.0015777136292201377,
+      "loss": 0.974,
+      "step": 6286
+    },
+    {
+      "epoch": 0.483578186293362,
+      "learning_rate": 0.0015773516501624283,
+      "loss": 0.7558,
+      "step": 6287
+    },
+    {
+      "epoch": 0.4836551034535805,
+      "learning_rate": 0.001576989666588075,
+      "loss": 1.4457,
+      "step": 6288
+    },
+    {
+      "epoch": 0.48373202061379894,
+      "learning_rate": 0.0015766276785182153,
+      "loss": 1.1475,
+      "step": 6289
+    },
+    {
+      "epoch": 0.4838089377740174,
+      "learning_rate": 0.0015762656859739854,
+      "loss": 1.1936,
+      "step": 6290
+    },
+    {
+      "epoch": 0.4838858549342358,
+      "learning_rate": 0.0015759036889765232,
+      "loss": 1.1512,
+      "step": 6291
+    },
+    {
+      "epoch": 0.48396277209445426,
+      "learning_rate": 0.0015755416875469649,
+      "loss": 0.9819,
+      "step": 6292
+    },
+    {
+      "epoch": 0.4840396892546727,
+      "learning_rate": 0.0015751796817064488,
+      "loss": 1.3392,
+      "step": 6293
+    },
+    {
+      "epoch": 0.4841166064148912,
+      "learning_rate": 0.0015748176714761126,
+      "loss": 1.0857,
+      "step": 6294
+    },
+    {
+      "epoch": 0.4841935235751096,
+      "learning_rate": 0.0015744556568770952,
+      "loss": 1.0391,
+      "step": 6295
+    },
+    {
+      "epoch": 0.48427044073532804,
+      "learning_rate": 0.0015740936379305336,
+      "loss": 1.0908,
+      "step": 6296
+    },
+    {
+      "epoch": 0.4843473578955465,
+      "learning_rate": 0.0015737316146575678,
+      "loss": 1.1738,
+      "step": 6297
+    },
+    {
+      "epoch": 0.48442427505576496,
+      "learning_rate": 0.0015733695870793357,
+      "loss": 1.1246,
+      "step": 6298
+    },
+    {
+      "epoch": 0.48450119221598337,
+      "learning_rate": 0.0015730075552169774,
+      "loss": 1.1251,
+      "step": 6299
+    },
+    {
+      "epoch": 0.4845781093762018,
+      "learning_rate": 0.001572645519091632,
+      "loss": 1.2144,
+      "step": 6300
+    },
+    {
+      "epoch": 0.4846550265364203,
+      "learning_rate": 0.0015722834787244384,
+      "loss": 1.242,
+      "step": 6301
+    },
+    {
+      "epoch": 0.48473194369663875,
+      "learning_rate": 0.0015719214341365377,
+      "loss": 1.0311,
+      "step": 6302
+    },
+    {
+      "epoch": 0.48480886085685715,
+      "learning_rate": 0.0015715593853490694,
+      "loss": 1.3294,
+      "step": 6303
+    },
+    {
+      "epoch": 0.4848857780170756,
+      "learning_rate": 0.001571197332383174,
+      "loss": 1.4097,
+      "step": 6304
+    },
+    {
+      "epoch": 0.48496269517729407,
+      "learning_rate": 0.001570835275259992,
+      "loss": 1.1468,
+      "step": 6305
+    },
+    {
+      "epoch": 0.48503961233751247,
+      "learning_rate": 0.0015704732140006646,
+      "loss": 1.0481,
+      "step": 6306
+    },
+    {
+      "epoch": 0.48511652949773093,
+      "learning_rate": 0.0015701111486263327,
+      "loss": 1.2476,
+      "step": 6307
+    },
+    {
+      "epoch": 0.4851934466579494,
+      "learning_rate": 0.001569749079158138,
+      "loss": 0.8719,
+      "step": 6308
+    },
+    {
+      "epoch": 0.48527036381816785,
+      "learning_rate": 0.0015693870056172213,
+      "loss": 0.9322,
+      "step": 6309
+    },
+    {
+      "epoch": 0.48534728097838625,
+      "learning_rate": 0.0015690249280247255,
+      "loss": 1.0444,
+      "step": 6310
+    },
+    {
+      "epoch": 0.4854241981386047,
+      "learning_rate": 0.0015686628464017922,
+      "loss": 1.0921,
+      "step": 6311
+    },
+    {
+      "epoch": 0.48550111529882317,
+      "learning_rate": 0.0015683007607695636,
+      "loss": 0.7263,
+      "step": 6312
+    },
+    {
+      "epoch": 0.48557803245904163,
+      "learning_rate": 0.0015679386711491825,
+      "loss": 1.0986,
+      "step": 6313
+    },
+    {
+      "epoch": 0.48565494961926003,
+      "learning_rate": 0.0015675765775617916,
+      "loss": 0.9929,
+      "step": 6314
+    },
+    {
+      "epoch": 0.4857318667794785,
+      "learning_rate": 0.0015672144800285338,
+      "loss": 1.319,
+      "step": 6315
+    },
+    {
+      "epoch": 0.48580878393969695,
+      "learning_rate": 0.0015668523785705526,
+      "loss": 1.1637,
+      "step": 6316
+    },
+    {
+      "epoch": 0.4858857010999154,
+      "learning_rate": 0.001566490273208991,
+      "loss": 0.7977,
+      "step": 6317
+    },
+    {
+      "epoch": 0.4859626182601338,
+      "learning_rate": 0.0015661281639649937,
+      "loss": 1.0797,
+      "step": 6318
+    },
+    {
+      "epoch": 0.4860395354203523,
+      "learning_rate": 0.0015657660508597035,
+      "loss": 1.0003,
+      "step": 6319
+    },
+    {
+      "epoch": 0.48611645258057073,
+      "learning_rate": 0.0015654039339142649,
+      "loss": 1.157,
+      "step": 6320
+    },
+    {
+      "epoch": 0.4861933697407892,
+      "learning_rate": 0.0015650418131498227,
+      "loss": 1.1297,
+      "step": 6321
+    },
+    {
+      "epoch": 0.4862702869010076,
+      "learning_rate": 0.0015646796885875217,
+      "loss": 1.2896,
+      "step": 6322
+    },
+    {
+      "epoch": 0.48634720406122606,
+      "learning_rate": 0.0015643175602485053,
+      "loss": 1.3231,
+      "step": 6323
+    },
+    {
+      "epoch": 0.4864241212214445,
+      "learning_rate": 0.0015639554281539204,
+      "loss": 1.3241,
+      "step": 6324
+    },
+    {
+      "epoch": 0.486501038381663,
+      "learning_rate": 0.0015635932923249106,
+      "loss": 1.056,
+      "step": 6325
+    },
+    {
+      "epoch": 0.4865779555418814,
+      "learning_rate": 0.001563231152782623,
+      "loss": 0.8163,
+      "step": 6326
+    },
+    {
+      "epoch": 0.48665487270209984,
+      "learning_rate": 0.001562869009548202,
+      "loss": 1.0155,
+      "step": 6327
+    },
+    {
+      "epoch": 0.4867317898623183,
+      "learning_rate": 0.0015625068626427942,
+      "loss": 1.0447,
+      "step": 6328
+    },
+    {
+      "epoch": 0.4868087070225367,
+      "learning_rate": 0.0015621447120875455,
+      "loss": 1.3451,
+      "step": 6329
+    },
+    {
+      "epoch": 0.48688562418275516,
+      "learning_rate": 0.0015617825579036023,
+      "loss": 1.021,
+      "step": 6330
+    },
+    {
+      "epoch": 0.4869625413429736,
+      "learning_rate": 0.0015614204001121108,
+      "loss": 0.8143,
+      "step": 6331
+    },
+    {
+      "epoch": 0.4870394585031921,
+      "learning_rate": 0.0015610582387342185,
+      "loss": 1.2286,
+      "step": 6332
+    },
+    {
+      "epoch": 0.4871163756634105,
+      "learning_rate": 0.001560696073791072,
+      "loss": 1.0498,
+      "step": 6333
+    },
+    {
+      "epoch": 0.48719329282362894,
+      "learning_rate": 0.0015603339053038186,
+      "loss": 0.8564,
+      "step": 6334
+    },
+    {
+      "epoch": 0.4872702099838474,
+      "learning_rate": 0.0015599717332936051,
+      "loss": 1.3406,
+      "step": 6335
+    },
+    {
+      "epoch": 0.48734712714406586,
+      "learning_rate": 0.00155960955778158,
+      "loss": 0.9921,
+      "step": 6336
+    },
+    {
+      "epoch": 0.48742404430428427,
+      "learning_rate": 0.001559247378788891,
+      "loss": 1.0311,
+      "step": 6337
+    },
+    {
+      "epoch": 0.4875009614645027,
+      "learning_rate": 0.0015588851963366853,
+      "loss": 1.3841,
+      "step": 6338
+    },
+    {
+      "epoch": 0.4875778786247212,
+      "learning_rate": 0.0015585230104461119,
+      "loss": 1.2471,
+      "step": 6339
+    },
+    {
+      "epoch": 0.48765479578493964,
+      "learning_rate": 0.0015581608211383184,
+      "loss": 1.1149,
+      "step": 6340
+    },
+    {
+      "epoch": 0.48773171294515805,
+      "learning_rate": 0.0015577986284344544,
+      "loss": 1.2192,
+      "step": 6341
+    },
+    {
+      "epoch": 0.4878086301053765,
+      "learning_rate": 0.0015574364323556683,
+      "loss": 1.0891,
+      "step": 6342
+    },
+    {
+      "epoch": 0.48788554726559497,
+      "learning_rate": 0.0015570742329231088,
+      "loss": 1.2197,
+      "step": 6343
+    },
+    {
+      "epoch": 0.4879624644258134,
+      "learning_rate": 0.0015567120301579257,
+      "loss": 1.225,
+      "step": 6344
+    },
+    {
+      "epoch": 0.48803938158603183,
+      "learning_rate": 0.0015563498240812675,
+      "loss": 1.234,
+      "step": 6345
+    },
+    {
+      "epoch": 0.4881162987462503,
+      "learning_rate": 0.001555987614714285,
+      "loss": 0.8916,
+      "step": 6346
+    },
+    {
+      "epoch": 0.48819321590646875,
+      "learning_rate": 0.0015556254020781268,
+      "loss": 1.0892,
+      "step": 6347
+    },
+    {
+      "epoch": 0.48827013306668715,
+      "learning_rate": 0.001555263186193944,
+      "loss": 1.1146,
+      "step": 6348
+    },
+    {
+      "epoch": 0.4883470502269056,
+      "learning_rate": 0.0015549009670828855,
+      "loss": 1.1832,
+      "step": 6349
+    },
+    {
+      "epoch": 0.48842396738712407,
+      "learning_rate": 0.0015545387447661028,
+      "loss": 1.2099,
+      "step": 6350
+    },
+    {
+      "epoch": 0.48850088454734253,
+      "learning_rate": 0.0015541765192647456,
+      "loss": 1.2221,
+      "step": 6351
+    },
+    {
+      "epoch": 0.48857780170756093,
+      "learning_rate": 0.0015538142905999652,
+      "loss": 1.2262,
+      "step": 6352
+    },
+    {
+      "epoch": 0.4886547188677794,
+      "learning_rate": 0.0015534520587929122,
+      "loss": 1.0718,
+      "step": 6353
+    },
+    {
+      "epoch": 0.48873163602799785,
+      "learning_rate": 0.0015530898238647374,
+      "loss": 1.2102,
+      "step": 6354
+    },
+    {
+      "epoch": 0.4888085531882163,
+      "learning_rate": 0.0015527275858365933,
+      "loss": 0.7523,
+      "step": 6355
+    },
+    {
+      "epoch": 0.4888854703484347,
+      "learning_rate": 0.0015523653447296298,
+      "loss": 1.2573,
+      "step": 6356
+    },
+    {
+      "epoch": 0.4889623875086532,
+      "learning_rate": 0.0015520031005649996,
+      "loss": 1.4722,
+      "step": 6357
+    },
+    {
+      "epoch": 0.48903930466887163,
+      "learning_rate": 0.001551640853363854,
+      "loss": 1.1594,
+      "step": 6358
+    },
+    {
+      "epoch": 0.4891162218290901,
+      "learning_rate": 0.0015512786031473453,
+      "loss": 1.0412,
+      "step": 6359
+    },
+    {
+      "epoch": 0.4891931389893085,
+      "learning_rate": 0.0015509163499366257,
+      "loss": 1.0809,
+      "step": 6360
+    },
+    {
+      "epoch": 0.48927005614952696,
+      "learning_rate": 0.0015505540937528473,
+      "loss": 1.103,
+      "step": 6361
+    },
+    {
+      "epoch": 0.4893469733097454,
+      "learning_rate": 0.0015501918346171619,
+      "loss": 1.2935,
+      "step": 6362
+    },
+    {
+      "epoch": 0.4894238904699639,
+      "learning_rate": 0.0015498295725507238,
+      "loss": 1.3264,
+      "step": 6363
+    },
+    {
+      "epoch": 0.4895008076301823,
+      "learning_rate": 0.001549467307574685,
+      "loss": 1.2447,
+      "step": 6364
+    },
+    {
+      "epoch": 0.48957772479040074,
+      "learning_rate": 0.001549105039710198,
+      "loss": 1.3307,
+      "step": 6365
+    },
+    {
+      "epoch": 0.4896546419506192,
+      "learning_rate": 0.0015487427689784168,
+      "loss": 1.1443,
+      "step": 6366
+    },
+    {
+      "epoch": 0.48973155911083766,
+      "learning_rate": 0.0015483804954004945,
+      "loss": 1.2331,
+      "step": 6367
+    },
+    {
+      "epoch": 0.48980847627105606,
+      "learning_rate": 0.0015480182189975852,
+      "loss": 1.1772,
+      "step": 6368
+    },
+    {
+      "epoch": 0.4898853934312745,
+      "learning_rate": 0.0015476559397908413,
+      "loss": 1.0711,
+      "step": 6369
+    },
+    {
+      "epoch": 0.489962310591493,
+      "learning_rate": 0.0015472936578014179,
+      "loss": 0.933,
+      "step": 6370
+    },
+    {
+      "epoch": 0.4900392277517114,
+      "learning_rate": 0.0015469313730504684,
+      "loss": 1.4777,
+      "step": 6371
+    },
+    {
+      "epoch": 0.49011614491192984,
+      "learning_rate": 0.0015465690855591472,
+      "loss": 1.0311,
+      "step": 6372
+    },
+    {
+      "epoch": 0.4901930620721483,
+      "learning_rate": 0.0015462067953486084,
+      "loss": 1.142,
+      "step": 6373
+    },
+    {
+      "epoch": 0.49026997923236676,
+      "learning_rate": 0.0015458445024400067,
+      "loss": 0.9315,
+      "step": 6374
+    },
+    {
+      "epoch": 0.49034689639258516,
+      "learning_rate": 0.0015454822068544968,
+      "loss": 1.1105,
+      "step": 6375
+    },
+    {
+      "epoch": 0.4904238135528036,
+      "learning_rate": 0.001545119908613233,
+      "loss": 1.25,
+      "step": 6376
+    },
+    {
+      "epoch": 0.4905007307130221,
+      "learning_rate": 0.0015447576077373716,
+      "loss": 1.1408,
+      "step": 6377
+    },
+    {
+      "epoch": 0.49057764787324054,
+      "learning_rate": 0.0015443953042480658,
+      "loss": 1.1946,
+      "step": 6378
+    },
+    {
+      "epoch": 0.49065456503345894,
+      "learning_rate": 0.0015440329981664727,
+      "loss": 0.9651,
+      "step": 6379
+    },
+    {
+      "epoch": 0.4907314821936774,
+      "learning_rate": 0.0015436706895137467,
+      "loss": 1.1148,
+      "step": 6380
+    },
+    {
+      "epoch": 0.49080839935389586,
+      "learning_rate": 0.0015433083783110435,
+      "loss": 1.015,
+      "step": 6381
+    },
+    {
+      "epoch": 0.4908853165141143,
+      "learning_rate": 0.0015429460645795192,
+      "loss": 1.265,
+      "step": 6382
+    },
+    {
+      "epoch": 0.4909622336743327,
+      "learning_rate": 0.0015425837483403297,
+      "loss": 0.9602,
+      "step": 6383
+    },
+    {
+      "epoch": 0.4910391508345512,
+      "learning_rate": 0.0015422214296146298,
+      "loss": 1.3532,
+      "step": 6384
+    },
+    {
+      "epoch": 0.49111606799476965,
+      "learning_rate": 0.0015418591084235774,
+      "loss": 1.2866,
+      "step": 6385
+    },
+    {
+      "epoch": 0.4911929851549881,
+      "learning_rate": 0.0015414967847883275,
+      "loss": 1.1965,
+      "step": 6386
+    },
+    {
+      "epoch": 0.4912699023152065,
+      "learning_rate": 0.0015411344587300375,
+      "loss": 1.3679,
+      "step": 6387
+    },
+    {
+      "epoch": 0.49134681947542497,
+      "learning_rate": 0.0015407721302698638,
+      "loss": 1.3245,
+      "step": 6388
+    },
+    {
+      "epoch": 0.4914237366356434,
+      "learning_rate": 0.0015404097994289622,
+      "loss": 1.0462,
+      "step": 6389
+    },
+    {
+      "epoch": 0.49150065379586183,
+      "learning_rate": 0.0015400474662284913,
+      "loss": 1.2234,
+      "step": 6390
+    },
+    {
+      "epoch": 0.4915775709560803,
+      "learning_rate": 0.0015396851306896061,
+      "loss": 1.1224,
+      "step": 6391
+    },
+    {
+      "epoch": 0.49165448811629875,
+      "learning_rate": 0.0015393227928334654,
+      "loss": 1.2868,
+      "step": 6392
+    },
+    {
+      "epoch": 0.4917314052765172,
+      "learning_rate": 0.0015389604526812252,
+      "loss": 1.1485,
+      "step": 6393
+    },
+    {
+      "epoch": 0.4918083224367356,
+      "learning_rate": 0.0015385981102540444,
+      "loss": 1.1772,
+      "step": 6394
+    },
+    {
+      "epoch": 0.49188523959695407,
+      "learning_rate": 0.0015382357655730793,
+      "loss": 1.0205,
+      "step": 6395
+    },
+    {
+      "epoch": 0.49196215675717253,
+      "learning_rate": 0.001537873418659488,
+      "loss": 1.3706,
+      "step": 6396
+    },
+    {
+      "epoch": 0.492039073917391,
+      "learning_rate": 0.001537511069534428,
+      "loss": 0.9855,
+      "step": 6397
+    },
+    {
+      "epoch": 0.4921159910776094,
+      "learning_rate": 0.0015371487182190581,
+      "loss": 1.1971,
+      "step": 6398
+    },
+    {
+      "epoch": 0.49219290823782785,
+      "learning_rate": 0.0015367863647345356,
+      "loss": 1.2389,
+      "step": 6399
+    },
+    {
+      "epoch": 0.4922698253980463,
+      "learning_rate": 0.0015364240091020188,
+      "loss": 1.0502,
+      "step": 6400
+    },
+    {
+      "epoch": 0.49234674255826477,
+      "learning_rate": 0.0015360616513426665,
+      "loss": 1.397,
+      "step": 6401
+    },
+    {
+      "epoch": 0.4924236597184832,
+      "learning_rate": 0.0015356992914776367,
+      "loss": 1.1998,
+      "step": 6402
+    },
+    {
+      "epoch": 0.49250057687870163,
+      "learning_rate": 0.0015353369295280877,
+      "loss": 1.0317,
+      "step": 6403
+    },
+    {
+      "epoch": 0.4925774940389201,
+      "learning_rate": 0.0015349745655151785,
+      "loss": 1.1539,
+      "step": 6404
+    },
+    {
+      "epoch": 0.49265441119913855,
+      "learning_rate": 0.0015346121994600688,
+      "loss": 0.9895,
+      "step": 6405
+    },
+    {
+      "epoch": 0.49273132835935696,
+      "learning_rate": 0.0015342498313839163,
+      "loss": 1.0169,
+      "step": 6406
+    },
+    {
+      "epoch": 0.4928082455195754,
+      "learning_rate": 0.00153388746130788,
+      "loss": 0.7782,
+      "step": 6407
+    },
+    {
+      "epoch": 0.4928851626797939,
+      "learning_rate": 0.0015335250892531196,
+      "loss": 0.9776,
+      "step": 6408
+    },
+    {
+      "epoch": 0.4929620798400123,
+      "learning_rate": 0.0015331627152407946,
+      "loss": 0.9775,
+      "step": 6409
+    },
+    {
+      "epoch": 0.49303899700023074,
+      "learning_rate": 0.0015328003392920642,
+      "loss": 1.3916,
+      "step": 6410
+    },
+    {
+      "epoch": 0.4931159141604492,
+      "learning_rate": 0.0015324379614280875,
+      "loss": 1.29,
+      "step": 6411
+    },
+    {
+      "epoch": 0.49319283132066766,
+      "learning_rate": 0.0015320755816700246,
+      "loss": 1.3708,
+      "step": 6412
+    },
+    {
+      "epoch": 0.49326974848088606,
+      "learning_rate": 0.0015317132000390351,
+      "loss": 1.2731,
+      "step": 6413
+    },
+    {
+      "epoch": 0.4933466656411045,
+      "learning_rate": 0.0015313508165562789,
+      "loss": 1.2686,
+      "step": 6414
+    },
+    {
+      "epoch": 0.493423582801323,
+      "learning_rate": 0.0015309884312429153,
+      "loss": 0.9308,
+      "step": 6415
+    },
+    {
+      "epoch": 0.49350049996154144,
+      "learning_rate": 0.0015306260441201056,
+      "loss": 0.9955,
+      "step": 6416
+    },
+    {
+      "epoch": 0.49357741712175984,
+      "learning_rate": 0.0015302636552090088,
+      "loss": 1.1347,
+      "step": 6417
+    },
+    {
+      "epoch": 0.4936543342819783,
+      "learning_rate": 0.001529901264530786,
+      "loss": 1.1132,
+      "step": 6418
+    },
+    {
+      "epoch": 0.49373125144219676,
+      "learning_rate": 0.0015295388721065968,
+      "loss": 0.8727,
+      "step": 6419
+    },
+    {
+      "epoch": 0.4938081686024152,
+      "learning_rate": 0.0015291764779576022,
+      "loss": 0.8353,
+      "step": 6420
+    },
+    {
+      "epoch": 0.4938850857626336,
+      "learning_rate": 0.0015288140821049629,
+      "loss": 0.89,
+      "step": 6421
+    },
+    {
+      "epoch": 0.4939620029228521,
+      "learning_rate": 0.0015284516845698389,
+      "loss": 1.022,
+      "step": 6422
+    },
+    {
+      "epoch": 0.49403892008307054,
+      "learning_rate": 0.0015280892853733916,
+      "loss": 0.8269,
+      "step": 6423
+    },
+    {
+      "epoch": 0.494115837243289,
+      "learning_rate": 0.0015277268845367814,
+      "loss": 1.1641,
+      "step": 6424
+    },
+    {
+      "epoch": 0.4941927544035074,
+      "learning_rate": 0.0015273644820811704,
+      "loss": 0.8085,
+      "step": 6425
+    },
+    {
+      "epoch": 0.49426967156372587,
+      "learning_rate": 0.0015270020780277177,
+      "loss": 1.4092,
+      "step": 6426
+    },
+    {
+      "epoch": 0.4943465887239443,
+      "learning_rate": 0.0015266396723975862,
+      "loss": 1.3734,
+      "step": 6427
+    },
+    {
+      "epoch": 0.4944235058841628,
+      "learning_rate": 0.0015262772652119357,
+      "loss": 1.3758,
+      "step": 6428
+    },
+    {
+      "epoch": 0.4945004230443812,
+      "learning_rate": 0.0015259148564919292,
+      "loss": 1.2438,
+      "step": 6429
+    },
+    {
+      "epoch": 0.49457734020459965,
+      "learning_rate": 0.0015255524462587267,
+      "loss": 1.1805,
+      "step": 6430
+    },
+    {
+      "epoch": 0.4946542573648181,
+      "learning_rate": 0.0015251900345334903,
+      "loss": 1.284,
+      "step": 6431
+    },
+    {
+      "epoch": 0.4947311745250365,
+      "learning_rate": 0.001524827621337382,
+      "loss": 1.5342,
+      "step": 6432
+    },
+    {
+      "epoch": 0.49480809168525497,
+      "learning_rate": 0.0015244652066915623,
+      "loss": 1.1638,
+      "step": 6433
+    },
+    {
+      "epoch": 0.49488500884547343,
+      "learning_rate": 0.001524102790617194,
+      "loss": 0.8618,
+      "step": 6434
+    },
+    {
+      "epoch": 0.4949619260056919,
+      "learning_rate": 0.0015237403731354386,
+      "loss": 1.1474,
+      "step": 6435
+    },
+    {
+      "epoch": 0.4950388431659103,
+      "learning_rate": 0.0015233779542674587,
+      "loss": 1.397,
+      "step": 6436
+    },
+    {
+      "epoch": 0.49511576032612875,
+      "learning_rate": 0.0015230155340344145,
+      "loss": 1.4225,
+      "step": 6437
+    },
+    {
+      "epoch": 0.4951926774863472,
+      "learning_rate": 0.0015226531124574703,
+      "loss": 1.0917,
+      "step": 6438
+    },
+    {
+      "epoch": 0.49526959464656567,
+      "learning_rate": 0.0015222906895577864,
+      "loss": 1.1323,
+      "step": 6439
+    },
+    {
+      "epoch": 0.4953465118067841,
+      "learning_rate": 0.0015219282653565265,
+      "loss": 1.2818,
+      "step": 6440
+    },
+    {
+      "epoch": 0.49542342896700253,
+      "learning_rate": 0.001521565839874852,
+      "loss": 1.0726,
+      "step": 6441
+    },
+    {
+      "epoch": 0.495500346127221,
+      "learning_rate": 0.0015212034131339258,
+      "loss": 1.1083,
+      "step": 6442
+    },
+    {
+      "epoch": 0.49557726328743945,
+      "learning_rate": 0.0015208409851549104,
+      "loss": 1.1327,
+      "step": 6443
+    },
+    {
+      "epoch": 0.49565418044765785,
+      "learning_rate": 0.001520478555958968,
+      "loss": 1.031,
+      "step": 6444
+    },
+    {
+      "epoch": 0.4957310976078763,
+      "learning_rate": 0.0015201161255672614,
+      "loss": 0.8077,
+      "step": 6445
+    },
+    {
+      "epoch": 0.4958080147680948,
+      "learning_rate": 0.001519753694000953,
+      "loss": 1.2604,
+      "step": 6446
+    },
+    {
+      "epoch": 0.49588493192831323,
+      "learning_rate": 0.0015193912612812062,
+      "loss": 1.7456,
+      "step": 6447
+    },
+    {
+      "epoch": 0.49596184908853164,
+      "learning_rate": 0.0015190288274291833,
+      "loss": 0.9868,
+      "step": 6448
+    },
+    {
+      "epoch": 0.4960387662487501,
+      "learning_rate": 0.0015186663924660476,
+      "loss": 1.3619,
+      "step": 6449
+    },
+    {
+      "epoch": 0.49611568340896856,
+      "learning_rate": 0.0015183039564129613,
+      "loss": 0.9619,
+      "step": 6450
+    },
+    {
+      "epoch": 0.49619260056918696,
+      "learning_rate": 0.0015179415192910879,
+      "loss": 1.1899,
+      "step": 6451
+    },
+    {
+      "epoch": 0.4962695177294054,
+      "learning_rate": 0.001517579081121591,
+      "loss": 1.2517,
+      "step": 6452
+    },
+    {
+      "epoch": 0.4963464348896239,
+      "learning_rate": 0.0015172166419256327,
+      "loss": 1.1316,
+      "step": 6453
+    },
+    {
+      "epoch": 0.49642335204984234,
+      "learning_rate": 0.001516854201724377,
+      "loss": 1.0495,
+      "step": 6454
+    },
+    {
+      "epoch": 0.49650026921006074,
+      "learning_rate": 0.001516491760538987,
+      "loss": 1.1894,
+      "step": 6455
+    },
+    {
+      "epoch": 0.4965771863702792,
+      "learning_rate": 0.0015161293183906259,
+      "loss": 1.2593,
+      "step": 6456
+    },
+    {
+      "epoch": 0.49665410353049766,
+      "learning_rate": 0.0015157668753004567,
+      "loss": 1.1953,
+      "step": 6457
+    },
+    {
+      "epoch": 0.4967310206907161,
+      "learning_rate": 0.0015154044312896435,
+      "loss": 1.6624,
+      "step": 6458
+    },
+    {
+      "epoch": 0.4968079378509345,
+      "learning_rate": 0.0015150419863793497,
+      "loss": 1.1629,
+      "step": 6459
+    },
+    {
+      "epoch": 0.496884855011153,
+      "learning_rate": 0.0015146795405907387,
+      "loss": 1.5651,
+      "step": 6460
+    },
+    {
+      "epoch": 0.49696177217137144,
+      "learning_rate": 0.0015143170939449734,
+      "loss": 1.0473,
+      "step": 6461
+    },
+    {
+      "epoch": 0.4970386893315899,
+      "learning_rate": 0.0015139546464632183,
+      "loss": 1.0656,
+      "step": 6462
+    },
+    {
+      "epoch": 0.4971156064918083,
+      "learning_rate": 0.0015135921981666372,
+      "loss": 0.934,
+      "step": 6463
+    },
+    {
+      "epoch": 0.49719252365202676,
+      "learning_rate": 0.001513229749076393,
+      "loss": 1.3089,
+      "step": 6464
+    },
+    {
+      "epoch": 0.4972694408122452,
+      "learning_rate": 0.0015128672992136506,
+      "loss": 1.0161,
+      "step": 6465
+    },
+    {
+      "epoch": 0.4973463579724637,
+      "learning_rate": 0.0015125048485995726,
+      "loss": 1.4793,
+      "step": 6466
+    },
+    {
+      "epoch": 0.4974232751326821,
+      "learning_rate": 0.0015121423972553239,
+      "loss": 1.2478,
+      "step": 6467
+    },
+    {
+      "epoch": 0.49750019229290054,
+      "learning_rate": 0.0015117799452020674,
+      "loss": 1.3143,
+      "step": 6468
+    },
+    {
+      "epoch": 0.497577109453119,
+      "learning_rate": 0.0015114174924609682,
+      "loss": 1.1109,
+      "step": 6469
+    },
+    {
+      "epoch": 0.4976540266133374,
+      "learning_rate": 0.0015110550390531896,
+      "loss": 0.9355,
+      "step": 6470
+    },
+    {
+      "epoch": 0.49773094377355587,
+      "learning_rate": 0.0015106925849998956,
+      "loss": 1.045,
+      "step": 6471
+    },
+    {
+      "epoch": 0.4978078609337743,
+      "learning_rate": 0.00151033013032225,
+      "loss": 1.1446,
+      "step": 6472
+    },
+    {
+      "epoch": 0.4978847780939928,
+      "learning_rate": 0.001509967675041418,
+      "loss": 1.1395,
+      "step": 6473
+    },
+    {
+      "epoch": 0.4979616952542112,
+      "learning_rate": 0.001509605219178563,
+      "loss": 1.1233,
+      "step": 6474
+    },
+    {
+      "epoch": 0.49803861241442965,
+      "learning_rate": 0.001509242762754849,
+      "loss": 1.1183,
+      "step": 6475
+    },
+    {
+      "epoch": 0.4981155295746481,
+      "learning_rate": 0.0015088803057914403,
+      "loss": 1.4301,
+      "step": 6476
+    },
+    {
+      "epoch": 0.49819244673486657,
+      "learning_rate": 0.0015085178483095014,
+      "loss": 1.1801,
+      "step": 6477
+    },
+    {
+      "epoch": 0.49826936389508497,
+      "learning_rate": 0.001508155390330197,
+      "loss": 0.963,
+      "step": 6478
+    },
+    {
+      "epoch": 0.49834628105530343,
+      "learning_rate": 0.00150779293187469,
+      "loss": 1.2661,
+      "step": 6479
+    },
+    {
+      "epoch": 0.4984231982155219,
+      "learning_rate": 0.0015074304729641463,
+      "loss": 1.1302,
+      "step": 6480
+    },
+    {
+      "epoch": 0.49850011537574035,
+      "learning_rate": 0.0015070680136197291,
+      "loss": 1.0476,
+      "step": 6481
+    },
+    {
+      "epoch": 0.49857703253595875,
+      "learning_rate": 0.0015067055538626037,
+      "loss": 1.2589,
+      "step": 6482
+    },
+    {
+      "epoch": 0.4986539496961772,
+      "learning_rate": 0.0015063430937139336,
+      "loss": 0.9057,
+      "step": 6483
+    },
+    {
+      "epoch": 0.49873086685639567,
+      "learning_rate": 0.0015059806331948838,
+      "loss": 0.9872,
+      "step": 6484
+    },
+    {
+      "epoch": 0.49880778401661413,
+      "learning_rate": 0.0015056181723266183,
+      "loss": 1.2148,
+      "step": 6485
+    },
+    {
+      "epoch": 0.49888470117683253,
+      "learning_rate": 0.001505255711130302,
+      "loss": 1.2279,
+      "step": 6486
+    },
+    {
+      "epoch": 0.498961618337051,
+      "learning_rate": 0.0015048932496270998,
+      "loss": 0.8585,
+      "step": 6487
+    },
+    {
+      "epoch": 0.49903853549726945,
+      "learning_rate": 0.0015045307878381747,
+      "loss": 1.2618,
+      "step": 6488
+    },
+    {
+      "epoch": 0.4991154526574879,
+      "learning_rate": 0.001504168325784693,
+      "loss": 1.0307,
+      "step": 6489
+    },
+    {
+      "epoch": 0.4991923698177063,
+      "learning_rate": 0.0015038058634878181,
+      "loss": 1.0074,
+      "step": 6490
+    },
+    {
+      "epoch": 0.4992692869779248,
+      "learning_rate": 0.001503443400968715,
+      "loss": 1.0053,
+      "step": 6491
+    },
+    {
+      "epoch": 0.49934620413814323,
+      "learning_rate": 0.0015030809382485478,
+      "loss": 1.1319,
+      "step": 6492
+    },
+    {
+      "epoch": 0.49942312129836164,
+      "learning_rate": 0.0015027184753484818,
+      "loss": 1.2984,
+      "step": 6493
+    },
+    {
+      "epoch": 0.4995000384585801,
+      "learning_rate": 0.0015023560122896808,
+      "loss": 0.7327,
+      "step": 6494
+    },
+    {
+      "epoch": 0.49957695561879856,
+      "learning_rate": 0.0015019935490933104,
+      "loss": 1.4445,
+      "step": 6495
+    },
+    {
+      "epoch": 0.499653872779017,
+      "learning_rate": 0.001501631085780534,
+      "loss": 1.1119,
+      "step": 6496
+    },
+    {
+      "epoch": 0.4997307899392354,
+      "learning_rate": 0.0015012686223725169,
+      "loss": 1.3338,
+      "step": 6497
+    },
+    {
+      "epoch": 0.4998077070994539,
+      "learning_rate": 0.0015009061588904236,
+      "loss": 1.3912,
+      "step": 6498
+    },
+    {
+      "epoch": 0.49988462425967234,
+      "learning_rate": 0.0015005436953554187,
+      "loss": 1.2257,
+      "step": 6499
+    },
+    {
+      "epoch": 0.4999615414198908,
+      "learning_rate": 0.001500181231788667,
+      "loss": 1.5399,
+      "step": 6500
+    },
+    {
+      "epoch": 0.5000384585801092,
+      "learning_rate": 0.001499818768211333,
+      "loss": 0.8163,
+      "step": 6501
+    },
+    {
+      "epoch": 0.5001153757403277,
+      "learning_rate": 0.0014994563046445812,
+      "loss": 0.9792,
+      "step": 6502
+    },
+    {
+      "epoch": 0.5001922929005461,
+      "learning_rate": 0.0014990938411095765,
+      "loss": 1.0609,
+      "step": 6503
+    },
+    {
+      "epoch": 0.5002692100607645,
+      "learning_rate": 0.0014987313776274834,
+      "loss": 1.3299,
+      "step": 6504
+    },
+    {
+      "epoch": 0.500346127220983,
+      "learning_rate": 0.0014983689142194661,
+      "loss": 1.3401,
+      "step": 6505
+    },
+    {
+      "epoch": 0.5004230443812014,
+      "learning_rate": 0.00149800645090669,
+      "loss": 1.1659,
+      "step": 6506
+    },
+    {
+      "epoch": 0.5004999615414198,
+      "learning_rate": 0.0014976439877103197,
+      "loss": 1.1513,
+      "step": 6507
+    },
+    {
+      "epoch": 0.5005768787016384,
+      "learning_rate": 0.0014972815246515187,
+      "loss": 1.3099,
+      "step": 6508
+    },
+    {
+      "epoch": 0.5006537958618568,
+      "learning_rate": 0.0014969190617514525,
+      "loss": 1.0427,
+      "step": 6509
+    },
+    {
+      "epoch": 0.5007307130220753,
+      "learning_rate": 0.001496556599031285,
+      "loss": 1.3597,
+      "step": 6510
+    },
+    {
+      "epoch": 0.5008076301822937,
+      "learning_rate": 0.0014961941365121822,
+      "loss": 0.9758,
+      "step": 6511
+    },
+    {
+      "epoch": 0.5008845473425121,
+      "learning_rate": 0.0014958316742153073,
+      "loss": 1.3063,
+      "step": 6512
+    },
+    {
+      "epoch": 0.5009614645027306,
+      "learning_rate": 0.0014954692121618251,
+      "loss": 1.4286,
+      "step": 6513
+    },
+    {
+      "epoch": 0.501038381662949,
+      "learning_rate": 0.0014951067503729005,
+      "loss": 1.4384,
+      "step": 6514
+    },
+    {
+      "epoch": 0.5011152988231674,
+      "learning_rate": 0.001494744288869698,
+      "loss": 1.0694,
+      "step": 6515
+    },
+    {
+      "epoch": 0.5011922159833859,
+      "learning_rate": 0.0014943818276733816,
+      "loss": 1.3276,
+      "step": 6516
+    },
+    {
+      "epoch": 0.5012691331436043,
+      "learning_rate": 0.0014940193668051165,
+      "loss": 0.8976,
+      "step": 6517
+    },
+    {
+      "epoch": 0.5013460503038227,
+      "learning_rate": 0.001493656906286067,
+      "loss": 1.0597,
+      "step": 6518
+    },
+    {
+      "epoch": 0.5014229674640412,
+      "learning_rate": 0.0014932944461373964,
+      "loss": 0.8895,
+      "step": 6519
+    },
+    {
+      "epoch": 0.5014998846242597,
+      "learning_rate": 0.0014929319863802712,
+      "loss": 0.9636,
+      "step": 6520
+    },
+    {
+      "epoch": 0.5015768017844782,
+      "learning_rate": 0.0014925695270358536,
+      "loss": 1.2442,
+      "step": 6521
+    },
+    {
+      "epoch": 0.5016537189446966,
+      "learning_rate": 0.00149220706812531,
+      "loss": 1.1827,
+      "step": 6522
+    },
+    {
+      "epoch": 0.501730636104915,
+      "learning_rate": 0.0014918446096698036,
+      "loss": 1.2397,
+      "step": 6523
+    },
+    {
+      "epoch": 0.5018075532651335,
+      "learning_rate": 0.0014914821516904984,
+      "loss": 1.1812,
+      "step": 6524
+    },
+    {
+      "epoch": 0.5018844704253519,
+      "learning_rate": 0.0014911196942085597,
+      "loss": 1.4535,
+      "step": 6525
+    },
+    {
+      "epoch": 0.5019613875855703,
+      "learning_rate": 0.0014907572372451514,
+      "loss": 1.0805,
+      "step": 6526
+    },
+    {
+      "epoch": 0.5020383047457888,
+      "learning_rate": 0.0014903947808214372,
+      "loss": 1.2827,
+      "step": 6527
+    },
+    {
+      "epoch": 0.5021152219060072,
+      "learning_rate": 0.0014900323249585823,
+      "loss": 1.1194,
+      "step": 6528
+    },
+    {
+      "epoch": 0.5021921390662257,
+      "learning_rate": 0.0014896698696777496,
+      "loss": 1.1704,
+      "step": 6529
+    },
+    {
+      "epoch": 0.5022690562264441,
+      "learning_rate": 0.0014893074150001046,
+      "loss": 1.235,
+      "step": 6530
+    },
+    {
+      "epoch": 0.5023459733866625,
+      "learning_rate": 0.0014889449609468111,
+      "loss": 1.0981,
+      "step": 6531
+    },
+    {
+      "epoch": 0.502422890546881,
+      "learning_rate": 0.0014885825075390316,
+      "loss": 1.2114,
+      "step": 6532
+    },
+    {
+      "epoch": 0.5024998077070995,
+      "learning_rate": 0.0014882200547979329,
+      "loss": 1.6672,
+      "step": 6533
+    },
+    {
+      "epoch": 0.5025767248673179,
+      "learning_rate": 0.0014878576027446766,
+      "loss": 0.639,
+      "step": 6534
+    },
+    {
+      "epoch": 0.5026536420275364,
+      "learning_rate": 0.0014874951514004275,
+      "loss": 1.3364,
+      "step": 6535
+    },
+    {
+      "epoch": 0.5027305591877548,
+      "learning_rate": 0.00148713270078635,
+      "loss": 1.3282,
+      "step": 6536
+    },
+    {
+      "epoch": 0.5028074763479732,
+      "learning_rate": 0.001486770250923607,
+      "loss": 0.9317,
+      "step": 6537
+    },
+    {
+      "epoch": 0.5028843935081917,
+      "learning_rate": 0.001486407801833363,
+      "loss": 0.6212,
+      "step": 6538
+    },
+    {
+      "epoch": 0.5029613106684101,
+      "learning_rate": 0.0014860453535367817,
+      "loss": 1.1881,
+      "step": 6539
+    },
+    {
+      "epoch": 0.5030382278286286,
+      "learning_rate": 0.0014856829060550267,
+      "loss": 1.2884,
+      "step": 6540
+    },
+    {
+      "epoch": 0.503115144988847,
+      "learning_rate": 0.0014853204594092618,
+      "loss": 1.2122,
+      "step": 6541
+    },
+    {
+      "epoch": 0.5031920621490654,
+      "learning_rate": 0.0014849580136206508,
+      "loss": 1.3868,
+      "step": 6542
+    },
+    {
+      "epoch": 0.5032689793092839,
+      "learning_rate": 0.0014845955687103564,
+      "loss": 1.3568,
+      "step": 6543
+    },
+    {
+      "epoch": 0.5033458964695023,
+      "learning_rate": 0.0014842331246995436,
+      "loss": 0.6795,
+      "step": 6544
+    },
+    {
+      "epoch": 0.5034228136297207,
+      "learning_rate": 0.0014838706816093748,
+      "loss": 1.0597,
+      "step": 6545
+    },
+    {
+      "epoch": 0.5034997307899393,
+      "learning_rate": 0.0014835082394610132,
+      "loss": 1.2906,
+      "step": 6546
+    },
+    {
+      "epoch": 0.5035766479501577,
+      "learning_rate": 0.0014831457982756233,
+      "loss": 1.0363,
+      "step": 6547
+    },
+    {
+      "epoch": 0.5036535651103762,
+      "learning_rate": 0.0014827833580743671,
+      "loss": 1.4559,
+      "step": 6548
+    },
+    {
+      "epoch": 0.5037304822705946,
+      "learning_rate": 0.0014824209188784093,
+      "loss": 0.9388,
+      "step": 6549
+    },
+    {
+      "epoch": 0.503807399430813,
+      "learning_rate": 0.0014820584807089122,
+      "loss": 1.0233,
+      "step": 6550
+    },
+    {
+      "epoch": 0.5038843165910315,
+      "learning_rate": 0.0014816960435870387,
+      "loss": 0.9559,
+      "step": 6551
+    },
+    {
+      "epoch": 0.5039612337512499,
+      "learning_rate": 0.0014813336075339529,
+      "loss": 1.1925,
+      "step": 6552
+    },
+    {
+      "epoch": 0.5040381509114683,
+      "learning_rate": 0.0014809711725708172,
+      "loss": 0.8224,
+      "step": 6553
+    },
+    {
+      "epoch": 0.5041150680716868,
+      "learning_rate": 0.0014806087387187938,
+      "loss": 1.1409,
+      "step": 6554
+    },
+    {
+      "epoch": 0.5041919852319052,
+      "learning_rate": 0.0014802463059990472,
+      "loss": 0.7648,
+      "step": 6555
+    },
+    {
+      "epoch": 0.5042689023921236,
+      "learning_rate": 0.001479883874432739,
+      "loss": 1.1594,
+      "step": 6556
+    },
+    {
+      "epoch": 0.5043458195523421,
+      "learning_rate": 0.0014795214440410322,
+      "loss": 1.0776,
+      "step": 6557
+    },
+    {
+      "epoch": 0.5044227367125605,
+      "learning_rate": 0.0014791590148450899,
+      "loss": 1.282,
+      "step": 6558
+    },
+    {
+      "epoch": 0.5044996538727791,
+      "learning_rate": 0.001478796586866074,
+      "loss": 1.106,
+      "step": 6559
+    },
+    {
+      "epoch": 0.5045765710329975,
+      "learning_rate": 0.001478434160125148,
+      "loss": 1.316,
+      "step": 6560
+    },
+    {
+      "epoch": 0.5046534881932159,
+      "learning_rate": 0.0014780717346434737,
+      "loss": 0.9918,
+      "step": 6561
+    },
+    {
+      "epoch": 0.5047304053534344,
+      "learning_rate": 0.0014777093104422136,
+      "loss": 0.8127,
+      "step": 6562
+    },
+    {
+      "epoch": 0.5048073225136528,
+      "learning_rate": 0.0014773468875425302,
+      "loss": 1.1305,
+      "step": 6563
+    },
+    {
+      "epoch": 0.5048842396738712,
+      "learning_rate": 0.001476984465965586,
+      "loss": 1.1134,
+      "step": 6564
+    },
+    {
+      "epoch": 0.5049611568340897,
+      "learning_rate": 0.0014766220457325418,
+      "loss": 1.299,
+      "step": 6565
+    },
+    {
+      "epoch": 0.5050380739943081,
+      "learning_rate": 0.0014762596268645617,
+      "loss": 1.4228,
+      "step": 6566
+    },
+    {
+      "epoch": 0.5051149911545266,
+      "learning_rate": 0.0014758972093828057,
+      "loss": 1.0898,
+      "step": 6567
+    },
+    {
+      "epoch": 0.505191908314745,
+      "learning_rate": 0.001475534793308438,
+      "loss": 1.192,
+      "step": 6568
+    },
+    {
+      "epoch": 0.5052688254749634,
+      "learning_rate": 0.0014751723786626187,
+      "loss": 1.4412,
+      "step": 6569
+    },
+    {
+      "epoch": 0.505345742635182,
+      "learning_rate": 0.0014748099654665096,
+      "loss": 0.7492,
+      "step": 6570
+    },
+    {
+      "epoch": 0.5054226597954004,
+      "learning_rate": 0.0014744475537412734,
+      "loss": 0.8845,
+      "step": 6571
+    },
+    {
+      "epoch": 0.5054995769556188,
+      "learning_rate": 0.0014740851435080713,
+      "loss": 1.1231,
+      "step": 6572
+    },
+    {
+      "epoch": 0.5055764941158373,
+      "learning_rate": 0.0014737227347880642,
+      "loss": 1.1297,
+      "step": 6573
+    },
+    {
+      "epoch": 0.5056534112760557,
+      "learning_rate": 0.0014733603276024143,
+      "loss": 1.6957,
+      "step": 6574
+    },
+    {
+      "epoch": 0.5057303284362741,
+      "learning_rate": 0.001472997921972283,
+      "loss": 1.1814,
+      "step": 6575
+    },
+    {
+      "epoch": 0.5058072455964926,
+      "learning_rate": 0.0014726355179188301,
+      "loss": 1.3075,
+      "step": 6576
+    },
+    {
+      "epoch": 0.505884162756711,
+      "learning_rate": 0.0014722731154632187,
+      "loss": 1.2026,
+      "step": 6577
+    },
+    {
+      "epoch": 0.5059610799169295,
+      "learning_rate": 0.0014719107146266082,
+      "loss": 1.096,
+      "step": 6578
+    },
+    {
+      "epoch": 0.5060379970771479,
+      "learning_rate": 0.0014715483154301614,
+      "loss": 0.7804,
+      "step": 6579
+    },
+    {
+      "epoch": 0.5061149142373663,
+      "learning_rate": 0.0014711859178950378,
+      "loss": 0.9692,
+      "step": 6580
+    },
+    {
+      "epoch": 0.5061918313975848,
+      "learning_rate": 0.001470823522042398,
+      "loss": 1.1861,
+      "step": 6581
+    },
+    {
+      "epoch": 0.5062687485578032,
+      "learning_rate": 0.0014704611278934035,
+      "loss": 1.0058,
+      "step": 6582
+    },
+    {
+      "epoch": 0.5063456657180216,
+      "learning_rate": 0.0014700987354692146,
+      "loss": 1.5371,
+      "step": 6583
+    },
+    {
+      "epoch": 0.5064225828782402,
+      "learning_rate": 0.0014697363447909912,
+      "loss": 1.0502,
+      "step": 6584
+    },
+    {
+      "epoch": 0.5064995000384586,
+      "learning_rate": 0.0014693739558798947,
+      "loss": 1.1506,
+      "step": 6585
+    },
+    {
+      "epoch": 0.5065764171986771,
+      "learning_rate": 0.0014690115687570845,
+      "loss": 1.1711,
+      "step": 6586
+    },
+    {
+      "epoch": 0.5066533343588955,
+      "learning_rate": 0.0014686491834437214,
+      "loss": 1.0638,
+      "step": 6587
+    },
+    {
+      "epoch": 0.5067302515191139,
+      "learning_rate": 0.0014682867999609654,
+      "loss": 1.0009,
+      "step": 6588
+    },
+    {
+      "epoch": 0.5068071686793324,
+      "learning_rate": 0.0014679244183299753,
+      "loss": 1.1536,
+      "step": 6589
+    },
+    {
+      "epoch": 0.5068840858395508,
+      "learning_rate": 0.0014675620385719128,
+      "loss": 0.9509,
+      "step": 6590
+    },
+    {
+      "epoch": 0.5069610029997692,
+      "learning_rate": 0.0014671996607079365,
+      "loss": 0.9983,
+      "step": 6591
+    },
+    {
+      "epoch": 0.5070379201599877,
+      "learning_rate": 0.0014668372847592055,
+      "loss": 1.2864,
+      "step": 6592
+    },
+    {
+      "epoch": 0.5071148373202061,
+      "learning_rate": 0.0014664749107468804,
+      "loss": 1.3432,
+      "step": 6593
+    },
+    {
+      "epoch": 0.5071917544804245,
+      "learning_rate": 0.0014661125386921205,
+      "loss": 1.1032,
+      "step": 6594
+    },
+    {
+      "epoch": 0.507268671640643,
+      "learning_rate": 0.0014657501686160842,
+      "loss": 1.4989,
+      "step": 6595
+    },
+    {
+      "epoch": 0.5073455888008614,
+      "learning_rate": 0.0014653878005399317,
+      "loss": 1.3576,
+      "step": 6596
+    },
+    {
+      "epoch": 0.50742250596108,
+      "learning_rate": 0.0014650254344848212,
+      "loss": 1.3586,
+      "step": 6597
+    },
+    {
+      "epoch": 0.5074994231212984,
+      "learning_rate": 0.0014646630704719123,
+      "loss": 1.2624,
+      "step": 6598
+    },
+    {
+      "epoch": 0.5075763402815168,
+      "learning_rate": 0.001464300708522364,
+      "loss": 1.0322,
+      "step": 6599
+    },
+    {
+      "epoch": 0.5076532574417353,
+      "learning_rate": 0.0014639383486573336,
+      "loss": 0.835,
+      "step": 6600
+    },
+    {
+      "epoch": 0.5077301746019537,
+      "learning_rate": 0.0014635759908979815,
+      "loss": 1.3156,
+      "step": 6601
+    },
+    {
+      "epoch": 0.5078070917621721,
+      "learning_rate": 0.001463213635265465,
+      "loss": 1.2155,
+      "step": 6602
+    },
+    {
+      "epoch": 0.5078840089223906,
+      "learning_rate": 0.0014628512817809422,
+      "loss": 1.1603,
+      "step": 6603
+    },
+    {
+      "epoch": 0.507960926082609,
+      "learning_rate": 0.001462488930465572,
+      "loss": 0.8567,
+      "step": 6604
+    },
+    {
+      "epoch": 0.5080378432428275,
+      "learning_rate": 0.0014621265813405122,
+      "loss": 1.3782,
+      "step": 6605
+    },
+    {
+      "epoch": 0.5081147604030459,
+      "learning_rate": 0.0014617642344269212,
+      "loss": 1.0919,
+      "step": 6606
+    },
+    {
+      "epoch": 0.5081916775632643,
+      "learning_rate": 0.001461401889745956,
+      "loss": 1.0697,
+      "step": 6607
+    },
+    {
+      "epoch": 0.5082685947234828,
+      "learning_rate": 0.0014610395473187746,
+      "loss": 1.1664,
+      "step": 6608
+    },
+    {
+      "epoch": 0.5083455118837013,
+      "learning_rate": 0.0014606772071665349,
+      "loss": 1.5752,
+      "step": 6609
+    },
+    {
+      "epoch": 0.5084224290439197,
+      "learning_rate": 0.0014603148693103944,
+      "loss": 1.3914,
+      "step": 6610
+    },
+    {
+      "epoch": 0.5084993462041382,
+      "learning_rate": 0.001459952533771509,
+      "loss": 1.3671,
+      "step": 6611
+    },
+    {
+      "epoch": 0.5085762633643566,
+      "learning_rate": 0.001459590200571038,
+      "loss": 1.3097,
+      "step": 6612
+    },
+    {
+      "epoch": 0.5086531805245751,
+      "learning_rate": 0.0014592278697301362,
+      "loss": 1.1959,
+      "step": 6613
+    },
+    {
+      "epoch": 0.5087300976847935,
+      "learning_rate": 0.0014588655412699626,
+      "loss": 1.3758,
+      "step": 6614
+    },
+    {
+      "epoch": 0.5088070148450119,
+      "learning_rate": 0.0014585032152116726,
+      "loss": 1.2366,
+      "step": 6615
+    },
+    {
+      "epoch": 0.5088839320052304,
+      "learning_rate": 0.0014581408915764228,
+      "loss": 1.2483,
+      "step": 6616
+    },
+    {
+      "epoch": 0.5089608491654488,
+      "learning_rate": 0.0014577785703853703,
+      "loss": 0.9804,
+      "step": 6617
+    },
+    {
+      "epoch": 0.5090377663256672,
+      "learning_rate": 0.001457416251659671,
+      "loss": 0.9902,
+      "step": 6618
+    },
+    {
+      "epoch": 0.5091146834858857,
+      "learning_rate": 0.001457053935420481,
+      "loss": 1.6953,
+      "step": 6619
+    },
+    {
+      "epoch": 0.5091916006461041,
+      "learning_rate": 0.0014566916216889566,
+      "loss": 1.0082,
+      "step": 6620
+    },
+    {
+      "epoch": 0.5092685178063225,
+      "learning_rate": 0.001456329310486254,
+      "loss": 1.1904,
+      "step": 6621
+    },
+    {
+      "epoch": 0.509345434966541,
+      "learning_rate": 0.0014559670018335273,
+      "loss": 1.3764,
+      "step": 6622
+    },
+    {
+      "epoch": 0.5094223521267595,
+      "learning_rate": 0.0014556046957519345,
+      "loss": 1.2398,
+      "step": 6623
+    },
+    {
+      "epoch": 0.509499269286978,
+      "learning_rate": 0.0014552423922626284,
+      "loss": 1.2004,
+      "step": 6624
+    },
+    {
+      "epoch": 0.5095761864471964,
+      "learning_rate": 0.001454880091386767,
+      "loss": 1.3528,
+      "step": 6625
+    },
+    {
+      "epoch": 0.5096531036074148,
+      "learning_rate": 0.0014545177931455037,
+      "loss": 1.0543,
+      "step": 6626
+    },
+    {
+      "epoch": 0.5097300207676333,
+      "learning_rate": 0.0014541554975599936,
+      "loss": 1.0902,
+      "step": 6627
+    },
+    {
+      "epoch": 0.5098069379278517,
+      "learning_rate": 0.001453793204651392,
+      "loss": 0.7525,
+      "step": 6628
+    },
+    {
+      "epoch": 0.5098838550880701,
+      "learning_rate": 0.0014534309144408532,
+      "loss": 1.1906,
+      "step": 6629
+    },
+    {
+      "epoch": 0.5099607722482886,
+      "learning_rate": 0.0014530686269495316,
+      "loss": 1.3126,
+      "step": 6630
+    },
+    {
+      "epoch": 0.510037689408507,
+      "learning_rate": 0.0014527063421985822,
+      "loss": 1.408,
+      "step": 6631
+    },
+    {
+      "epoch": 0.5101146065687255,
+      "learning_rate": 0.0014523440602091583,
+      "loss": 1.0331,
+      "step": 6632
+    },
+    {
+      "epoch": 0.5101915237289439,
+      "learning_rate": 0.001451981781002415,
+      "loss": 0.8259,
+      "step": 6633
+    },
+    {
+      "epoch": 0.5102684408891623,
+      "learning_rate": 0.0014516195045995058,
+      "loss": 1.2364,
+      "step": 6634
+    },
+    {
+      "epoch": 0.5103453580493809,
+      "learning_rate": 0.001451257231021583,
+      "loss": 1.0736,
+      "step": 6635
+    },
+    {
+      "epoch": 0.5104222752095993,
+      "learning_rate": 0.0014508949602898022,
+      "loss": 1.1846,
+      "step": 6636
+    },
+    {
+      "epoch": 0.5104991923698177,
+      "learning_rate": 0.0014505326924253157,
+      "loss": 1.4337,
+      "step": 6637
+    },
+    {
+      "epoch": 0.5105761095300362,
+      "learning_rate": 0.0014501704274492765,
+      "loss": 1.5369,
+      "step": 6638
+    },
+    {
+      "epoch": 0.5106530266902546,
+      "learning_rate": 0.0014498081653828382,
+      "loss": 1.0255,
+      "step": 6639
+    },
+    {
+      "epoch": 0.510729943850473,
+      "learning_rate": 0.0014494459062471537,
+      "loss": 1.2547,
+      "step": 6640
+    },
+    {
+      "epoch": 0.5108068610106915,
+      "learning_rate": 0.0014490836500633748,
+      "loss": 0.8558,
+      "step": 6641
+    },
+    {
+      "epoch": 0.5108837781709099,
+      "learning_rate": 0.001448721396852655,
+      "loss": 1.011,
+      "step": 6642
+    },
+    {
+      "epoch": 0.5109606953311284,
+      "learning_rate": 0.001448359146636146,
+      "loss": 0.6644,
+      "step": 6643
+    },
+    {
+      "epoch": 0.5110376124913468,
+      "learning_rate": 0.0014479968994350004,
+      "loss": 1.4301,
+      "step": 6644
+    },
+    {
+      "epoch": 0.5111145296515652,
+      "learning_rate": 0.0014476346552703705,
+      "loss": 1.2024,
+      "step": 6645
+    },
+    {
+      "epoch": 0.5111914468117837,
+      "learning_rate": 0.0014472724141634068,
+      "loss": 1.0296,
+      "step": 6646
+    },
+    {
+      "epoch": 0.5112683639720021,
+      "learning_rate": 0.0014469101761352627,
+      "loss": 1.4159,
+      "step": 6647
+    },
+    {
+      "epoch": 0.5113452811322206,
+      "learning_rate": 0.0014465479412070885,
+      "loss": 1.272,
+      "step": 6648
+    },
+    {
+      "epoch": 0.5114221982924391,
+      "learning_rate": 0.0014461857094000352,
+      "loss": 1.3374,
+      "step": 6649
+    },
+    {
+      "epoch": 0.5114991154526575,
+      "learning_rate": 0.001445823480735255,
+      "loss": 1.012,
+      "step": 6650
+    },
+    {
+      "epoch": 0.511576032612876,
+      "learning_rate": 0.0014454612552338977,
+      "loss": 1.4119,
+      "step": 6651
+    },
+    {
+      "epoch": 0.5116529497730944,
+      "learning_rate": 0.0014450990329171146,
+      "loss": 0.8706,
+      "step": 6652
+    },
+    {
+      "epoch": 0.5117298669333128,
+      "learning_rate": 0.0014447368138060565,
+      "loss": 1.0146,
+      "step": 6653
+    },
+    {
+      "epoch": 0.5118067840935313,
+      "learning_rate": 0.0014443745979218733,
+      "loss": 1.1647,
+      "step": 6654
+    },
+    {
+      "epoch": 0.5118837012537497,
+      "learning_rate": 0.0014440123852857152,
+      "loss": 1.1763,
+      "step": 6655
+    },
+    {
+      "epoch": 0.5119606184139681,
+      "learning_rate": 0.0014436501759187326,
+      "loss": 1.1731,
+      "step": 6656
+    },
+    {
+      "epoch": 0.5120375355741866,
+      "learning_rate": 0.0014432879698420744,
+      "loss": 0.9966,
+      "step": 6657
+    },
+    {
+      "epoch": 0.512114452734405,
+      "learning_rate": 0.0014429257670768912,
+      "loss": 1.0666,
+      "step": 6658
+    },
+    {
+      "epoch": 0.5121913698946234,
+      "learning_rate": 0.0014425635676443322,
+      "loss": 1.2646,
+      "step": 6659
+    },
+    {
+      "epoch": 0.512268287054842,
+      "learning_rate": 0.0014422013715655459,
+      "loss": 1.1913,
+      "step": 6660
+    },
+    {
+      "epoch": 0.5123452042150604,
+      "learning_rate": 0.0014418391788616819,
+      "loss": 1.1502,
+      "step": 6661
+    },
+    {
+      "epoch": 0.5124221213752789,
+      "learning_rate": 0.0014414769895538884,
+      "loss": 1.2237,
+      "step": 6662
+    },
+    {
+      "epoch": 0.5124990385354973,
+      "learning_rate": 0.001441114803663315,
+      "loss": 1.0391,
+      "step": 6663
+    },
+    {
+      "epoch": 0.5125759556957157,
+      "learning_rate": 0.0014407526212111095,
+      "loss": 1.1168,
+      "step": 6664
+    },
+    {
+      "epoch": 0.5126528728559342,
+      "learning_rate": 0.0014403904422184198,
+      "loss": 0.949,
+      "step": 6665
+    },
+    {
+      "epoch": 0.5127297900161526,
+      "learning_rate": 0.001440028266706395,
+      "loss": 1.1032,
+      "step": 6666
+    },
+    {
+      "epoch": 0.512806707176371,
+      "learning_rate": 0.001439666094696182,
+      "loss": 1.0011,
+      "step": 6667
+    },
+    {
+      "epoch": 0.5128836243365895,
+      "learning_rate": 0.001439303926208928,
+      "loss": 1.0922,
+      "step": 6668
+    },
+    {
+      "epoch": 0.5129605414968079,
+      "learning_rate": 0.0014389417612657815,
+      "loss": 1.4588,
+      "step": 6669
+    },
+    {
+      "epoch": 0.5130374586570264,
+      "learning_rate": 0.001438579599887889,
+      "loss": 0.9473,
+      "step": 6670
+    },
+    {
+      "epoch": 0.5131143758172448,
+      "learning_rate": 0.0014382174420963978,
+      "loss": 1.2888,
+      "step": 6671
+    },
+    {
+      "epoch": 0.5131912929774632,
+      "learning_rate": 0.0014378552879124552,
+      "loss": 1.0715,
+      "step": 6672
+    },
+    {
+      "epoch": 0.5132682101376818,
+      "learning_rate": 0.0014374931373572059,
+      "loss": 1.0945,
+      "step": 6673
+    },
+    {
+      "epoch": 0.5133451272979002,
+      "learning_rate": 0.0014371309904517985,
+      "loss": 1.2115,
+      "step": 6674
+    },
+    {
+      "epoch": 0.5134220444581186,
+      "learning_rate": 0.0014367688472173776,
+      "loss": 0.9833,
+      "step": 6675
+    },
+    {
+      "epoch": 0.5134989616183371,
+      "learning_rate": 0.0014364067076750894,
+      "loss": 1.0794,
+      "step": 6676
+    },
+    {
+      "epoch": 0.5135758787785555,
+      "learning_rate": 0.00143604457184608,
+      "loss": 0.9514,
+      "step": 6677
+    },
+    {
+      "epoch": 0.5136527959387739,
+      "learning_rate": 0.001435682439751495,
+      "loss": 1.0807,
+      "step": 6678
+    },
+    {
+      "epoch": 0.5137297130989924,
+      "learning_rate": 0.0014353203114124788,
+      "loss": 1.1926,
+      "step": 6679
+    },
+    {
+      "epoch": 0.5138066302592108,
+      "learning_rate": 0.0014349581868501775,
+      "loss": 1.0125,
+      "step": 6680
+    },
+    {
+      "epoch": 0.5138835474194293,
+      "learning_rate": 0.001434596066085735,
+      "loss": 1.0901,
+      "step": 6681
+    },
+    {
+      "epoch": 0.5139604645796477,
+      "learning_rate": 0.0014342339491402968,
+      "loss": 1.1668,
+      "step": 6682
+    },
+    {
+      "epoch": 0.5140373817398661,
+      "learning_rate": 0.001433871836035007,
+      "loss": 1.5974,
+      "step": 6683
+    },
+    {
+      "epoch": 0.5141142989000846,
+      "learning_rate": 0.0014335097267910088,
+      "loss": 1.4438,
+      "step": 6684
+    },
+    {
+      "epoch": 0.514191216060303,
+      "learning_rate": 0.001433147621429448,
+      "loss": 1.3482,
+      "step": 6685
+    },
+    {
+      "epoch": 0.5142681332205215,
+      "learning_rate": 0.0014327855199714667,
+      "loss": 1.2907,
+      "step": 6686
+    },
+    {
+      "epoch": 0.51434505038074,
+      "learning_rate": 0.0014324234224382087,
+      "loss": 1.0154,
+      "step": 6687
+    },
+    {
+      "epoch": 0.5144219675409584,
+      "learning_rate": 0.0014320613288508178,
+      "loss": 1.1176,
+      "step": 6688
+    },
+    {
+      "epoch": 0.5144988847011769,
+      "learning_rate": 0.0014316992392304363,
+      "loss": 1.0381,
+      "step": 6689
+    },
+    {
+      "epoch": 0.5145758018613953,
+      "learning_rate": 0.001431337153598208,
+      "loss": 1.1203,
+      "step": 6690
+    },
+    {
+      "epoch": 0.5146527190216137,
+      "learning_rate": 0.0014309750719752746,
+      "loss": 1.1187,
+      "step": 6691
+    },
+    {
+      "epoch": 0.5147296361818322,
+      "learning_rate": 0.0014306129943827785,
+      "loss": 1.0744,
+      "step": 6692
+    },
+    {
+      "epoch": 0.5148065533420506,
+      "learning_rate": 0.0014302509208418623,
+      "loss": 1.0064,
+      "step": 6693
+    },
+    {
+      "epoch": 0.514883470502269,
+      "learning_rate": 0.0014298888513736678,
+      "loss": 1.0792,
+      "step": 6694
+    },
+    {
+      "epoch": 0.5149603876624875,
+      "learning_rate": 0.0014295267859993355,
+      "loss": 0.9389,
+      "step": 6695
+    },
+    {
+      "epoch": 0.5150373048227059,
+      "learning_rate": 0.0014291647247400084,
+      "loss": 1.1881,
+      "step": 6696
+    },
+    {
+      "epoch": 0.5151142219829243,
+      "learning_rate": 0.0014288026676168267,
+      "loss": 0.8957,
+      "step": 6697
+    },
+    {
+      "epoch": 0.5151911391431429,
+      "learning_rate": 0.0014284406146509309,
+      "loss": 1.0012,
+      "step": 6698
+    },
+    {
+      "epoch": 0.5152680563033613,
+      "learning_rate": 0.0014280785658634626,
+      "loss": 0.9666,
+      "step": 6699
+    },
+    {
+      "epoch": 0.5153449734635798,
+      "learning_rate": 0.0014277165212755614,
+      "loss": 0.8031,
+      "step": 6700
+    },
+    {
+      "epoch": 0.5154218906237982,
+      "learning_rate": 0.0014273544809083683,
+      "loss": 1.2961,
+      "step": 6701
+    },
+    {
+      "epoch": 0.5154988077840166,
+      "learning_rate": 0.0014269924447830227,
+      "loss": 0.9829,
+      "step": 6702
+    },
+    {
+      "epoch": 0.5155757249442351,
+      "learning_rate": 0.0014266304129206642,
+      "loss": 1.4266,
+      "step": 6703
+    },
+    {
+      "epoch": 0.5156526421044535,
+      "learning_rate": 0.0014262683853424325,
+      "loss": 1.1295,
+      "step": 6704
+    },
+    {
+      "epoch": 0.5157295592646719,
+      "learning_rate": 0.0014259063620694669,
+      "loss": 0.7262,
+      "step": 6705
+    },
+    {
+      "epoch": 0.5158064764248904,
+      "learning_rate": 0.0014255443431229052,
+      "loss": 1.3216,
+      "step": 6706
+    },
+    {
+      "epoch": 0.5158833935851088,
+      "learning_rate": 0.0014251823285238877,
+      "loss": 1.3511,
+      "step": 6707
+    },
+    {
+      "epoch": 0.5159603107453273,
+      "learning_rate": 0.001424820318293551,
+      "loss": 1.5622,
+      "step": 6708
+    },
+    {
+      "epoch": 0.5160372279055457,
+      "learning_rate": 0.0014244583124530356,
+      "loss": 1.1363,
+      "step": 6709
+    },
+    {
+      "epoch": 0.5161141450657641,
+      "learning_rate": 0.0014240963110234775,
+      "loss": 1.114,
+      "step": 6710
+    },
+    {
+      "epoch": 0.5161910622259827,
+      "learning_rate": 0.0014237343140260145,
+      "loss": 1.316,
+      "step": 6711
+    },
+    {
+      "epoch": 0.5162679793862011,
+      "learning_rate": 0.0014233723214817848,
+      "loss": 1.119,
+      "step": 6712
+    },
+    {
+      "epoch": 0.5163448965464195,
+      "learning_rate": 0.0014230103334119252,
+      "loss": 1.3046,
+      "step": 6713
+    },
+    {
+      "epoch": 0.516421813706638,
+      "learning_rate": 0.001422648349837572,
+      "loss": 1.1005,
+      "step": 6714
+    },
+    {
+      "epoch": 0.5164987308668564,
+      "learning_rate": 0.0014222863707798626,
+      "loss": 1.3312,
+      "step": 6715
+    },
+    {
+      "epoch": 0.5165756480270748,
+      "learning_rate": 0.0014219243962599333,
+      "loss": 1.1983,
+      "step": 6716
+    },
+    {
+      "epoch": 0.5166525651872933,
+      "learning_rate": 0.001421562426298919,
+      "loss": 0.9307,
+      "step": 6717
+    },
+    {
+      "epoch": 0.5167294823475117,
+      "learning_rate": 0.0014212004609179573,
+      "loss": 1.1534,
+      "step": 6718
+    },
+    {
+      "epoch": 0.5168063995077302,
+      "learning_rate": 0.001420838500138182,
+      "loss": 0.843,
+      "step": 6719
+    },
+    {
+      "epoch": 0.5168833166679486,
+      "learning_rate": 0.00142047654398073,
+      "loss": 1.1494,
+      "step": 6720
+    },
+    {
+      "epoch": 0.516960233828167,
+      "learning_rate": 0.001420114592466735,
+      "loss": 1.1551,
+      "step": 6721
+    },
+    {
+      "epoch": 0.5170371509883855,
+      "learning_rate": 0.001419752645617332,
+      "loss": 1.0119,
+      "step": 6722
+    },
+    {
+      "epoch": 0.517114068148604,
+      "learning_rate": 0.0014193907034536562,
+      "loss": 1.1552,
+      "step": 6723
+    },
+    {
+      "epoch": 0.5171909853088223,
+      "learning_rate": 0.001419028765996841,
+      "loss": 1.1905,
+      "step": 6724
+    },
+    {
+      "epoch": 0.5172679024690409,
+      "learning_rate": 0.0014186668332680204,
+      "loss": 1.2755,
+      "step": 6725
+    },
+    {
+      "epoch": 0.5173448196292593,
+      "learning_rate": 0.0014183049052883282,
+      "loss": 0.8684,
+      "step": 6726
+    },
+    {
+      "epoch": 0.5174217367894778,
+      "learning_rate": 0.0014179429820788977,
+      "loss": 1.0045,
+      "step": 6727
+    },
+    {
+      "epoch": 0.5174986539496962,
+      "learning_rate": 0.001417581063660862,
+      "loss": 1.4103,
+      "step": 6728
+    },
+    {
+      "epoch": 0.5175755711099146,
+      "learning_rate": 0.0014172191500553544,
+      "loss": 1.1899,
+      "step": 6729
+    },
+    {
+      "epoch": 0.5176524882701331,
+      "learning_rate": 0.001416857241283506,
+      "loss": 1.3045,
+      "step": 6730
+    },
+    {
+      "epoch": 0.5177294054303515,
+      "learning_rate": 0.0014164953373664508,
+      "loss": 1.2077,
+      "step": 6731
+    },
+    {
+      "epoch": 0.5178063225905699,
+      "learning_rate": 0.0014161334383253196,
+      "loss": 1.3035,
+      "step": 6732
+    },
+    {
+      "epoch": 0.5178832397507884,
+      "learning_rate": 0.001415771544181244,
+      "loss": 1.2366,
+      "step": 6733
+    },
+    {
+      "epoch": 0.5179601569110068,
+      "learning_rate": 0.0014154096549553559,
+      "loss": 1.2346,
+      "step": 6734
+    },
+    {
+      "epoch": 0.5180370740712253,
+      "learning_rate": 0.0014150477706687862,
+      "loss": 0.8414,
+      "step": 6735
+    },
+    {
+      "epoch": 0.5181139912314437,
+      "learning_rate": 0.0014146858913426656,
+      "loss": 0.974,
+      "step": 6736
+    },
+    {
+      "epoch": 0.5181909083916622,
+      "learning_rate": 0.0014143240169981251,
+      "loss": 0.9438,
+      "step": 6737
+    },
+    {
+      "epoch": 0.5182678255518807,
+      "learning_rate": 0.001413962147656294,
+      "loss": 1.0879,
+      "step": 6738
+    },
+    {
+      "epoch": 0.5183447427120991,
+      "learning_rate": 0.001413600283338303,
+      "loss": 1.079,
+      "step": 6739
+    },
+    {
+      "epoch": 0.5184216598723175,
+      "learning_rate": 0.001413238424065282,
+      "loss": 1.144,
+      "step": 6740
+    },
+    {
+      "epoch": 0.518498577032536,
+      "learning_rate": 0.001412876569858359,
+      "loss": 1.3145,
+      "step": 6741
+    },
+    {
+      "epoch": 0.5185754941927544,
+      "learning_rate": 0.0014125147207386646,
+      "loss": 1.2907,
+      "step": 6742
+    },
+    {
+      "epoch": 0.5186524113529728,
+      "learning_rate": 0.0014121528767273267,
+      "loss": 1.1897,
+      "step": 6743
+    },
+    {
+      "epoch": 0.5187293285131913,
+      "learning_rate": 0.0014117910378454734,
+      "loss": 1.2253,
+      "step": 6744
+    },
+    {
+      "epoch": 0.5188062456734097,
+      "learning_rate": 0.0014114292041142336,
+      "loss": 1.3954,
+      "step": 6745
+    },
+    {
+      "epoch": 0.5188831628336282,
+      "learning_rate": 0.0014110673755547349,
+      "loss": 0.9285,
+      "step": 6746
+    },
+    {
+      "epoch": 0.5189600799938466,
+      "learning_rate": 0.0014107055521881049,
+      "loss": 1.0801,
+      "step": 6747
+    },
+    {
+      "epoch": 0.519036997154065,
+      "learning_rate": 0.0014103437340354707,
+      "loss": 0.8494,
+      "step": 6748
+    },
+    {
+      "epoch": 0.5191139143142836,
+      "learning_rate": 0.001409981921117959,
+      "loss": 1.2647,
+      "step": 6749
+    },
+    {
+      "epoch": 0.519190831474502,
+      "learning_rate": 0.0014096201134566973,
+      "loss": 0.828,
+      "step": 6750
+    },
+    {
+      "epoch": 0.5192677486347204,
+      "learning_rate": 0.0014092583110728114,
+      "loss": 1.0505,
+      "step": 6751
+    },
+    {
+      "epoch": 0.5193446657949389,
+      "learning_rate": 0.0014088965139874264,
+      "loss": 1.2376,
+      "step": 6752
+    },
+    {
+      "epoch": 0.5194215829551573,
+      "learning_rate": 0.00140853472222167,
+      "loss": 1.5328,
+      "step": 6753
+    },
+    {
+      "epoch": 0.5194985001153758,
+      "learning_rate": 0.0014081729357966655,
+      "loss": 1.3196,
+      "step": 6754
+    },
+    {
+      "epoch": 0.5195754172755942,
+      "learning_rate": 0.0014078111547335397,
+      "loss": 1.0608,
+      "step": 6755
+    },
+    {
+      "epoch": 0.5196523344358126,
+      "learning_rate": 0.0014074493790534165,
+      "loss": 1.0171,
+      "step": 6756
+    },
+    {
+      "epoch": 0.5197292515960311,
+      "learning_rate": 0.0014070876087774202,
+      "loss": 1.4048,
+      "step": 6757
+    },
+    {
+      "epoch": 0.5198061687562495,
+      "learning_rate": 0.0014067258439266757,
+      "loss": 1.2421,
+      "step": 6758
+    },
+    {
+      "epoch": 0.5198830859164679,
+      "learning_rate": 0.001406364084522306,
+      "loss": 0.8108,
+      "step": 6759
+    },
+    {
+      "epoch": 0.5199600030766864,
+      "learning_rate": 0.001406002330585435,
+      "loss": 1.175,
+      "step": 6760
+    },
+    {
+      "epoch": 0.5200369202369048,
+      "learning_rate": 0.0014056405821371862,
+      "loss": 1.3451,
+      "step": 6761
+    },
+    {
+      "epoch": 0.5201138373971232,
+      "learning_rate": 0.0014052788391986824,
+      "loss": 1.3186,
+      "step": 6762
+    },
+    {
+      "epoch": 0.5201907545573418,
+      "learning_rate": 0.0014049171017910449,
+      "loss": 0.9307,
+      "step": 6763
+    },
+    {
+      "epoch": 0.5202676717175602,
+      "learning_rate": 0.001404555369935398,
+      "loss": 1.0457,
+      "step": 6764
+    },
+    {
+      "epoch": 0.5203445888777787,
+      "learning_rate": 0.0014041936436528612,
+      "loss": 0.9869,
+      "step": 6765
+    },
+    {
+      "epoch": 0.5204215060379971,
+      "learning_rate": 0.0014038319229645586,
+      "loss": 1.3627,
+      "step": 6766
+    },
+    {
+      "epoch": 0.5204984231982155,
+      "learning_rate": 0.0014034702078916102,
+      "loss": 1.1022,
+      "step": 6767
+    },
+    {
+      "epoch": 0.520575340358434,
+      "learning_rate": 0.0014031084984551362,
+      "loss": 0.9449,
+      "step": 6768
+    },
+    {
+      "epoch": 0.5206522575186524,
+      "learning_rate": 0.0014027467946762585,
+      "loss": 0.9179,
+      "step": 6769
+    },
+    {
+      "epoch": 0.5207291746788708,
+      "learning_rate": 0.0014023850965760969,
+      "loss": 0.9102,
+      "step": 6770
+    },
+    {
+      "epoch": 0.5208060918390893,
+      "learning_rate": 0.0014020234041757706,
+      "loss": 0.8934,
+      "step": 6771
+    },
+    {
+      "epoch": 0.5208830089993077,
+      "learning_rate": 0.0014016617174964004,
+      "loss": 0.9073,
+      "step": 6772
+    },
+    {
+      "epoch": 0.5209599261595262,
+      "learning_rate": 0.0014013000365591046,
+      "loss": 1.0801,
+      "step": 6773
+    },
+    {
+      "epoch": 0.5210368433197446,
+      "learning_rate": 0.0014009383613850029,
+      "loss": 0.8615,
+      "step": 6774
+    },
+    {
+      "epoch": 0.521113760479963,
+      "learning_rate": 0.0014005766919952137,
+      "loss": 0.8591,
+      "step": 6775
+    },
+    {
+      "epoch": 0.5211906776401816,
+      "learning_rate": 0.0014002150284108542,
+      "loss": 0.8669,
+      "step": 6776
+    },
+    {
+      "epoch": 0.5212675948004,
+      "learning_rate": 0.0013998533706530442,
+      "loss": 1.0367,
+      "step": 6777
+    },
+    {
+      "epoch": 0.5213445119606184,
+      "learning_rate": 0.0013994917187429002,
+      "loss": 1.3112,
+      "step": 6778
+    },
+    {
+      "epoch": 0.5214214291208369,
+      "learning_rate": 0.0013991300727015385,
+      "loss": 1.364,
+      "step": 6779
+    },
+    {
+      "epoch": 0.5214983462810553,
+      "learning_rate": 0.0013987684325500779,
+      "loss": 1.0634,
+      "step": 6780
+    },
+    {
+      "epoch": 0.5215752634412737,
+      "learning_rate": 0.0013984067983096337,
+      "loss": 1.1844,
+      "step": 6781
+    },
+    {
+      "epoch": 0.5216521806014922,
+      "learning_rate": 0.0013980451700013222,
+      "loss": 1.1443,
+      "step": 6782
+    },
+    {
+      "epoch": 0.5217290977617106,
+      "learning_rate": 0.00139768354764626,
+      "loss": 1.3297,
+      "step": 6783
+    },
+    {
+      "epoch": 0.5218060149219291,
+      "learning_rate": 0.0013973219312655614,
+      "loss": 1.2687,
+      "step": 6784
+    },
+    {
+      "epoch": 0.5218829320821475,
+      "learning_rate": 0.0013969603208803427,
+      "loss": 1.268,
+      "step": 6785
+    },
+    {
+      "epoch": 0.5219598492423659,
+      "learning_rate": 0.0013965987165117186,
+      "loss": 1.2805,
+      "step": 6786
+    },
+    {
+      "epoch": 0.5220367664025845,
+      "learning_rate": 0.0013962371181808022,
+      "loss": 1.2949,
+      "step": 6787
+    },
+    {
+      "epoch": 0.5221136835628029,
+      "learning_rate": 0.0013958755259087098,
+      "loss": 1.1942,
+      "step": 6788
+    },
+    {
+      "epoch": 0.5221906007230213,
+      "learning_rate": 0.0013955139397165534,
+      "loss": 1.2702,
+      "step": 6789
+    },
+    {
+      "epoch": 0.5222675178832398,
+      "learning_rate": 0.0013951523596254466,
+      "loss": 1.4694,
+      "step": 6790
+    },
+    {
+      "epoch": 0.5223444350434582,
+      "learning_rate": 0.0013947907856565033,
+      "loss": 1.4483,
+      "step": 6791
+    },
+    {
+      "epoch": 0.5224213522036767,
+      "learning_rate": 0.0013944292178308352,
+      "loss": 1.2538,
+      "step": 6792
+    },
+    {
+      "epoch": 0.5224982693638951,
+      "learning_rate": 0.0013940676561695555,
+      "loss": 1.3387,
+      "step": 6793
+    },
+    {
+      "epoch": 0.5225751865241135,
+      "learning_rate": 0.0013937061006937756,
+      "loss": 1.1385,
+      "step": 6794
+    },
+    {
+      "epoch": 0.522652103684332,
+      "learning_rate": 0.001393344551424607,
+      "loss": 1.3387,
+      "step": 6795
+    },
+    {
+      "epoch": 0.5227290208445504,
+      "learning_rate": 0.0013929830083831616,
+      "loss": 0.9344,
+      "step": 6796
+    },
+    {
+      "epoch": 0.5228059380047688,
+      "learning_rate": 0.00139262147159055,
+      "loss": 1.1744,
+      "step": 6797
+    },
+    {
+      "epoch": 0.5228828551649873,
+      "learning_rate": 0.001392259941067882,
+      "loss": 0.8946,
+      "step": 6798
+    },
+    {
+      "epoch": 0.5229597723252057,
+      "learning_rate": 0.0013918984168362692,
+      "loss": 1.1324,
+      "step": 6799
+    },
+    {
+      "epoch": 0.5230366894854241,
+      "learning_rate": 0.00139153689891682,
+      "loss": 1.1671,
+      "step": 6800
+    },
+    {
+      "epoch": 0.5231136066456427,
+      "learning_rate": 0.0013911753873306438,
+      "loss": 1.1916,
+      "step": 6801
+    },
+    {
+      "epoch": 0.5231905238058611,
+      "learning_rate": 0.0013908138820988509,
+      "loss": 1.0871,
+      "step": 6802
+    },
+    {
+      "epoch": 0.5232674409660796,
+      "learning_rate": 0.0013904523832425487,
+      "loss": 1.4429,
+      "step": 6803
+    },
+    {
+      "epoch": 0.523344358126298,
+      "learning_rate": 0.0013900908907828462,
+      "loss": 1.2224,
+      "step": 6804
+    },
+    {
+      "epoch": 0.5234212752865164,
+      "learning_rate": 0.0013897294047408514,
+      "loss": 1.1442,
+      "step": 6805
+    },
+    {
+      "epoch": 0.5234981924467349,
+      "learning_rate": 0.001389367925137671,
+      "loss": 1.4412,
+      "step": 6806
+    },
+    {
+      "epoch": 0.5235751096069533,
+      "learning_rate": 0.0013890064519944134,
+      "loss": 1.1639,
+      "step": 6807
+    },
+    {
+      "epoch": 0.5236520267671717,
+      "learning_rate": 0.0013886449853321844,
+      "loss": 0.8159,
+      "step": 6808
+    },
+    {
+      "epoch": 0.5237289439273902,
+      "learning_rate": 0.0013882835251720906,
+      "loss": 1.1721,
+      "step": 6809
+    },
+    {
+      "epoch": 0.5238058610876086,
+      "learning_rate": 0.0013879220715352384,
+      "loss": 1.1325,
+      "step": 6810
+    },
+    {
+      "epoch": 0.5238827782478271,
+      "learning_rate": 0.001387560624442733,
+      "loss": 0.9891,
+      "step": 6811
+    },
+    {
+      "epoch": 0.5239596954080455,
+      "learning_rate": 0.0013871991839156804,
+      "loss": 1.3041,
+      "step": 6812
+    },
+    {
+      "epoch": 0.524036612568264,
+      "learning_rate": 0.0013868377499751852,
+      "loss": 1.0041,
+      "step": 6813
+    },
+    {
+      "epoch": 0.5241135297284825,
+      "learning_rate": 0.001386476322642351,
+      "loss": 1.5879,
+      "step": 6814
+    },
+    {
+      "epoch": 0.5241904468887009,
+      "learning_rate": 0.0013861149019382829,
+      "loss": 1.1554,
+      "step": 6815
+    },
+    {
+      "epoch": 0.5242673640489193,
+      "learning_rate": 0.0013857534878840845,
+      "loss": 1.5309,
+      "step": 6816
+    },
+    {
+      "epoch": 0.5243442812091378,
+      "learning_rate": 0.0013853920805008586,
+      "loss": 1.2539,
+      "step": 6817
+    },
+    {
+      "epoch": 0.5244211983693562,
+      "learning_rate": 0.0013850306798097092,
+      "loss": 1.0906,
+      "step": 6818
+    },
+    {
+      "epoch": 0.5244981155295746,
+      "learning_rate": 0.0013846692858317379,
+      "loss": 0.8813,
+      "step": 6819
+    },
+    {
+      "epoch": 0.5245750326897931,
+      "learning_rate": 0.0013843078985880468,
+      "loss": 0.9908,
+      "step": 6820
+    },
+    {
+      "epoch": 0.5246519498500115,
+      "learning_rate": 0.0013839465180997385,
+      "loss": 1.2657,
+      "step": 6821
+    },
+    {
+      "epoch": 0.52472886701023,
+      "learning_rate": 0.0013835851443879136,
+      "loss": 1.4675,
+      "step": 6822
+    },
+    {
+      "epoch": 0.5248057841704484,
+      "learning_rate": 0.0013832237774736738,
+      "loss": 1.2915,
+      "step": 6823
+    },
+    {
+      "epoch": 0.5248827013306668,
+      "learning_rate": 0.0013828624173781198,
+      "loss": 1.2075,
+      "step": 6824
+    },
+    {
+      "epoch": 0.5249596184908853,
+      "learning_rate": 0.0013825010641223502,
+      "loss": 1.3164,
+      "step": 6825
+    },
+    {
+      "epoch": 0.5250365356511038,
+      "learning_rate": 0.0013821397177274672,
+      "loss": 1.0839,
+      "step": 6826
+    },
+    {
+      "epoch": 0.5251134528113222,
+      "learning_rate": 0.0013817783782145681,
+      "loss": 1.4518,
+      "step": 6827
+    },
+    {
+      "epoch": 0.5251903699715407,
+      "learning_rate": 0.0013814170456047526,
+      "loss": 1.2361,
+      "step": 6828
+    },
+    {
+      "epoch": 0.5252672871317591,
+      "learning_rate": 0.0013810557199191197,
+      "loss": 1.4809,
+      "step": 6829
+    },
+    {
+      "epoch": 0.5253442042919776,
+      "learning_rate": 0.001380694401178767,
+      "loss": 1.3586,
+      "step": 6830
+    },
+    {
+      "epoch": 0.525421121452196,
+      "learning_rate": 0.001380333089404793,
+      "loss": 1.0907,
+      "step": 6831
+    },
+    {
+      "epoch": 0.5254980386124144,
+      "learning_rate": 0.0013799717846182944,
+      "loss": 1.0633,
+      "step": 6832
+    },
+    {
+      "epoch": 0.5255749557726329,
+      "learning_rate": 0.0013796104868403682,
+      "loss": 1.0793,
+      "step": 6833
+    },
+    {
+      "epoch": 0.5256518729328513,
+      "learning_rate": 0.0013792491960921114,
+      "loss": 1.1081,
+      "step": 6834
+    },
+    {
+      "epoch": 0.5257287900930697,
+      "learning_rate": 0.00137888791239462,
+      "loss": 0.8416,
+      "step": 6835
+    },
+    {
+      "epoch": 0.5258057072532882,
+      "learning_rate": 0.001378526635768989,
+      "loss": 1.1909,
+      "step": 6836
+    },
+    {
+      "epoch": 0.5258826244135066,
+      "learning_rate": 0.0013781653662363151,
+      "loss": 1.2034,
+      "step": 6837
+    },
+    {
+      "epoch": 0.5259595415737252,
+      "learning_rate": 0.0013778041038176915,
+      "loss": 1.0422,
+      "step": 6838
+    },
+    {
+      "epoch": 0.5260364587339436,
+      "learning_rate": 0.0013774428485342147,
+      "loss": 1.3183,
+      "step": 6839
+    },
+    {
+      "epoch": 0.526113375894162,
+      "learning_rate": 0.0013770816004069774,
+      "loss": 1.0863,
+      "step": 6840
+    },
+    {
+      "epoch": 0.5261902930543805,
+      "learning_rate": 0.001376720359457073,
+      "loss": 1.3833,
+      "step": 6841
+    },
+    {
+      "epoch": 0.5262672102145989,
+      "learning_rate": 0.0013763591257055958,
+      "loss": 1.0414,
+      "step": 6842
+    },
+    {
+      "epoch": 0.5263441273748173,
+      "learning_rate": 0.001375997899173638,
+      "loss": 1.3602,
+      "step": 6843
+    },
+    {
+      "epoch": 0.5264210445350358,
+      "learning_rate": 0.001375636679882292,
+      "loss": 1.2499,
+      "step": 6844
+    },
+    {
+      "epoch": 0.5264979616952542,
+      "learning_rate": 0.0013752754678526499,
+      "loss": 1.0176,
+      "step": 6845
+    },
+    {
+      "epoch": 0.5265748788554726,
+      "learning_rate": 0.0013749142631058035,
+      "loss": 1.4699,
+      "step": 6846
+    },
+    {
+      "epoch": 0.5266517960156911,
+      "learning_rate": 0.001374553065662843,
+      "loss": 1.3072,
+      "step": 6847
+    },
+    {
+      "epoch": 0.5267287131759095,
+      "learning_rate": 0.0013741918755448607,
+      "loss": 1.1194,
+      "step": 6848
+    },
+    {
+      "epoch": 0.526805630336128,
+      "learning_rate": 0.0013738306927729444,
+      "loss": 0.7833,
+      "step": 6849
+    },
+    {
+      "epoch": 0.5268825474963464,
+      "learning_rate": 0.001373469517368187,
+      "loss": 0.8987,
+      "step": 6850
+    },
+    {
+      "epoch": 0.5269594646565648,
+      "learning_rate": 0.0013731083493516757,
+      "loss": 1.1682,
+      "step": 6851
+    },
+    {
+      "epoch": 0.5270363818167834,
+      "learning_rate": 0.0013727471887445,
+      "loss": 1.3395,
+      "step": 6852
+    },
+    {
+      "epoch": 0.5271132989770018,
+      "learning_rate": 0.0013723860355677488,
+      "loss": 1.4319,
+      "step": 6853
+    },
+    {
+      "epoch": 0.5271902161372202,
+      "learning_rate": 0.00137202488984251,
+      "loss": 1.2318,
+      "step": 6854
+    },
+    {
+      "epoch": 0.5272671332974387,
+      "learning_rate": 0.0013716637515898708,
+      "loss": 1.2094,
+      "step": 6855
+    },
+    {
+      "epoch": 0.5273440504576571,
+      "learning_rate": 0.0013713026208309192,
+      "loss": 1.1864,
+      "step": 6856
+    },
+    {
+      "epoch": 0.5274209676178756,
+      "learning_rate": 0.0013709414975867416,
+      "loss": 1.1114,
+      "step": 6857
+    },
+    {
+      "epoch": 0.527497884778094,
+      "learning_rate": 0.0013705803818784246,
+      "loss": 1.0764,
+      "step": 6858
+    },
+    {
+      "epoch": 0.5275748019383124,
+      "learning_rate": 0.0013702192737270545,
+      "loss": 0.8763,
+      "step": 6859
+    },
+    {
+      "epoch": 0.5276517190985309,
+      "learning_rate": 0.0013698581731537152,
+      "loss": 1.092,
+      "step": 6860
+    },
+    {
+      "epoch": 0.5277286362587493,
+      "learning_rate": 0.0013694970801794938,
+      "loss": 1.1911,
+      "step": 6861
+    },
+    {
+      "epoch": 0.5278055534189677,
+      "learning_rate": 0.0013691359948254734,
+      "loss": 1.1457,
+      "step": 6862
+    },
+    {
+      "epoch": 0.5278824705791862,
+      "learning_rate": 0.0013687749171127385,
+      "loss": 1.11,
+      "step": 6863
+    },
+    {
+      "epoch": 0.5279593877394047,
+      "learning_rate": 0.0013684138470623731,
+      "loss": 1.1553,
+      "step": 6864
+    },
+    {
+      "epoch": 0.528036304899623,
+      "learning_rate": 0.0013680527846954605,
+      "loss": 1.1272,
+      "step": 6865
+    },
+    {
+      "epoch": 0.5281132220598416,
+      "learning_rate": 0.001367691730033083,
+      "loss": 1.3221,
+      "step": 6866
+    },
+    {
+      "epoch": 0.52819013922006,
+      "learning_rate": 0.0013673306830963234,
+      "loss": 0.8401,
+      "step": 6867
+    },
+    {
+      "epoch": 0.5282670563802785,
+      "learning_rate": 0.0013669696439062632,
+      "loss": 1.0266,
+      "step": 6868
+    },
+    {
+      "epoch": 0.5283439735404969,
+      "learning_rate": 0.0013666086124839844,
+      "loss": 1.0802,
+      "step": 6869
+    },
+    {
+      "epoch": 0.5284208907007153,
+      "learning_rate": 0.0013662475888505683,
+      "loss": 0.9563,
+      "step": 6870
+    },
+    {
+      "epoch": 0.5284978078609338,
+      "learning_rate": 0.0013658865730270936,
+      "loss": 1.1702,
+      "step": 6871
+    },
+    {
+      "epoch": 0.5285747250211522,
+      "learning_rate": 0.001365525565034643,
+      "loss": 1.075,
+      "step": 6872
+    },
+    {
+      "epoch": 0.5286516421813706,
+      "learning_rate": 0.0013651645648942942,
+      "loss": 1.3054,
+      "step": 6873
+    },
+    {
+      "epoch": 0.5287285593415891,
+      "learning_rate": 0.0013648035726271266,
+      "loss": 1.3343,
+      "step": 6874
+    },
+    {
+      "epoch": 0.5288054765018075,
+      "learning_rate": 0.0013644425882542199,
+      "loss": 1.1683,
+      "step": 6875
+    },
+    {
+      "epoch": 0.528882393662026,
+      "learning_rate": 0.0013640816117966514,
+      "loss": 0.9975,
+      "step": 6876
+    },
+    {
+      "epoch": 0.5289593108222445,
+      "learning_rate": 0.0013637206432754994,
+      "loss": 0.877,
+      "step": 6877
+    },
+    {
+      "epoch": 0.5290362279824629,
+      "learning_rate": 0.001363359682711841,
+      "loss": 1.0053,
+      "step": 6878
+    },
+    {
+      "epoch": 0.5291131451426814,
+      "learning_rate": 0.001362998730126753,
+      "loss": 1.2998,
+      "step": 6879
+    },
+    {
+      "epoch": 0.5291900623028998,
+      "learning_rate": 0.0013626377855413122,
+      "loss": 1.1646,
+      "step": 6880
+    },
+    {
+      "epoch": 0.5292669794631182,
+      "learning_rate": 0.0013622768489765945,
+      "loss": 1.3411,
+      "step": 6881
+    },
+    {
+      "epoch": 0.5293438966233367,
+      "learning_rate": 0.0013619159204536742,
+      "loss": 1.1243,
+      "step": 6882
+    },
+    {
+      "epoch": 0.5294208137835551,
+      "learning_rate": 0.0013615549999936281,
+      "loss": 1.1901,
+      "step": 6883
+    },
+    {
+      "epoch": 0.5294977309437735,
+      "learning_rate": 0.0013611940876175298,
+      "loss": 0.9948,
+      "step": 6884
+    },
+    {
+      "epoch": 0.529574648103992,
+      "learning_rate": 0.0013608331833464527,
+      "loss": 1.031,
+      "step": 6885
+    },
+    {
+      "epoch": 0.5296515652642104,
+      "learning_rate": 0.0013604722872014715,
+      "loss": 1.2082,
+      "step": 6886
+    },
+    {
+      "epoch": 0.5297284824244289,
+      "learning_rate": 0.0013601113992036586,
+      "loss": 1.1903,
+      "step": 6887
+    },
+    {
+      "epoch": 0.5298053995846473,
+      "learning_rate": 0.0013597505193740872,
+      "loss": 1.2638,
+      "step": 6888
+    },
+    {
+      "epoch": 0.5298823167448657,
+      "learning_rate": 0.0013593896477338292,
+      "loss": 1.3412,
+      "step": 6889
+    },
+    {
+      "epoch": 0.5299592339050843,
+      "learning_rate": 0.0013590287843039557,
+      "loss": 1.2244,
+      "step": 6890
+    },
+    {
+      "epoch": 0.5300361510653027,
+      "learning_rate": 0.0013586679291055389,
+      "loss": 1.0126,
+      "step": 6891
+    },
+    {
+      "epoch": 0.5301130682255211,
+      "learning_rate": 0.0013583070821596493,
+      "loss": 0.8857,
+      "step": 6892
+    },
+    {
+      "epoch": 0.5301899853857396,
+      "learning_rate": 0.0013579462434873558,
+      "loss": 0.9358,
+      "step": 6893
+    },
+    {
+      "epoch": 0.530266902545958,
+      "learning_rate": 0.0013575854131097301,
+      "loss": 1.2337,
+      "step": 6894
+    },
+    {
+      "epoch": 0.5303438197061765,
+      "learning_rate": 0.0013572245910478398,
+      "loss": 0.8602,
+      "step": 6895
+    },
+    {
+      "epoch": 0.5304207368663949,
+      "learning_rate": 0.0013568637773227552,
+      "loss": 1.2004,
+      "step": 6896
+    },
+    {
+      "epoch": 0.5304976540266133,
+      "learning_rate": 0.0013565029719555437,
+      "loss": 1.0215,
+      "step": 6897
+    },
+    {
+      "epoch": 0.5305745711868318,
+      "learning_rate": 0.0013561421749672727,
+      "loss": 1.0866,
+      "step": 6898
+    },
+    {
+      "epoch": 0.5306514883470502,
+      "learning_rate": 0.0013557813863790103,
+      "loss": 1.3846,
+      "step": 6899
+    },
+    {
+      "epoch": 0.5307284055072686,
+      "learning_rate": 0.0013554206062118233,
+      "loss": 1.0447,
+      "step": 6900
+    },
+    {
+      "epoch": 0.5308053226674871,
+      "learning_rate": 0.0013550598344867772,
+      "loss": 1.0576,
+      "step": 6901
+    },
+    {
+      "epoch": 0.5308822398277055,
+      "learning_rate": 0.001354699071224939,
+      "loss": 0.9923,
+      "step": 6902
+    },
+    {
+      "epoch": 0.530959156987924,
+      "learning_rate": 0.0013543383164473738,
+      "loss": 0.9693,
+      "step": 6903
+    },
+    {
+      "epoch": 0.5310360741481425,
+      "learning_rate": 0.0013539775701751449,
+      "loss": 0.7004,
+      "step": 6904
+    },
+    {
+      "epoch": 0.5311129913083609,
+      "learning_rate": 0.0013536168324293188,
+      "loss": 1.1429,
+      "step": 6905
+    },
+    {
+      "epoch": 0.5311899084685794,
+      "learning_rate": 0.0013532561032309576,
+      "loss": 0.8434,
+      "step": 6906
+    },
+    {
+      "epoch": 0.5312668256287978,
+      "learning_rate": 0.0013528953826011264,
+      "loss": 1.2499,
+      "step": 6907
+    },
+    {
+      "epoch": 0.5313437427890162,
+      "learning_rate": 0.0013525346705608868,
+      "loss": 1.3637,
+      "step": 6908
+    },
+    {
+      "epoch": 0.5314206599492347,
+      "learning_rate": 0.0013521739671313012,
+      "loss": 1.5843,
+      "step": 6909
+    },
+    {
+      "epoch": 0.5314975771094531,
+      "learning_rate": 0.001351813272333432,
+      "loss": 1.0776,
+      "step": 6910
+    },
+    {
+      "epoch": 0.5315744942696715,
+      "learning_rate": 0.0013514525861883402,
+      "loss": 1.336,
+      "step": 6911
+    },
+    {
+      "epoch": 0.53165141142989,
+      "learning_rate": 0.0013510919087170864,
+      "loss": 1.6208,
+      "step": 6912
+    },
+    {
+      "epoch": 0.5317283285901084,
+      "learning_rate": 0.0013507312399407315,
+      "loss": 1.1827,
+      "step": 6913
+    },
+    {
+      "epoch": 0.531805245750327,
+      "learning_rate": 0.001350370579880335,
+      "loss": 1.1826,
+      "step": 6914
+    },
+    {
+      "epoch": 0.5318821629105454,
+      "learning_rate": 0.0013500099285569563,
+      "loss": 1.0885,
+      "step": 6915
+    },
+    {
+      "epoch": 0.5319590800707638,
+      "learning_rate": 0.0013496492859916547,
+      "loss": 1.3705,
+      "step": 6916
+    },
+    {
+      "epoch": 0.5320359972309823,
+      "learning_rate": 0.0013492886522054868,
+      "loss": 1.2497,
+      "step": 6917
+    },
+    {
+      "epoch": 0.5321129143912007,
+      "learning_rate": 0.0013489280272195128,
+      "loss": 1.2999,
+      "step": 6918
+    },
+    {
+      "epoch": 0.5321898315514191,
+      "learning_rate": 0.0013485674110547882,
+      "loss": 0.7348,
+      "step": 6919
+    },
+    {
+      "epoch": 0.5322667487116376,
+      "learning_rate": 0.0013482068037323702,
+      "loss": 1.4769,
+      "step": 6920
+    },
+    {
+      "epoch": 0.532343665871856,
+      "learning_rate": 0.0013478462052733153,
+      "loss": 0.9537,
+      "step": 6921
+    },
+    {
+      "epoch": 0.5324205830320744,
+      "learning_rate": 0.0013474856156986792,
+      "loss": 1.2866,
+      "step": 6922
+    },
+    {
+      "epoch": 0.5324975001922929,
+      "learning_rate": 0.0013471250350295164,
+      "loss": 1.0531,
+      "step": 6923
+    },
+    {
+      "epoch": 0.5325744173525113,
+      "learning_rate": 0.0013467644632868825,
+      "loss": 1.0772,
+      "step": 6924
+    },
+    {
+      "epoch": 0.5326513345127298,
+      "learning_rate": 0.001346403900491831,
+      "loss": 1.0888,
+      "step": 6925
+    },
+    {
+      "epoch": 0.5327282516729482,
+      "learning_rate": 0.001346043346665416,
+      "loss": 1.1258,
+      "step": 6926
+    },
+    {
+      "epoch": 0.5328051688331666,
+      "learning_rate": 0.0013456828018286909,
+      "loss": 1.2859,
+      "step": 6927
+    },
+    {
+      "epoch": 0.5328820859933852,
+      "learning_rate": 0.001345322266002707,
+      "loss": 1.1934,
+      "step": 6928
+    },
+    {
+      "epoch": 0.5329590031536036,
+      "learning_rate": 0.001344961739208518,
+      "loss": 1.1608,
+      "step": 6929
+    },
+    {
+      "epoch": 0.533035920313822,
+      "learning_rate": 0.0013446012214671742,
+      "loss": 1.4867,
+      "step": 6930
+    },
+    {
+      "epoch": 0.5331128374740405,
+      "learning_rate": 0.0013442407127997268,
+      "loss": 0.9409,
+      "step": 6931
+    },
+    {
+      "epoch": 0.5331897546342589,
+      "learning_rate": 0.0013438802132272269,
+      "loss": 1.2508,
+      "step": 6932
+    },
+    {
+      "epoch": 0.5332666717944774,
+      "learning_rate": 0.0013435197227707237,
+      "loss": 1.0573,
+      "step": 6933
+    },
+    {
+      "epoch": 0.5333435889546958,
+      "learning_rate": 0.0013431592414512674,
+      "loss": 1.1462,
+      "step": 6934
+    },
+    {
+      "epoch": 0.5334205061149142,
+      "learning_rate": 0.0013427987692899062,
+      "loss": 1.4857,
+      "step": 6935
+    },
+    {
+      "epoch": 0.5334974232751327,
+      "learning_rate": 0.0013424383063076887,
+      "loss": 1.1281,
+      "step": 6936
+    },
+    {
+      "epoch": 0.5335743404353511,
+      "learning_rate": 0.0013420778525256626,
+      "loss": 1.1199,
+      "step": 6937
+    },
+    {
+      "epoch": 0.5336512575955695,
+      "learning_rate": 0.001341717407964876,
+      "loss": 0.9395,
+      "step": 6938
+    },
+    {
+      "epoch": 0.533728174755788,
+      "learning_rate": 0.0013413569726463735,
+      "loss": 1.01,
+      "step": 6939
+    },
+    {
+      "epoch": 0.5338050919160064,
+      "learning_rate": 0.001340996546591204,
+      "loss": 1.0332,
+      "step": 6940
+    },
+    {
+      "epoch": 0.5338820090762249,
+      "learning_rate": 0.0013406361298204114,
+      "loss": 1.2737,
+      "step": 6941
+    },
+    {
+      "epoch": 0.5339589262364434,
+      "learning_rate": 0.0013402757223550412,
+      "loss": 1.146,
+      "step": 6942
+    },
+    {
+      "epoch": 0.5340358433966618,
+      "learning_rate": 0.001339915324216138,
+      "loss": 0.9185,
+      "step": 6943
+    },
+    {
+      "epoch": 0.5341127605568803,
+      "learning_rate": 0.0013395549354247459,
+      "loss": 1.1744,
+      "step": 6944
+    },
+    {
+      "epoch": 0.5341896777170987,
+      "learning_rate": 0.0013391945560019083,
+      "loss": 0.9265,
+      "step": 6945
+    },
+    {
+      "epoch": 0.5342665948773171,
+      "learning_rate": 0.0013388341859686683,
+      "loss": 1.1616,
+      "step": 6946
+    },
+    {
+      "epoch": 0.5343435120375356,
+      "learning_rate": 0.001338473825346068,
+      "loss": 0.9231,
+      "step": 6947
+    },
+    {
+      "epoch": 0.534420429197754,
+      "learning_rate": 0.0013381134741551493,
+      "loss": 1.2317,
+      "step": 6948
+    },
+    {
+      "epoch": 0.5344973463579724,
+      "learning_rate": 0.0013377531324169538,
+      "loss": 1.4175,
+      "step": 6949
+    },
+    {
+      "epoch": 0.5345742635181909,
+      "learning_rate": 0.0013373928001525217,
+      "loss": 0.8812,
+      "step": 6950
+    },
+    {
+      "epoch": 0.5346511806784093,
+      "learning_rate": 0.001337032477382894,
+      "loss": 1.2232,
+      "step": 6951
+    },
+    {
+      "epoch": 0.5347280978386278,
+      "learning_rate": 0.0013366721641291089,
+      "loss": 1.4312,
+      "step": 6952
+    },
+    {
+      "epoch": 0.5348050149988463,
+      "learning_rate": 0.0013363118604122074,
+      "loss": 1.1784,
+      "step": 6953
+    },
+    {
+      "epoch": 0.5348819321590647,
+      "learning_rate": 0.0013359515662532265,
+      "loss": 1.0246,
+      "step": 6954
+    },
+    {
+      "epoch": 0.5349588493192832,
+      "learning_rate": 0.0013355912816732044,
+      "loss": 0.789,
+      "step": 6955
+    },
+    {
+      "epoch": 0.5350357664795016,
+      "learning_rate": 0.0013352310066931789,
+      "loss": 1.348,
+      "step": 6956
+    },
+    {
+      "epoch": 0.53511268363972,
+      "learning_rate": 0.0013348707413341867,
+      "loss": 1.016,
+      "step": 6957
+    },
+    {
+      "epoch": 0.5351896007999385,
+      "learning_rate": 0.0013345104856172635,
+      "loss": 1.295,
+      "step": 6958
+    },
+    {
+      "epoch": 0.5352665179601569,
+      "learning_rate": 0.0013341502395634462,
+      "loss": 1.066,
+      "step": 6959
+    },
+    {
+      "epoch": 0.5353434351203754,
+      "learning_rate": 0.0013337900031937687,
+      "loss": 1.0589,
+      "step": 6960
+    },
+    {
+      "epoch": 0.5354203522805938,
+      "learning_rate": 0.0013334297765292666,
+      "loss": 1.19,
+      "step": 6961
+    },
+    {
+      "epoch": 0.5354972694408122,
+      "learning_rate": 0.0013330695595909733,
+      "loss": 0.81,
+      "step": 6962
+    },
+    {
+      "epoch": 0.5355741866010307,
+      "learning_rate": 0.001332709352399922,
+      "loss": 1.1576,
+      "step": 6963
+    },
+    {
+      "epoch": 0.5356511037612491,
+      "learning_rate": 0.0013323491549771464,
+      "loss": 1.3964,
+      "step": 6964
+    },
+    {
+      "epoch": 0.5357280209214675,
+      "learning_rate": 0.0013319889673436788,
+      "loss": 1.2666,
+      "step": 6965
+    },
+    {
+      "epoch": 0.535804938081686,
+      "learning_rate": 0.0013316287895205497,
+      "loss": 1.2331,
+      "step": 6966
+    },
+    {
+      "epoch": 0.5358818552419045,
+      "learning_rate": 0.0013312686215287914,
+      "loss": 1.0216,
+      "step": 6967
+    },
+    {
+      "epoch": 0.5359587724021229,
+      "learning_rate": 0.001330908463389434,
+      "loss": 1.2304,
+      "step": 6968
+    },
+    {
+      "epoch": 0.5360356895623414,
+      "learning_rate": 0.0013305483151235076,
+      "loss": 1.0852,
+      "step": 6969
+    },
+    {
+      "epoch": 0.5361126067225598,
+      "learning_rate": 0.001330188176752042,
+      "loss": 0.8415,
+      "step": 6970
+    },
+    {
+      "epoch": 0.5361895238827783,
+      "learning_rate": 0.0013298280482960652,
+      "loss": 1.28,
+      "step": 6971
+    },
+    {
+      "epoch": 0.5362664410429967,
+      "learning_rate": 0.0013294679297766062,
+      "loss": 1.2948,
+      "step": 6972
+    },
+    {
+      "epoch": 0.5363433582032151,
+      "learning_rate": 0.0013291078212146928,
+      "loss": 0.992,
+      "step": 6973
+    },
+    {
+      "epoch": 0.5364202753634336,
+      "learning_rate": 0.0013287477226313514,
+      "loss": 0.9287,
+      "step": 6974
+    },
+    {
+      "epoch": 0.536497192523652,
+      "learning_rate": 0.0013283876340476094,
+      "loss": 1.3024,
+      "step": 6975
+    },
+    {
+      "epoch": 0.5365741096838704,
+      "learning_rate": 0.0013280275554844925,
+      "loss": 1.1904,
+      "step": 6976
+    },
+    {
+      "epoch": 0.5366510268440889,
+      "learning_rate": 0.0013276674869630248,
+      "loss": 1.2229,
+      "step": 6977
+    },
+    {
+      "epoch": 0.5367279440043073,
+      "learning_rate": 0.0013273074285042332,
+      "loss": 1.329,
+      "step": 6978
+    },
+    {
+      "epoch": 0.5368048611645259,
+      "learning_rate": 0.0013269473801291402,
+      "loss": 1.1713,
+      "step": 6979
+    },
+    {
+      "epoch": 0.5368817783247443,
+      "learning_rate": 0.0013265873418587702,
+      "loss": 1.2024,
+      "step": 6980
+    },
+    {
+      "epoch": 0.5369586954849627,
+      "learning_rate": 0.0013262273137141464,
+      "loss": 1.3499,
+      "step": 6981
+    },
+    {
+      "epoch": 0.5370356126451812,
+      "learning_rate": 0.00132586729571629,
+      "loss": 1.3022,
+      "step": 6982
+    },
+    {
+      "epoch": 0.5371125298053996,
+      "learning_rate": 0.0013255072878862245,
+      "loss": 1.024,
+      "step": 6983
+    },
+    {
+      "epoch": 0.537189446965618,
+      "learning_rate": 0.00132514729024497,
+      "loss": 1.078,
+      "step": 6984
+    },
+    {
+      "epoch": 0.5372663641258365,
+      "learning_rate": 0.0013247873028135473,
+      "loss": 1.61,
+      "step": 6985
+    },
+    {
+      "epoch": 0.5373432812860549,
+      "learning_rate": 0.0013244273256129767,
+      "loss": 1.3039,
+      "step": 6986
+    },
+    {
+      "epoch": 0.5374201984462733,
+      "learning_rate": 0.0013240673586642781,
+      "loss": 0.8892,
+      "step": 6987
+    },
+    {
+      "epoch": 0.5374971156064918,
+      "learning_rate": 0.0013237074019884687,
+      "loss": 1.1584,
+      "step": 6988
+    },
+    {
+      "epoch": 0.5375740327667102,
+      "learning_rate": 0.0013233474556065689,
+      "loss": 1.2637,
+      "step": 6989
+    },
+    {
+      "epoch": 0.5376509499269287,
+      "learning_rate": 0.0013229875195395941,
+      "loss": 0.8415,
+      "step": 6990
+    },
+    {
+      "epoch": 0.5377278670871471,
+      "learning_rate": 0.0013226275938085638,
+      "loss": 1.1912,
+      "step": 6991
+    },
+    {
+      "epoch": 0.5378047842473656,
+      "learning_rate": 0.0013222676784344924,
+      "loss": 1.0295,
+      "step": 6992
+    },
+    {
+      "epoch": 0.5378817014075841,
+      "learning_rate": 0.0013219077734383964,
+      "loss": 1.2195,
+      "step": 6993
+    },
+    {
+      "epoch": 0.5379586185678025,
+      "learning_rate": 0.0013215478788412916,
+      "loss": 1.1863,
+      "step": 6994
+    },
+    {
+      "epoch": 0.5380355357280209,
+      "learning_rate": 0.0013211879946641919,
+      "loss": 1.0444,
+      "step": 6995
+    },
+    {
+      "epoch": 0.5381124528882394,
+      "learning_rate": 0.0013208281209281113,
+      "loss": 1.185,
+      "step": 6996
+    },
+    {
+      "epoch": 0.5381893700484578,
+      "learning_rate": 0.0013204682576540637,
+      "loss": 1.2222,
+      "step": 6997
+    },
+    {
+      "epoch": 0.5382662872086763,
+      "learning_rate": 0.0013201084048630614,
+      "loss": 1.0717,
+      "step": 6998
+    },
+    {
+      "epoch": 0.5383432043688947,
+      "learning_rate": 0.001319748562576117,
+      "loss": 0.9739,
+      "step": 6999
+    },
+    {
+      "epoch": 0.5384201215291131,
+      "learning_rate": 0.0013193887308142424,
+      "loss": 1.1365,
+      "step": 7000
+    },
+    {
+      "epoch": 0.5384970386893316,
+      "learning_rate": 0.0013190289095984468,
+      "loss": 1.292,
+      "step": 7001
+    },
+    {
+      "epoch": 0.53857395584955,
+      "learning_rate": 0.001318669098949743,
+      "loss": 1.0203,
+      "step": 7002
+    },
+    {
+      "epoch": 0.5386508730097684,
+      "learning_rate": 0.001318309298889139,
+      "loss": 0.9898,
+      "step": 7003
+    },
+    {
+      "epoch": 0.538727790169987,
+      "learning_rate": 0.001317949509437644,
+      "loss": 1.2468,
+      "step": 7004
+    },
+    {
+      "epoch": 0.5388047073302054,
+      "learning_rate": 0.0013175897306162677,
+      "loss": 1.4557,
+      "step": 7005
+    },
+    {
+      "epoch": 0.5388816244904238,
+      "learning_rate": 0.0013172299624460166,
+      "loss": 0.8888,
+      "step": 7006
+    },
+    {
+      "epoch": 0.5389585416506423,
+      "learning_rate": 0.0013168702049478984,
+      "loss": 0.9741,
+      "step": 7007
+    },
+    {
+      "epoch": 0.5390354588108607,
+      "learning_rate": 0.00131651045814292,
+      "loss": 1.2969,
+      "step": 7008
+    },
+    {
+      "epoch": 0.5391123759710792,
+      "learning_rate": 0.0013161507220520868,
+      "loss": 1.4163,
+      "step": 7009
+    },
+    {
+      "epoch": 0.5391892931312976,
+      "learning_rate": 0.001315790996696405,
+      "loss": 1.3695,
+      "step": 7010
+    },
+    {
+      "epoch": 0.539266210291516,
+      "learning_rate": 0.0013154312820968792,
+      "loss": 1.3334,
+      "step": 7011
+    },
+    {
+      "epoch": 0.5393431274517345,
+      "learning_rate": 0.0013150715782745124,
+      "loss": 1.105,
+      "step": 7012
+    },
+    {
+      "epoch": 0.5394200446119529,
+      "learning_rate": 0.0013147118852503093,
+      "loss": 1.2386,
+      "step": 7013
+    },
+    {
+      "epoch": 0.5394969617721713,
+      "learning_rate": 0.0013143522030452726,
+      "loss": 1.2102,
+      "step": 7014
+    },
+    {
+      "epoch": 0.5395738789323898,
+      "learning_rate": 0.0013139925316804034,
+      "loss": 0.9835,
+      "step": 7015
+    },
+    {
+      "epoch": 0.5396507960926082,
+      "learning_rate": 0.001313632871176705,
+      "loss": 1.3669,
+      "step": 7016
+    },
+    {
+      "epoch": 0.5397277132528268,
+      "learning_rate": 0.0013132732215551768,
+      "loss": 1.003,
+      "step": 7017
+    },
+    {
+      "epoch": 0.5398046304130452,
+      "learning_rate": 0.00131291358283682,
+      "loss": 1.1371,
+      "step": 7018
+    },
+    {
+      "epoch": 0.5398815475732636,
+      "learning_rate": 0.0013125539550426343,
+      "loss": 1.3373,
+      "step": 7019
+    },
+    {
+      "epoch": 0.5399584647334821,
+      "learning_rate": 0.0013121943381936181,
+      "loss": 1.4093,
+      "step": 7020
+    },
+    {
+      "epoch": 0.5400353818937005,
+      "learning_rate": 0.0013118347323107706,
+      "loss": 1.2889,
+      "step": 7021
+    },
+    {
+      "epoch": 0.5401122990539189,
+      "learning_rate": 0.0013114751374150894,
+      "loss": 1.3281,
+      "step": 7022
+    },
+    {
+      "epoch": 0.5401892162141374,
+      "learning_rate": 0.0013111155535275706,
+      "loss": 1.0294,
+      "step": 7023
+    },
+    {
+      "epoch": 0.5402661333743558,
+      "learning_rate": 0.0013107559806692123,
+      "loss": 0.8113,
+      "step": 7024
+    },
+    {
+      "epoch": 0.5403430505345742,
+      "learning_rate": 0.0013103964188610092,
+      "loss": 1.1516,
+      "step": 7025
+    },
+    {
+      "epoch": 0.5404199676947927,
+      "learning_rate": 0.0013100368681239567,
+      "loss": 0.7669,
+      "step": 7026
+    },
+    {
+      "epoch": 0.5404968848550111,
+      "learning_rate": 0.0013096773284790495,
+      "loss": 0.9348,
+      "step": 7027
+    },
+    {
+      "epoch": 0.5405738020152296,
+      "learning_rate": 0.0013093177999472813,
+      "loss": 1.3785,
+      "step": 7028
+    },
+    {
+      "epoch": 0.540650719175448,
+      "learning_rate": 0.0013089582825496456,
+      "loss": 0.9523,
+      "step": 7029
+    },
+    {
+      "epoch": 0.5407276363356665,
+      "learning_rate": 0.0013085987763071351,
+      "loss": 1.2283,
+      "step": 7030
+    },
+    {
+      "epoch": 0.540804553495885,
+      "learning_rate": 0.001308239281240741,
+      "loss": 1.0576,
+      "step": 7031
+    },
+    {
+      "epoch": 0.5408814706561034,
+      "learning_rate": 0.0013078797973714558,
+      "loss": 1.2732,
+      "step": 7032
+    },
+    {
+      "epoch": 0.5409583878163218,
+      "learning_rate": 0.0013075203247202693,
+      "loss": 0.8747,
+      "step": 7033
+    },
+    {
+      "epoch": 0.5410353049765403,
+      "learning_rate": 0.001307160863308171,
+      "loss": 0.8084,
+      "step": 7034
+    },
+    {
+      "epoch": 0.5411122221367587,
+      "learning_rate": 0.0013068014131561517,
+      "loss": 1.3837,
+      "step": 7035
+    },
+    {
+      "epoch": 0.5411891392969772,
+      "learning_rate": 0.0013064419742851983,
+      "loss": 1.02,
+      "step": 7036
+    },
+    {
+      "epoch": 0.5412660564571956,
+      "learning_rate": 0.0013060825467163008,
+      "loss": 0.8834,
+      "step": 7037
+    },
+    {
+      "epoch": 0.541342973617414,
+      "learning_rate": 0.0013057231304704452,
+      "loss": 1.6106,
+      "step": 7038
+    },
+    {
+      "epoch": 0.5414198907776325,
+      "learning_rate": 0.001305363725568618,
+      "loss": 1.4662,
+      "step": 7039
+    },
+    {
+      "epoch": 0.5414968079378509,
+      "learning_rate": 0.001305004332031806,
+      "loss": 1.2767,
+      "step": 7040
+    },
+    {
+      "epoch": 0.5415737250980693,
+      "learning_rate": 0.0013046449498809942,
+      "loss": 1.1696,
+      "step": 7041
+    },
+    {
+      "epoch": 0.5416506422582879,
+      "learning_rate": 0.0013042855791371671,
+      "loss": 1.274,
+      "step": 7042
+    },
+    {
+      "epoch": 0.5417275594185063,
+      "learning_rate": 0.0013039262198213095,
+      "loss": 1.2978,
+      "step": 7043
+    },
+    {
+      "epoch": 0.5418044765787247,
+      "learning_rate": 0.0013035668719544047,
+      "loss": 1.1238,
+      "step": 7044
+    },
+    {
+      "epoch": 0.5418813937389432,
+      "learning_rate": 0.0013032075355574337,
+      "loss": 1.4077,
+      "step": 7045
+    },
+    {
+      "epoch": 0.5419583108991616,
+      "learning_rate": 0.0013028482106513808,
+      "loss": 1.2791,
+      "step": 7046
+    },
+    {
+      "epoch": 0.5420352280593801,
+      "learning_rate": 0.0013024888972572254,
+      "loss": 1.475,
+      "step": 7047
+    },
+    {
+      "epoch": 0.5421121452195985,
+      "learning_rate": 0.0013021295953959502,
+      "loss": 1.2005,
+      "step": 7048
+    },
+    {
+      "epoch": 0.5421890623798169,
+      "learning_rate": 0.0013017703050885337,
+      "loss": 1.0672,
+      "step": 7049
+    },
+    {
+      "epoch": 0.5422659795400354,
+      "learning_rate": 0.0013014110263559555,
+      "loss": 1.0102,
+      "step": 7050
+    },
+    {
+      "epoch": 0.5423428967002538,
+      "learning_rate": 0.0013010517592191946,
+      "loss": 1.1015,
+      "step": 7051
+    },
+    {
+      "epoch": 0.5424198138604722,
+      "learning_rate": 0.0013006925036992288,
+      "loss": 1.1986,
+      "step": 7052
+    },
+    {
+      "epoch": 0.5424967310206907,
+      "learning_rate": 0.0013003332598170352,
+      "loss": 1.0695,
+      "step": 7053
+    },
+    {
+      "epoch": 0.5425736481809091,
+      "learning_rate": 0.001299974027593591,
+      "loss": 1.3769,
+      "step": 7054
+    },
+    {
+      "epoch": 0.5426505653411277,
+      "learning_rate": 0.0012996148070498715,
+      "loss": 0.9897,
+      "step": 7055
+    },
+    {
+      "epoch": 0.5427274825013461,
+      "learning_rate": 0.0012992555982068526,
+      "loss": 1.3528,
+      "step": 7056
+    },
+    {
+      "epoch": 0.5428043996615645,
+      "learning_rate": 0.0012988964010855087,
+      "loss": 1.4178,
+      "step": 7057
+    },
+    {
+      "epoch": 0.542881316821783,
+      "learning_rate": 0.0012985372157068124,
+      "loss": 1.0812,
+      "step": 7058
+    },
+    {
+      "epoch": 0.5429582339820014,
+      "learning_rate": 0.0012981780420917393,
+      "loss": 0.9687,
+      "step": 7059
+    },
+    {
+      "epoch": 0.5430351511422198,
+      "learning_rate": 0.0012978188802612604,
+      "loss": 1.2151,
+      "step": 7060
+    },
+    {
+      "epoch": 0.5431120683024383,
+      "learning_rate": 0.0012974597302363472,
+      "loss": 0.9797,
+      "step": 7061
+    },
+    {
+      "epoch": 0.5431889854626567,
+      "learning_rate": 0.0012971005920379718,
+      "loss": 1.2642,
+      "step": 7062
+    },
+    {
+      "epoch": 0.5432659026228752,
+      "learning_rate": 0.001296741465687104,
+      "loss": 1.4976,
+      "step": 7063
+    },
+    {
+      "epoch": 0.5433428197830936,
+      "learning_rate": 0.0012963823512047142,
+      "loss": 1.1779,
+      "step": 7064
+    },
+    {
+      "epoch": 0.543419736943312,
+      "learning_rate": 0.0012960232486117712,
+      "loss": 1.2746,
+      "step": 7065
+    },
+    {
+      "epoch": 0.5434966541035305,
+      "learning_rate": 0.0012956641579292428,
+      "loss": 1.263,
+      "step": 7066
+    },
+    {
+      "epoch": 0.543573571263749,
+      "learning_rate": 0.0012953050791780977,
+      "loss": 1.3056,
+      "step": 7067
+    },
+    {
+      "epoch": 0.5436504884239673,
+      "learning_rate": 0.0012949460123793026,
+      "loss": 1.1749,
+      "step": 7068
+    },
+    {
+      "epoch": 0.5437274055841859,
+      "learning_rate": 0.0012945869575538226,
+      "loss": 1.3322,
+      "step": 7069
+    },
+    {
+      "epoch": 0.5438043227444043,
+      "learning_rate": 0.0012942279147226251,
+      "loss": 1.2806,
+      "step": 7070
+    },
+    {
+      "epoch": 0.5438812399046227,
+      "learning_rate": 0.0012938688839066739,
+      "loss": 1.1828,
+      "step": 7071
+    },
+    {
+      "epoch": 0.5439581570648412,
+      "learning_rate": 0.0012935098651269327,
+      "loss": 1.3037,
+      "step": 7072
+    },
+    {
+      "epoch": 0.5440350742250596,
+      "learning_rate": 0.0012931508584043663,
+      "loss": 1.0784,
+      "step": 7073
+    },
+    {
+      "epoch": 0.5441119913852781,
+      "learning_rate": 0.0012927918637599368,
+      "loss": 1.3263,
+      "step": 7074
+    },
+    {
+      "epoch": 0.5441889085454965,
+      "learning_rate": 0.001292432881214606,
+      "loss": 1.3612,
+      "step": 7075
+    },
+    {
+      "epoch": 0.5442658257057149,
+      "learning_rate": 0.001292073910789336,
+      "loss": 1.1196,
+      "step": 7076
+    },
+    {
+      "epoch": 0.5443427428659334,
+      "learning_rate": 0.0012917149525050868,
+      "loss": 0.9935,
+      "step": 7077
+    },
+    {
+      "epoch": 0.5444196600261518,
+      "learning_rate": 0.0012913560063828188,
+      "loss": 1.1576,
+      "step": 7078
+    },
+    {
+      "epoch": 0.5444965771863702,
+      "learning_rate": 0.0012909970724434914,
+      "loss": 0.9746,
+      "step": 7079
+    },
+    {
+      "epoch": 0.5445734943465887,
+      "learning_rate": 0.0012906381507080617,
+      "loss": 1.3635,
+      "step": 7080
+    },
+    {
+      "epoch": 0.5446504115068072,
+      "learning_rate": 0.0012902792411974896,
+      "loss": 1.0496,
+      "step": 7081
+    },
+    {
+      "epoch": 0.5447273286670257,
+      "learning_rate": 0.0012899203439327301,
+      "loss": 1.1322,
+      "step": 7082
+    },
+    {
+      "epoch": 0.5448042458272441,
+      "learning_rate": 0.0012895614589347416,
+      "loss": 0.9644,
+      "step": 7083
+    },
+    {
+      "epoch": 0.5448811629874625,
+      "learning_rate": 0.0012892025862244786,
+      "loss": 1.4189,
+      "step": 7084
+    },
+    {
+      "epoch": 0.544958080147681,
+      "learning_rate": 0.0012888437258228962,
+      "loss": 1.4761,
+      "step": 7085
+    },
+    {
+      "epoch": 0.5450349973078994,
+      "learning_rate": 0.0012884848777509486,
+      "loss": 1.3381,
+      "step": 7086
+    },
+    {
+      "epoch": 0.5451119144681178,
+      "learning_rate": 0.0012881260420295896,
+      "loss": 1.1425,
+      "step": 7087
+    },
+    {
+      "epoch": 0.5451888316283363,
+      "learning_rate": 0.0012877672186797715,
+      "loss": 1.0682,
+      "step": 7088
+    },
+    {
+      "epoch": 0.5452657487885547,
+      "learning_rate": 0.001287408407722447,
+      "loss": 1.4989,
+      "step": 7089
+    },
+    {
+      "epoch": 0.5453426659487731,
+      "learning_rate": 0.0012870496091785676,
+      "loss": 1.2307,
+      "step": 7090
+    },
+    {
+      "epoch": 0.5454195831089916,
+      "learning_rate": 0.0012866908230690823,
+      "loss": 1.219,
+      "step": 7091
+    },
+    {
+      "epoch": 0.54549650026921,
+      "learning_rate": 0.0012863320494149431,
+      "loss": 1.0517,
+      "step": 7092
+    },
+    {
+      "epoch": 0.5455734174294286,
+      "learning_rate": 0.0012859732882370973,
+      "loss": 0.8331,
+      "step": 7093
+    },
+    {
+      "epoch": 0.545650334589647,
+      "learning_rate": 0.0012856145395564952,
+      "loss": 1.5791,
+      "step": 7094
+    },
+    {
+      "epoch": 0.5457272517498654,
+      "learning_rate": 0.0012852558033940834,
+      "loss": 0.8782,
+      "step": 7095
+    },
+    {
+      "epoch": 0.5458041689100839,
+      "learning_rate": 0.0012848970797708086,
+      "loss": 1.0444,
+      "step": 7096
+    },
+    {
+      "epoch": 0.5458810860703023,
+      "learning_rate": 0.0012845383687076176,
+      "loss": 1.0185,
+      "step": 7097
+    },
+    {
+      "epoch": 0.5459580032305207,
+      "learning_rate": 0.0012841796702254562,
+      "loss": 1.4756,
+      "step": 7098
+    },
+    {
+      "epoch": 0.5460349203907392,
+      "learning_rate": 0.0012838209843452682,
+      "loss": 1.3353,
+      "step": 7099
+    },
+    {
+      "epoch": 0.5461118375509576,
+      "learning_rate": 0.0012834623110879986,
+      "loss": 1.3954,
+      "step": 7100
+    },
+    {
+      "epoch": 0.5461887547111761,
+      "learning_rate": 0.0012831036504745898,
+      "loss": 1.1106,
+      "step": 7101
+    },
+    {
+      "epoch": 0.5462656718713945,
+      "learning_rate": 0.0012827450025259858,
+      "loss": 0.9238,
+      "step": 7102
+    },
+    {
+      "epoch": 0.5463425890316129,
+      "learning_rate": 0.001282386367263127,
+      "loss": 1.2487,
+      "step": 7103
+    },
+    {
+      "epoch": 0.5464195061918314,
+      "learning_rate": 0.0012820277447069548,
+      "loss": 1.0021,
+      "step": 7104
+    },
+    {
+      "epoch": 0.5464964233520498,
+      "learning_rate": 0.0012816691348784103,
+      "loss": 1.3208,
+      "step": 7105
+    },
+    {
+      "epoch": 0.5465733405122682,
+      "learning_rate": 0.0012813105377984324,
+      "loss": 1.0015,
+      "step": 7106
+    },
+    {
+      "epoch": 0.5466502576724868,
+      "learning_rate": 0.0012809519534879598,
+      "loss": 0.9199,
+      "step": 7107
+    },
+    {
+      "epoch": 0.5467271748327052,
+      "learning_rate": 0.0012805933819679313,
+      "loss": 0.8717,
+      "step": 7108
+    },
+    {
+      "epoch": 0.5468040919929236,
+      "learning_rate": 0.0012802348232592835,
+      "loss": 0.9197,
+      "step": 7109
+    },
+    {
+      "epoch": 0.5468810091531421,
+      "learning_rate": 0.0012798762773829535,
+      "loss": 0.992,
+      "step": 7110
+    },
+    {
+      "epoch": 0.5469579263133605,
+      "learning_rate": 0.001279517744359877,
+      "loss": 1.1453,
+      "step": 7111
+    },
+    {
+      "epoch": 0.547034843473579,
+      "learning_rate": 0.0012791592242109894,
+      "loss": 0.9082,
+      "step": 7112
+    },
+    {
+      "epoch": 0.5471117606337974,
+      "learning_rate": 0.0012788007169572248,
+      "loss": 1.2377,
+      "step": 7113
+    },
+    {
+      "epoch": 0.5471886777940158,
+      "learning_rate": 0.001278442222619517,
+      "loss": 0.9316,
+      "step": 7114
+    },
+    {
+      "epoch": 0.5472655949542343,
+      "learning_rate": 0.0012780837412187983,
+      "loss": 1.4862,
+      "step": 7115
+    },
+    {
+      "epoch": 0.5473425121144527,
+      "learning_rate": 0.0012777252727760016,
+      "loss": 1.0822,
+      "step": 7116
+    },
+    {
+      "epoch": 0.5474194292746711,
+      "learning_rate": 0.0012773668173120581,
+      "loss": 1.1401,
+      "step": 7117
+    },
+    {
+      "epoch": 0.5474963464348896,
+      "learning_rate": 0.0012770083748478978,
+      "loss": 1.0458,
+      "step": 7118
+    },
+    {
+      "epoch": 0.547573263595108,
+      "learning_rate": 0.0012766499454044508,
+      "loss": 1.1473,
+      "step": 7119
+    },
+    {
+      "epoch": 0.5476501807553266,
+      "learning_rate": 0.0012762915290026463,
+      "loss": 1.023,
+      "step": 7120
+    },
+    {
+      "epoch": 0.547727097915545,
+      "learning_rate": 0.0012759331256634128,
+      "loss": 1.2007,
+      "step": 7121
+    },
+    {
+      "epoch": 0.5478040150757634,
+      "learning_rate": 0.0012755747354076775,
+      "loss": 1.1641,
+      "step": 7122
+    },
+    {
+      "epoch": 0.5478809322359819,
+      "learning_rate": 0.0012752163582563669,
+      "loss": 1.3878,
+      "step": 7123
+    },
+    {
+      "epoch": 0.5479578493962003,
+      "learning_rate": 0.0012748579942304078,
+      "loss": 1.1641,
+      "step": 7124
+    },
+    {
+      "epoch": 0.5480347665564187,
+      "learning_rate": 0.0012744996433507252,
+      "loss": 1.1184,
+      "step": 7125
+    },
+    {
+      "epoch": 0.5481116837166372,
+      "learning_rate": 0.0012741413056382431,
+      "loss": 1.3532,
+      "step": 7126
+    },
+    {
+      "epoch": 0.5481886008768556,
+      "learning_rate": 0.0012737829811138857,
+      "loss": 0.9796,
+      "step": 7127
+    },
+    {
+      "epoch": 0.548265518037074,
+      "learning_rate": 0.0012734246697985762,
+      "loss": 1.048,
+      "step": 7128
+    },
+    {
+      "epoch": 0.5483424351972925,
+      "learning_rate": 0.0012730663717132355,
+      "loss": 1.1809,
+      "step": 7129
+    },
+    {
+      "epoch": 0.5484193523575109,
+      "learning_rate": 0.0012727080868787867,
+      "loss": 1.5304,
+      "step": 7130
+    },
+    {
+      "epoch": 0.5484962695177295,
+      "learning_rate": 0.0012723498153161487,
+      "loss": 0.6619,
+      "step": 7131
+    },
+    {
+      "epoch": 0.5485731866779479,
+      "learning_rate": 0.001271991557046243,
+      "loss": 1.0458,
+      "step": 7132
+    },
+    {
+      "epoch": 0.5486501038381663,
+      "learning_rate": 0.0012716333120899876,
+      "loss": 1.1986,
+      "step": 7133
+    },
+    {
+      "epoch": 0.5487270209983848,
+      "learning_rate": 0.0012712750804683012,
+      "loss": 1.0602,
+      "step": 7134
+    },
+    {
+      "epoch": 0.5488039381586032,
+      "learning_rate": 0.0012709168622021011,
+      "loss": 0.8825,
+      "step": 7135
+    },
+    {
+      "epoch": 0.5488808553188216,
+      "learning_rate": 0.0012705586573123044,
+      "loss": 1.0024,
+      "step": 7136
+    },
+    {
+      "epoch": 0.5489577724790401,
+      "learning_rate": 0.0012702004658198263,
+      "loss": 1.1447,
+      "step": 7137
+    },
+    {
+      "epoch": 0.5490346896392585,
+      "learning_rate": 0.001269842287745583,
+      "loss": 1.1722,
+      "step": 7138
+    },
+    {
+      "epoch": 0.549111606799477,
+      "learning_rate": 0.001269484123110488,
+      "loss": 1.1509,
+      "step": 7139
+    },
+    {
+      "epoch": 0.5491885239596954,
+      "learning_rate": 0.0012691259719354554,
+      "loss": 1.1132,
+      "step": 7140
+    },
+    {
+      "epoch": 0.5492654411199138,
+      "learning_rate": 0.0012687678342413985,
+      "loss": 1.3512,
+      "step": 7141
+    },
+    {
+      "epoch": 0.5493423582801323,
+      "learning_rate": 0.0012684097100492278,
+      "loss": 1.3356,
+      "step": 7142
+    },
+    {
+      "epoch": 0.5494192754403507,
+      "learning_rate": 0.0012680515993798564,
+      "loss": 1.1692,
+      "step": 7143
+    },
+    {
+      "epoch": 0.5494961926005691,
+      "learning_rate": 0.0012676935022541934,
+      "loss": 1.1104,
+      "step": 7144
+    },
+    {
+      "epoch": 0.5495731097607877,
+      "learning_rate": 0.0012673354186931488,
+      "loss": 1.212,
+      "step": 7145
+    },
+    {
+      "epoch": 0.5496500269210061,
+      "learning_rate": 0.0012669773487176317,
+      "loss": 1.2503,
+      "step": 7146
+    },
+    {
+      "epoch": 0.5497269440812245,
+      "learning_rate": 0.0012666192923485502,
+      "loss": 1.2845,
+      "step": 7147
+    },
+    {
+      "epoch": 0.549803861241443,
+      "learning_rate": 0.0012662612496068108,
+      "loss": 1.7123,
+      "step": 7148
+    },
+    {
+      "epoch": 0.5498807784016614,
+      "learning_rate": 0.0012659032205133212,
+      "loss": 1.2454,
+      "step": 7149
+    },
+    {
+      "epoch": 0.5499576955618799,
+      "learning_rate": 0.0012655452050889862,
+      "loss": 1.0506,
+      "step": 7150
+    },
+    {
+      "epoch": 0.5500346127220983,
+      "learning_rate": 0.0012651872033547114,
+      "loss": 1.3297,
+      "step": 7151
+    },
+    {
+      "epoch": 0.5501115298823167,
+      "learning_rate": 0.0012648292153314008,
+      "loss": 1.1001,
+      "step": 7152
+    },
+    {
+      "epoch": 0.5501884470425352,
+      "learning_rate": 0.0012644712410399565,
+      "loss": 1.0706,
+      "step": 7153
+    },
+    {
+      "epoch": 0.5502653642027536,
+      "learning_rate": 0.0012641132805012827,
+      "loss": 1.1781,
+      "step": 7154
+    },
+    {
+      "epoch": 0.550342281362972,
+      "learning_rate": 0.00126375533373628,
+      "loss": 1.3003,
+      "step": 7155
+    },
+    {
+      "epoch": 0.5504191985231905,
+      "learning_rate": 0.0012633974007658494,
+      "loss": 1.2581,
+      "step": 7156
+    },
+    {
+      "epoch": 0.550496115683409,
+      "learning_rate": 0.0012630394816108912,
+      "loss": 1.2777,
+      "step": 7157
+    },
+    {
+      "epoch": 0.5505730328436275,
+      "learning_rate": 0.0012626815762923045,
+      "loss": 0.9019,
+      "step": 7158
+    },
+    {
+      "epoch": 0.5506499500038459,
+      "learning_rate": 0.0012623236848309882,
+      "loss": 1.0706,
+      "step": 7159
+    },
+    {
+      "epoch": 0.5507268671640643,
+      "learning_rate": 0.0012619658072478396,
+      "loss": 1.2874,
+      "step": 7160
+    },
+    {
+      "epoch": 0.5508037843242828,
+      "learning_rate": 0.0012616079435637554,
+      "loss": 1.0621,
+      "step": 7161
+    },
+    {
+      "epoch": 0.5508807014845012,
+      "learning_rate": 0.0012612500937996321,
+      "loss": 1.3226,
+      "step": 7162
+    },
+    {
+      "epoch": 0.5509576186447196,
+      "learning_rate": 0.0012608922579763649,
+      "loss": 1.4676,
+      "step": 7163
+    },
+    {
+      "epoch": 0.5510345358049381,
+      "learning_rate": 0.0012605344361148467,
+      "loss": 1.2182,
+      "step": 7164
+    },
+    {
+      "epoch": 0.5511114529651565,
+      "learning_rate": 0.0012601766282359737,
+      "loss": 1.0978,
+      "step": 7165
+    },
+    {
+      "epoch": 0.5511883701253749,
+      "learning_rate": 0.001259818834360637,
+      "loss": 1.3387,
+      "step": 7166
+    },
+    {
+      "epoch": 0.5512652872855934,
+      "learning_rate": 0.0012594610545097282,
+      "loss": 0.8153,
+      "step": 7167
+    },
+    {
+      "epoch": 0.5513422044458118,
+      "learning_rate": 0.0012591032887041397,
+      "loss": 1.6278,
+      "step": 7168
+    },
+    {
+      "epoch": 0.5514191216060303,
+      "learning_rate": 0.001258745536964761,
+      "loss": 1.0743,
+      "step": 7169
+    },
+    {
+      "epoch": 0.5514960387662488,
+      "learning_rate": 0.001258387799312482,
+      "loss": 1.4928,
+      "step": 7170
+    },
+    {
+      "epoch": 0.5515729559264672,
+      "learning_rate": 0.0012580300757681912,
+      "loss": 0.801,
+      "step": 7171
+    },
+    {
+      "epoch": 0.5516498730866857,
+      "learning_rate": 0.001257672366352776,
+      "loss": 1.5612,
+      "step": 7172
+    },
+    {
+      "epoch": 0.5517267902469041,
+      "learning_rate": 0.0012573146710871245,
+      "loss": 0.7027,
+      "step": 7173
+    },
+    {
+      "epoch": 0.5518037074071225,
+      "learning_rate": 0.0012569569899921224,
+      "loss": 1.6275,
+      "step": 7174
+    },
+    {
+      "epoch": 0.551880624567341,
+      "learning_rate": 0.001256599323088654,
+      "loss": 1.2391,
+      "step": 7175
+    },
+    {
+      "epoch": 0.5519575417275594,
+      "learning_rate": 0.0012562416703976059,
+      "loss": 1.2503,
+      "step": 7176
+    },
+    {
+      "epoch": 0.5520344588877779,
+      "learning_rate": 0.0012558840319398595,
+      "loss": 0.9731,
+      "step": 7177
+    },
+    {
+      "epoch": 0.5521113760479963,
+      "learning_rate": 0.0012555264077363,
+      "loss": 0.9332,
+      "step": 7178
+    },
+    {
+      "epoch": 0.5521882932082147,
+      "learning_rate": 0.001255168797807808,
+      "loss": 0.7701,
+      "step": 7179
+    },
+    {
+      "epoch": 0.5522652103684332,
+      "learning_rate": 0.0012548112021752648,
+      "loss": 1.1712,
+      "step": 7180
+    },
+    {
+      "epoch": 0.5523421275286516,
+      "learning_rate": 0.0012544536208595513,
+      "loss": 1.1259,
+      "step": 7181
+    },
+    {
+      "epoch": 0.55241904468887,
+      "learning_rate": 0.0012540960538815467,
+      "loss": 1.2825,
+      "step": 7182
+    },
+    {
+      "epoch": 0.5524959618490886,
+      "learning_rate": 0.0012537385012621294,
+      "loss": 1.1437,
+      "step": 7183
+    },
+    {
+      "epoch": 0.552572879009307,
+      "learning_rate": 0.0012533809630221782,
+      "loss": 1.2397,
+      "step": 7184
+    },
+    {
+      "epoch": 0.5526497961695255,
+      "learning_rate": 0.001253023439182569,
+      "loss": 0.9709,
+      "step": 7185
+    },
+    {
+      "epoch": 0.5527267133297439,
+      "learning_rate": 0.0012526659297641793,
+      "loss": 1.7203,
+      "step": 7186
+    },
+    {
+      "epoch": 0.5528036304899623,
+      "learning_rate": 0.0012523084347878838,
+      "loss": 0.988,
+      "step": 7187
+    },
+    {
+      "epoch": 0.5528805476501808,
+      "learning_rate": 0.001251950954274556,
+      "loss": 0.9961,
+      "step": 7188
+    },
+    {
+      "epoch": 0.5529574648103992,
+      "learning_rate": 0.0012515934882450714,
+      "loss": 1.0273,
+      "step": 7189
+    },
+    {
+      "epoch": 0.5530343819706176,
+      "learning_rate": 0.0012512360367203017,
+      "loss": 1.0816,
+      "step": 7190
+    },
+    {
+      "epoch": 0.5531112991308361,
+      "learning_rate": 0.0012508785997211184,
+      "loss": 1.0205,
+      "step": 7191
+    },
+    {
+      "epoch": 0.5531882162910545,
+      "learning_rate": 0.001250521177268394,
+      "loss": 1.035,
+      "step": 7192
+    },
+    {
+      "epoch": 0.5532651334512729,
+      "learning_rate": 0.001250163769382998,
+      "loss": 0.9753,
+      "step": 7193
+    },
+    {
+      "epoch": 0.5533420506114914,
+      "learning_rate": 0.0012498063760857994,
+      "loss": 1.0268,
+      "step": 7194
+    },
+    {
+      "epoch": 0.5534189677717098,
+      "learning_rate": 0.0012494489973976677,
+      "loss": 0.8882,
+      "step": 7195
+    },
+    {
+      "epoch": 0.5534958849319284,
+      "learning_rate": 0.0012490916333394698,
+      "loss": 1.1814,
+      "step": 7196
+    },
+    {
+      "epoch": 0.5535728020921468,
+      "learning_rate": 0.0012487342839320734,
+      "loss": 1.0913,
+      "step": 7197
+    },
+    {
+      "epoch": 0.5536497192523652,
+      "learning_rate": 0.0012483769491963442,
+      "loss": 1.0421,
+      "step": 7198
+    },
+    {
+      "epoch": 0.5537266364125837,
+      "learning_rate": 0.001248019629153146,
+      "loss": 1.2698,
+      "step": 7199
+    },
+    {
+      "epoch": 0.5538035535728021,
+      "learning_rate": 0.0012476623238233454,
+      "loss": 1.0751,
+      "step": 7200
+    },
+    {
+      "epoch": 0.5538804707330205,
+      "learning_rate": 0.0012473050332278044,
+      "loss": 0.8832,
+      "step": 7201
+    },
+    {
+      "epoch": 0.553957387893239,
+      "learning_rate": 0.0012469477573873856,
+      "loss": 1.2194,
+      "step": 7202
+    },
+    {
+      "epoch": 0.5540343050534574,
+      "learning_rate": 0.0012465904963229514,
+      "loss": 1.1259,
+      "step": 7203
+    },
+    {
+      "epoch": 0.5541112222136759,
+      "learning_rate": 0.0012462332500553616,
+      "loss": 1.3259,
+      "step": 7204
+    },
+    {
+      "epoch": 0.5541881393738943,
+      "learning_rate": 0.0012458760186054771,
+      "loss": 1.3579,
+      "step": 7205
+    },
+    {
+      "epoch": 0.5542650565341127,
+      "learning_rate": 0.001245518801994157,
+      "loss": 1.0214,
+      "step": 7206
+    },
+    {
+      "epoch": 0.5543419736943312,
+      "learning_rate": 0.001245161600242259,
+      "loss": 1.3077,
+      "step": 7207
+    },
+    {
+      "epoch": 0.5544188908545497,
+      "learning_rate": 0.0012448044133706408,
+      "loss": 1.1951,
+      "step": 7208
+    },
+    {
+      "epoch": 0.554495808014768,
+      "learning_rate": 0.001244447241400159,
+      "loss": 1.2022,
+      "step": 7209
+    },
+    {
+      "epoch": 0.5545727251749866,
+      "learning_rate": 0.0012440900843516684,
+      "loss": 1.1184,
+      "step": 7210
+    },
+    {
+      "epoch": 0.554649642335205,
+      "learning_rate": 0.0012437329422460256,
+      "loss": 1.0422,
+      "step": 7211
+    },
+    {
+      "epoch": 0.5547265594954234,
+      "learning_rate": 0.0012433758151040829,
+      "loss": 1.0733,
+      "step": 7212
+    },
+    {
+      "epoch": 0.5548034766556419,
+      "learning_rate": 0.0012430187029466933,
+      "loss": 1.2094,
+      "step": 7213
+    },
+    {
+      "epoch": 0.5548803938158603,
+      "learning_rate": 0.0012426616057947101,
+      "loss": 1.262,
+      "step": 7214
+    },
+    {
+      "epoch": 0.5549573109760788,
+      "learning_rate": 0.0012423045236689835,
+      "loss": 0.8998,
+      "step": 7215
+    },
+    {
+      "epoch": 0.5550342281362972,
+      "learning_rate": 0.0012419474565903647,
+      "loss": 1.2346,
+      "step": 7216
+    },
+    {
+      "epoch": 0.5551111452965156,
+      "learning_rate": 0.0012415904045797027,
+      "loss": 0.9243,
+      "step": 7217
+    },
+    {
+      "epoch": 0.5551880624567341,
+      "learning_rate": 0.001241233367657846,
+      "loss": 1.2205,
+      "step": 7218
+    },
+    {
+      "epoch": 0.5552649796169525,
+      "learning_rate": 0.001240876345845643,
+      "loss": 0.9868,
+      "step": 7219
+    },
+    {
+      "epoch": 0.5553418967771709,
+      "learning_rate": 0.0012405193391639405,
+      "loss": 0.7864,
+      "step": 7220
+    },
+    {
+      "epoch": 0.5554188139373895,
+      "learning_rate": 0.0012401623476335833,
+      "loss": 1.3593,
+      "step": 7221
+    },
+    {
+      "epoch": 0.5554957310976079,
+      "learning_rate": 0.0012398053712754183,
+      "loss": 1.4042,
+      "step": 7222
+    },
+    {
+      "epoch": 0.5555726482578264,
+      "learning_rate": 0.001239448410110288,
+      "loss": 1.0608,
+      "step": 7223
+    },
+    {
+      "epoch": 0.5556495654180448,
+      "learning_rate": 0.0012390914641590375,
+      "loss": 1.2384,
+      "step": 7224
+    },
+    {
+      "epoch": 0.5557264825782632,
+      "learning_rate": 0.0012387345334425078,
+      "loss": 1.219,
+      "step": 7225
+    },
+    {
+      "epoch": 0.5558033997384817,
+      "learning_rate": 0.0012383776179815409,
+      "loss": 1.5466,
+      "step": 7226
+    },
+    {
+      "epoch": 0.5558803168987001,
+      "learning_rate": 0.0012380207177969777,
+      "loss": 1.2514,
+      "step": 7227
+    },
+    {
+      "epoch": 0.5559572340589185,
+      "learning_rate": 0.001237663832909658,
+      "loss": 1.0605,
+      "step": 7228
+    },
+    {
+      "epoch": 0.556034151219137,
+      "learning_rate": 0.0012373069633404199,
+      "loss": 1.1861,
+      "step": 7229
+    },
+    {
+      "epoch": 0.5561110683793554,
+      "learning_rate": 0.0012369501091101026,
+      "loss": 1.3284,
+      "step": 7230
+    },
+    {
+      "epoch": 0.5561879855395738,
+      "learning_rate": 0.0012365932702395428,
+      "loss": 1.1109,
+      "step": 7231
+    },
+    {
+      "epoch": 0.5562649026997923,
+      "learning_rate": 0.0012362364467495754,
+      "loss": 1.0749,
+      "step": 7232
+    },
+    {
+      "epoch": 0.5563418198600107,
+      "learning_rate": 0.001235879638661038,
+      "loss": 1.0757,
+      "step": 7233
+    },
+    {
+      "epoch": 0.5564187370202293,
+      "learning_rate": 0.0012355228459947625,
+      "loss": 1.2782,
+      "step": 7234
+    },
+    {
+      "epoch": 0.5564956541804477,
+      "learning_rate": 0.0012351660687715846,
+      "loss": 0.998,
+      "step": 7235
+    },
+    {
+      "epoch": 0.5565725713406661,
+      "learning_rate": 0.0012348093070123358,
+      "loss": 1.2242,
+      "step": 7236
+    },
+    {
+      "epoch": 0.5566494885008846,
+      "learning_rate": 0.0012344525607378475,
+      "loss": 0.8711,
+      "step": 7237
+    },
+    {
+      "epoch": 0.556726405661103,
+      "learning_rate": 0.0012340958299689514,
+      "loss": 1.0938,
+      "step": 7238
+    },
+    {
+      "epoch": 0.5568033228213214,
+      "learning_rate": 0.0012337391147264768,
+      "loss": 1.3598,
+      "step": 7239
+    },
+    {
+      "epoch": 0.5568802399815399,
+      "learning_rate": 0.0012333824150312525,
+      "loss": 1.0687,
+      "step": 7240
+    },
+    {
+      "epoch": 0.5569571571417583,
+      "learning_rate": 0.001233025730904107,
+      "loss": 0.8455,
+      "step": 7241
+    },
+    {
+      "epoch": 0.5570340743019768,
+      "learning_rate": 0.001232669062365867,
+      "loss": 1.1195,
+      "step": 7242
+    },
+    {
+      "epoch": 0.5571109914621952,
+      "learning_rate": 0.0012323124094373598,
+      "loss": 1.5632,
+      "step": 7243
+    },
+    {
+      "epoch": 0.5571879086224136,
+      "learning_rate": 0.0012319557721394098,
+      "loss": 1.1215,
+      "step": 7244
+    },
+    {
+      "epoch": 0.5572648257826321,
+      "learning_rate": 0.001231599150492841,
+      "loss": 1.6228,
+      "step": 7245
+    },
+    {
+      "epoch": 0.5573417429428505,
+      "learning_rate": 0.0012312425445184783,
+      "loss": 1.1486,
+      "step": 7246
+    },
+    {
+      "epoch": 0.557418660103069,
+      "learning_rate": 0.0012308859542371432,
+      "loss": 1.2171,
+      "step": 7247
+    },
+    {
+      "epoch": 0.5574955772632875,
+      "learning_rate": 0.0012305293796696574,
+      "loss": 1.3485,
+      "step": 7248
+    },
+    {
+      "epoch": 0.5575724944235059,
+      "learning_rate": 0.0012301728208368424,
+      "loss": 1.272,
+      "step": 7249
+    },
+    {
+      "epoch": 0.5576494115837243,
+      "learning_rate": 0.0012298162777595175,
+      "loss": 1.0851,
+      "step": 7250
+    },
+    {
+      "epoch": 0.5577263287439428,
+      "learning_rate": 0.0012294597504585014,
+      "loss": 1.2688,
+      "step": 7251
+    },
+    {
+      "epoch": 0.5578032459041612,
+      "learning_rate": 0.0012291032389546127,
+      "loss": 1.1037,
+      "step": 7252
+    },
+    {
+      "epoch": 0.5578801630643797,
+      "learning_rate": 0.001228746743268668,
+      "loss": 1.1284,
+      "step": 7253
+    },
+    {
+      "epoch": 0.5579570802245981,
+      "learning_rate": 0.0012283902634214838,
+      "loss": 1.0962,
+      "step": 7254
+    },
+    {
+      "epoch": 0.5580339973848165,
+      "learning_rate": 0.0012280337994338751,
+      "loss": 1.0351,
+      "step": 7255
+    },
+    {
+      "epoch": 0.558110914545035,
+      "learning_rate": 0.0012276773513266563,
+      "loss": 1.6946,
+      "step": 7256
+    },
+    {
+      "epoch": 0.5581878317052534,
+      "learning_rate": 0.0012273209191206414,
+      "loss": 1.1502,
+      "step": 7257
+    },
+    {
+      "epoch": 0.5582647488654718,
+      "learning_rate": 0.0012269645028366416,
+      "loss": 1.1958,
+      "step": 7258
+    },
+    {
+      "epoch": 0.5583416660256904,
+      "learning_rate": 0.0012266081024954688,
+      "loss": 1.3933,
+      "step": 7259
+    },
+    {
+      "epoch": 0.5584185831859088,
+      "learning_rate": 0.0012262517181179343,
+      "loss": 1.235,
+      "step": 7260
+    },
+    {
+      "epoch": 0.5584955003461273,
+      "learning_rate": 0.0012258953497248468,
+      "loss": 1.2645,
+      "step": 7261
+    },
+    {
+      "epoch": 0.5585724175063457,
+      "learning_rate": 0.001225538997337016,
+      "loss": 1.2166,
+      "step": 7262
+    },
+    {
+      "epoch": 0.5586493346665641,
+      "learning_rate": 0.001225182660975249,
+      "loss": 1.1861,
+      "step": 7263
+    },
+    {
+      "epoch": 0.5587262518267826,
+      "learning_rate": 0.0012248263406603524,
+      "loss": 1.0994,
+      "step": 7264
+    },
+    {
+      "epoch": 0.558803168987001,
+      "learning_rate": 0.001224470036413133,
+      "loss": 0.7398,
+      "step": 7265
+    },
+    {
+      "epoch": 0.5588800861472194,
+      "learning_rate": 0.0012241137482543953,
+      "loss": 1.2622,
+      "step": 7266
+    },
+    {
+      "epoch": 0.5589570033074379,
+      "learning_rate": 0.0012237574762049431,
+      "loss": 0.9521,
+      "step": 7267
+    },
+    {
+      "epoch": 0.5590339204676563,
+      "learning_rate": 0.0012234012202855802,
+      "loss": 0.9145,
+      "step": 7268
+    },
+    {
+      "epoch": 0.5591108376278747,
+      "learning_rate": 0.0012230449805171084,
+      "loss": 0.8893,
+      "step": 7269
+    },
+    {
+      "epoch": 0.5591877547880932,
+      "learning_rate": 0.0012226887569203283,
+      "loss": 1.2631,
+      "step": 7270
+    },
+    {
+      "epoch": 0.5592646719483116,
+      "learning_rate": 0.001222332549516041,
+      "loss": 1.0852,
+      "step": 7271
+    },
+    {
+      "epoch": 0.5593415891085302,
+      "learning_rate": 0.001221976358325045,
+      "loss": 1.3567,
+      "step": 7272
+    },
+    {
+      "epoch": 0.5594185062687486,
+      "learning_rate": 0.0012216201833681398,
+      "loss": 1.3905,
+      "step": 7273
+    },
+    {
+      "epoch": 0.559495423428967,
+      "learning_rate": 0.0012212640246661221,
+      "loss": 0.9227,
+      "step": 7274
+    },
+    {
+      "epoch": 0.5595723405891855,
+      "learning_rate": 0.0012209078822397882,
+      "loss": 1.4219,
+      "step": 7275
+    },
+    {
+      "epoch": 0.5596492577494039,
+      "learning_rate": 0.001220551756109934,
+      "loss": 0.6822,
+      "step": 7276
+    },
+    {
+      "epoch": 0.5597261749096223,
+      "learning_rate": 0.0012201956462973545,
+      "loss": 1.2539,
+      "step": 7277
+    },
+    {
+      "epoch": 0.5598030920698408,
+      "learning_rate": 0.001219839552822842,
+      "loss": 1.2788,
+      "step": 7278
+    },
+    {
+      "epoch": 0.5598800092300592,
+      "learning_rate": 0.0012194834757071906,
+      "loss": 1.1444,
+      "step": 7279
+    },
+    {
+      "epoch": 0.5599569263902777,
+      "learning_rate": 0.001219127414971191,
+      "loss": 1.0939,
+      "step": 7280
+    },
+    {
+      "epoch": 0.5600338435504961,
+      "learning_rate": 0.0012187713706356346,
+      "loss": 1.2404,
+      "step": 7281
+    },
+    {
+      "epoch": 0.5601107607107145,
+      "learning_rate": 0.0012184153427213112,
+      "loss": 1.2991,
+      "step": 7282
+    },
+    {
+      "epoch": 0.560187677870933,
+      "learning_rate": 0.0012180593312490088,
+      "loss": 1.1995,
+      "step": 7283
+    },
+    {
+      "epoch": 0.5602645950311514,
+      "learning_rate": 0.0012177033362395166,
+      "loss": 1.3429,
+      "step": 7284
+    },
+    {
+      "epoch": 0.5603415121913699,
+      "learning_rate": 0.0012173473577136206,
+      "loss": 1.1776,
+      "step": 7285
+    },
+    {
+      "epoch": 0.5604184293515884,
+      "learning_rate": 0.0012169913956921064,
+      "loss": 1.239,
+      "step": 7286
+    },
+    {
+      "epoch": 0.5604953465118068,
+      "learning_rate": 0.00121663545019576,
+      "loss": 1.0809,
+      "step": 7287
+    },
+    {
+      "epoch": 0.5605722636720253,
+      "learning_rate": 0.0012162795212453648,
+      "loss": 1.4057,
+      "step": 7288
+    },
+    {
+      "epoch": 0.5606491808322437,
+      "learning_rate": 0.0012159236088617042,
+      "loss": 1.3556,
+      "step": 7289
+    },
+    {
+      "epoch": 0.5607260979924621,
+      "learning_rate": 0.00121556771306556,
+      "loss": 0.7773,
+      "step": 7290
+    },
+    {
+      "epoch": 0.5608030151526806,
+      "learning_rate": 0.0012152118338777133,
+      "loss": 1.2532,
+      "step": 7291
+    },
+    {
+      "epoch": 0.560879932312899,
+      "learning_rate": 0.0012148559713189447,
+      "loss": 1.1577,
+      "step": 7292
+    },
+    {
+      "epoch": 0.5609568494731174,
+      "learning_rate": 0.0012145001254100336,
+      "loss": 1.322,
+      "step": 7293
+    },
+    {
+      "epoch": 0.5610337666333359,
+      "learning_rate": 0.0012141442961717567,
+      "loss": 1.1817,
+      "step": 7294
+    },
+    {
+      "epoch": 0.5611106837935543,
+      "learning_rate": 0.001213788483624893,
+      "loss": 1.1667,
+      "step": 7295
+    },
+    {
+      "epoch": 0.5611876009537727,
+      "learning_rate": 0.001213432687790218,
+      "loss": 1.346,
+      "step": 7296
+    },
+    {
+      "epoch": 0.5612645181139913,
+      "learning_rate": 0.0012130769086885064,
+      "loss": 1.2367,
+      "step": 7297
+    },
+    {
+      "epoch": 0.5613414352742097,
+      "learning_rate": 0.0012127211463405336,
+      "loss": 0.867,
+      "step": 7298
+    },
+    {
+      "epoch": 0.5614183524344282,
+      "learning_rate": 0.0012123654007670723,
+      "loss": 1.1833,
+      "step": 7299
+    },
+    {
+      "epoch": 0.5614952695946466,
+      "learning_rate": 0.001212009671988895,
+      "loss": 1.3038,
+      "step": 7300
+    },
+    {
+      "epoch": 0.561572186754865,
+      "learning_rate": 0.0012116539600267734,
+      "loss": 1.0554,
+      "step": 7301
+    },
+    {
+      "epoch": 0.5616491039150835,
+      "learning_rate": 0.001211298264901477,
+      "loss": 1.0425,
+      "step": 7302
+    },
+    {
+      "epoch": 0.5617260210753019,
+      "learning_rate": 0.0012109425866337763,
+      "loss": 1.0325,
+      "step": 7303
+    },
+    {
+      "epoch": 0.5618029382355203,
+      "learning_rate": 0.0012105869252444393,
+      "loss": 0.7702,
+      "step": 7304
+    },
+    {
+      "epoch": 0.5618798553957388,
+      "learning_rate": 0.0012102312807542326,
+      "loss": 0.9658,
+      "step": 7305
+    },
+    {
+      "epoch": 0.5619567725559572,
+      "learning_rate": 0.0012098756531839245,
+      "loss": 1.1969,
+      "step": 7306
+    },
+    {
+      "epoch": 0.5620336897161757,
+      "learning_rate": 0.0012095200425542782,
+      "loss": 1.1286,
+      "step": 7307
+    },
+    {
+      "epoch": 0.5621106068763941,
+      "learning_rate": 0.0012091644488860603,
+      "loss": 0.6555,
+      "step": 7308
+    },
+    {
+      "epoch": 0.5621875240366125,
+      "learning_rate": 0.001208808872200033,
+      "loss": 0.943,
+      "step": 7309
+    },
+    {
+      "epoch": 0.562264441196831,
+      "learning_rate": 0.0012084533125169589,
+      "loss": 1.2899,
+      "step": 7310
+    },
+    {
+      "epoch": 0.5623413583570495,
+      "learning_rate": 0.0012080977698576,
+      "loss": 1.1096,
+      "step": 7311
+    },
+    {
+      "epoch": 0.5624182755172679,
+      "learning_rate": 0.0012077422442427165,
+      "loss": 1.2236,
+      "step": 7312
+    },
+    {
+      "epoch": 0.5624951926774864,
+      "learning_rate": 0.0012073867356930676,
+      "loss": 0.8613,
+      "step": 7313
+    },
+    {
+      "epoch": 0.5625721098377048,
+      "learning_rate": 0.0012070312442294124,
+      "loss": 1.2096,
+      "step": 7314
+    },
+    {
+      "epoch": 0.5626490269979232,
+      "learning_rate": 0.0012066757698725088,
+      "loss": 1.2344,
+      "step": 7315
+    },
+    {
+      "epoch": 0.5627259441581417,
+      "learning_rate": 0.0012063203126431116,
+      "loss": 1.5587,
+      "step": 7316
+    },
+    {
+      "epoch": 0.5628028613183601,
+      "learning_rate": 0.0012059648725619783,
+      "loss": 1.2057,
+      "step": 7317
+    },
+    {
+      "epoch": 0.5628797784785786,
+      "learning_rate": 0.0012056094496498615,
+      "loss": 1.0798,
+      "step": 7318
+    },
+    {
+      "epoch": 0.562956695638797,
+      "learning_rate": 0.0012052540439275168,
+      "loss": 1.092,
+      "step": 7319
+    },
+    {
+      "epoch": 0.5630336127990154,
+      "learning_rate": 0.0012048986554156953,
+      "loss": 0.8852,
+      "step": 7320
+    },
+    {
+      "epoch": 0.5631105299592339,
+      "learning_rate": 0.0012045432841351488,
+      "loss": 1.3227,
+      "step": 7321
+    },
+    {
+      "epoch": 0.5631874471194523,
+      "learning_rate": 0.0012041879301066278,
+      "loss": 1.0863,
+      "step": 7322
+    },
+    {
+      "epoch": 0.5632643642796707,
+      "learning_rate": 0.0012038325933508818,
+      "loss": 1.3352,
+      "step": 7323
+    },
+    {
+      "epoch": 0.5633412814398893,
+      "learning_rate": 0.0012034772738886593,
+      "loss": 0.8323,
+      "step": 7324
+    },
+    {
+      "epoch": 0.5634181986001077,
+      "learning_rate": 0.001203121971740708,
+      "loss": 1.1595,
+      "step": 7325
+    },
+    {
+      "epoch": 0.5634951157603262,
+      "learning_rate": 0.001202766686927774,
+      "loss": 0.9934,
+      "step": 7326
+    },
+    {
+      "epoch": 0.5635720329205446,
+      "learning_rate": 0.001202411419470603,
+      "loss": 1.0507,
+      "step": 7327
+    },
+    {
+      "epoch": 0.563648950080763,
+      "learning_rate": 0.0012020561693899397,
+      "loss": 1.1643,
+      "step": 7328
+    },
+    {
+      "epoch": 0.5637258672409815,
+      "learning_rate": 0.0012017009367065264,
+      "loss": 1.1667,
+      "step": 7329
+    },
+    {
+      "epoch": 0.5638027844011999,
+      "learning_rate": 0.001201345721441107,
+      "loss": 0.7244,
+      "step": 7330
+    },
+    {
+      "epoch": 0.5638797015614183,
+      "learning_rate": 0.0012009905236144223,
+      "loss": 1.1365,
+      "step": 7331
+    },
+    {
+      "epoch": 0.5639566187216368,
+      "learning_rate": 0.001200635343247212,
+      "loss": 0.9453,
+      "step": 7332
+    },
+    {
+      "epoch": 0.5640335358818552,
+      "learning_rate": 0.001200280180360216,
+      "loss": 0.9196,
+      "step": 7333
+    },
+    {
+      "epoch": 0.5641104530420736,
+      "learning_rate": 0.001199925034974173,
+      "loss": 1.1879,
+      "step": 7334
+    },
+    {
+      "epoch": 0.5641873702022921,
+      "learning_rate": 0.0011995699071098195,
+      "loss": 1.3436,
+      "step": 7335
+    },
+    {
+      "epoch": 0.5642642873625106,
+      "learning_rate": 0.0011992147967878925,
+      "loss": 1.1447,
+      "step": 7336
+    },
+    {
+      "epoch": 0.5643412045227291,
+      "learning_rate": 0.0011988597040291267,
+      "loss": 1.1356,
+      "step": 7337
+    },
+    {
+      "epoch": 0.5644181216829475,
+      "learning_rate": 0.001198504628854257,
+      "loss": 0.844,
+      "step": 7338
+    },
+    {
+      "epoch": 0.5644950388431659,
+      "learning_rate": 0.0011981495712840165,
+      "loss": 1.4253,
+      "step": 7339
+    },
+    {
+      "epoch": 0.5645719560033844,
+      "learning_rate": 0.0011977945313391362,
+      "loss": 0.9571,
+      "step": 7340
+    },
+    {
+      "epoch": 0.5646488731636028,
+      "learning_rate": 0.001197439509040349,
+      "loss": 1.2405,
+      "step": 7341
+    },
+    {
+      "epoch": 0.5647257903238212,
+      "learning_rate": 0.0011970845044083842,
+      "loss": 1.1636,
+      "step": 7342
+    },
+    {
+      "epoch": 0.5648027074840397,
+      "learning_rate": 0.0011967295174639702,
+      "loss": 1.3601,
+      "step": 7343
+    },
+    {
+      "epoch": 0.5648796246442581,
+      "learning_rate": 0.0011963745482278362,
+      "loss": 1.2052,
+      "step": 7344
+    },
+    {
+      "epoch": 0.5649565418044766,
+      "learning_rate": 0.0011960195967207084,
+      "loss": 1.1753,
+      "step": 7345
+    },
+    {
+      "epoch": 0.565033458964695,
+      "learning_rate": 0.0011956646629633135,
+      "loss": 0.7865,
+      "step": 7346
+    },
+    {
+      "epoch": 0.5651103761249134,
+      "learning_rate": 0.001195309746976376,
+      "loss": 1.497,
+      "step": 7347
+    },
+    {
+      "epoch": 0.565187293285132,
+      "learning_rate": 0.0011949548487806196,
+      "loss": 1.139,
+      "step": 7348
+    },
+    {
+      "epoch": 0.5652642104453504,
+      "learning_rate": 0.001194599968396768,
+      "loss": 1.2221,
+      "step": 7349
+    },
+    {
+      "epoch": 0.5653411276055688,
+      "learning_rate": 0.0011942451058455426,
+      "loss": 1.0699,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5654180447657873,
+      "learning_rate": 0.0011938902611476634,
+      "loss": 1.2305,
+      "step": 7351
+    },
+    {
+      "epoch": 0.5654949619260057,
+      "learning_rate": 0.0011935354343238515,
+      "loss": 1.4869,
+      "step": 7352
+    },
+    {
+      "epoch": 0.5655718790862241,
+      "learning_rate": 0.001193180625394825,
+      "loss": 1.2495,
+      "step": 7353
+    },
+    {
+      "epoch": 0.5656487962464426,
+      "learning_rate": 0.001192825834381301,
+      "loss": 1.107,
+      "step": 7354
+    },
+    {
+      "epoch": 0.565725713406661,
+      "learning_rate": 0.0011924710613039973,
+      "loss": 1.0977,
+      "step": 7355
+    },
+    {
+      "epoch": 0.5658026305668795,
+      "learning_rate": 0.0011921163061836284,
+      "loss": 0.936,
+      "step": 7356
+    },
+    {
+      "epoch": 0.5658795477270979,
+      "learning_rate": 0.0011917615690409092,
+      "loss": 1.2081,
+      "step": 7357
+    },
+    {
+      "epoch": 0.5659564648873163,
+      "learning_rate": 0.0011914068498965536,
+      "loss": 0.9764,
+      "step": 7358
+    },
+    {
+      "epoch": 0.5660333820475348,
+      "learning_rate": 0.0011910521487712732,
+      "loss": 1.202,
+      "step": 7359
+    },
+    {
+      "epoch": 0.5661102992077532,
+      "learning_rate": 0.0011906974656857802,
+      "loss": 1.345,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5661872163679716,
+      "learning_rate": 0.0011903428006607847,
+      "loss": 0.8083,
+      "step": 7361
+    },
+    {
+      "epoch": 0.5662641335281902,
+      "learning_rate": 0.001189988153716995,
+      "loss": 1.1411,
+      "step": 7362
+    },
+    {
+      "epoch": 0.5663410506884086,
+      "learning_rate": 0.0011896335248751209,
+      "loss": 1.2688,
+      "step": 7363
+    },
+    {
+      "epoch": 0.5664179678486271,
+      "learning_rate": 0.001189278914155868,
+      "loss": 1.2281,
+      "step": 7364
+    },
+    {
+      "epoch": 0.5664948850088455,
+      "learning_rate": 0.0011889243215799437,
+      "loss": 0.998,
+      "step": 7365
+    },
+    {
+      "epoch": 0.5665718021690639,
+      "learning_rate": 0.0011885697471680525,
+      "loss": 1.0692,
+      "step": 7366
+    },
+    {
+      "epoch": 0.5666487193292824,
+      "learning_rate": 0.001188215190940898,
+      "loss": 1.176,
+      "step": 7367
+    },
+    {
+      "epoch": 0.5667256364895008,
+      "learning_rate": 0.0011878606529191836,
+      "loss": 1.1958,
+      "step": 7368
+    },
+    {
+      "epoch": 0.5668025536497192,
+      "learning_rate": 0.0011875061331236107,
+      "loss": 1.1545,
+      "step": 7369
+    },
+    {
+      "epoch": 0.5668794708099377,
+      "learning_rate": 0.0011871516315748805,
+      "loss": 1.2527,
+      "step": 7370
+    },
+    {
+      "epoch": 0.5669563879701561,
+      "learning_rate": 0.0011867971482936927,
+      "loss": 1.0219,
+      "step": 7371
+    },
+    {
+      "epoch": 0.5670333051303745,
+      "learning_rate": 0.0011864426833007463,
+      "loss": 1.2691,
+      "step": 7372
+    },
+    {
+      "epoch": 0.567110222290593,
+      "learning_rate": 0.0011860882366167373,
+      "loss": 1.3616,
+      "step": 7373
+    },
+    {
+      "epoch": 0.5671871394508115,
+      "learning_rate": 0.0011857338082623644,
+      "loss": 1.5308,
+      "step": 7374
+    },
+    {
+      "epoch": 0.56726405661103,
+      "learning_rate": 0.0011853793982583207,
+      "loss": 1.2038,
+      "step": 7375
+    },
+    {
+      "epoch": 0.5673409737712484,
+      "learning_rate": 0.001185025006625303,
+      "loss": 0.9733,
+      "step": 7376
+    },
+    {
+      "epoch": 0.5674178909314668,
+      "learning_rate": 0.001184670633384003,
+      "loss": 1.5879,
+      "step": 7377
+    },
+    {
+      "epoch": 0.5674948080916853,
+      "learning_rate": 0.0011843162785551134,
+      "loss": 1.2297,
+      "step": 7378
+    },
+    {
+      "epoch": 0.5675717252519037,
+      "learning_rate": 0.0011839619421593253,
+      "loss": 1.0965,
+      "step": 7379
+    },
+    {
+      "epoch": 0.5676486424121221,
+      "learning_rate": 0.0011836076242173288,
+      "loss": 1.139,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5677255595723406,
+      "learning_rate": 0.0011832533247498123,
+      "loss": 1.139,
+      "step": 7381
+    },
+    {
+      "epoch": 0.567802476732559,
+      "learning_rate": 0.001182899043777465,
+      "loss": 1.253,
+      "step": 7382
+    },
+    {
+      "epoch": 0.5678793938927775,
+      "learning_rate": 0.0011825447813209723,
+      "loss": 1.1121,
+      "step": 7383
+    },
+    {
+      "epoch": 0.5679563110529959,
+      "learning_rate": 0.0011821905374010213,
+      "loss": 1.2031,
+      "step": 7384
+    },
+    {
+      "epoch": 0.5680332282132143,
+      "learning_rate": 0.0011818363120382964,
+      "loss": 0.6744,
+      "step": 7385
+    },
+    {
+      "epoch": 0.5681101453734329,
+      "learning_rate": 0.0011814821052534795,
+      "loss": 0.9784,
+      "step": 7386
+    },
+    {
+      "epoch": 0.5681870625336513,
+      "learning_rate": 0.0011811279170672558,
+      "loss": 1.1948,
+      "step": 7387
+    },
+    {
+      "epoch": 0.5682639796938697,
+      "learning_rate": 0.0011807737475003047,
+      "loss": 0.8045,
+      "step": 7388
+    },
+    {
+      "epoch": 0.5683408968540882,
+      "learning_rate": 0.001180419596573307,
+      "loss": 0.9747,
+      "step": 7389
+    },
+    {
+      "epoch": 0.5684178140143066,
+      "learning_rate": 0.0011800654643069424,
+      "loss": 0.9116,
+      "step": 7390
+    },
+    {
+      "epoch": 0.5684947311745251,
+      "learning_rate": 0.0011797113507218883,
+      "loss": 0.9082,
+      "step": 7391
+    },
+    {
+      "epoch": 0.5685716483347435,
+      "learning_rate": 0.0011793572558388228,
+      "loss": 0.9714,
+      "step": 7392
+    },
+    {
+      "epoch": 0.5686485654949619,
+      "learning_rate": 0.0011790031796784213,
+      "loss": 1.3846,
+      "step": 7393
+    },
+    {
+      "epoch": 0.5687254826551804,
+      "learning_rate": 0.0011786491222613583,
+      "loss": 1.0717,
+      "step": 7394
+    },
+    {
+      "epoch": 0.5688023998153988,
+      "learning_rate": 0.0011782950836083082,
+      "loss": 1.1803,
+      "step": 7395
+    },
+    {
+      "epoch": 0.5688793169756172,
+      "learning_rate": 0.0011779410637399437,
+      "loss": 1.6949,
+      "step": 7396
+    },
+    {
+      "epoch": 0.5689562341358357,
+      "learning_rate": 0.0011775870626769354,
+      "loss": 0.9212,
+      "step": 7397
+    },
+    {
+      "epoch": 0.5690331512960541,
+      "learning_rate": 0.0011772330804399552,
+      "loss": 1.1032,
+      "step": 7398
+    },
+    {
+      "epoch": 0.5691100684562725,
+      "learning_rate": 0.0011768791170496716,
+      "loss": 1.0047,
+      "step": 7399
+    },
+    {
+      "epoch": 0.5691869856164911,
+      "learning_rate": 0.001176525172526753,
+      "loss": 1.2054,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5692639027767095,
+      "learning_rate": 0.0011761712468918667,
+      "loss": 1.2594,
+      "step": 7401
+    },
+    {
+      "epoch": 0.569340819936928,
+      "learning_rate": 0.0011758173401656788,
+      "loss": 1.2848,
+      "step": 7402
+    },
+    {
+      "epoch": 0.5694177370971464,
+      "learning_rate": 0.0011754634523688543,
+      "loss": 1.403,
+      "step": 7403
+    },
+    {
+      "epoch": 0.5694946542573648,
+      "learning_rate": 0.0011751095835220572,
+      "loss": 1.1681,
+      "step": 7404
+    },
+    {
+      "epoch": 0.5695715714175833,
+      "learning_rate": 0.0011747557336459495,
+      "loss": 1.1703,
+      "step": 7405
+    },
+    {
+      "epoch": 0.5696484885778017,
+      "learning_rate": 0.001174401902761194,
+      "loss": 0.9024,
+      "step": 7406
+    },
+    {
+      "epoch": 0.5697254057380201,
+      "learning_rate": 0.0011740480908884509,
+      "loss": 1.1178,
+      "step": 7407
+    },
+    {
+      "epoch": 0.5698023228982386,
+      "learning_rate": 0.0011736942980483788,
+      "loss": 1.1575,
+      "step": 7408
+    },
+    {
+      "epoch": 0.569879240058457,
+      "learning_rate": 0.0011733405242616375,
+      "loss": 1.1706,
+      "step": 7409
+    },
+    {
+      "epoch": 0.5699561572186755,
+      "learning_rate": 0.0011729867695488825,
+      "loss": 1.3981,
+      "step": 7410
+    },
+    {
+      "epoch": 0.570033074378894,
+      "learning_rate": 0.001172633033930772,
+      "loss": 1.055,
+      "step": 7411
+    },
+    {
+      "epoch": 0.5701099915391123,
+      "learning_rate": 0.0011722793174279592,
+      "loss": 1.102,
+      "step": 7412
+    },
+    {
+      "epoch": 0.5701869086993309,
+      "learning_rate": 0.0011719256200610983,
+      "loss": 0.9951,
+      "step": 7413
+    },
+    {
+      "epoch": 0.5702638258595493,
+      "learning_rate": 0.001171571941850843,
+      "loss": 1.2232,
+      "step": 7414
+    },
+    {
+      "epoch": 0.5703407430197677,
+      "learning_rate": 0.0011712182828178437,
+      "loss": 1.2202,
+      "step": 7415
+    },
+    {
+      "epoch": 0.5704176601799862,
+      "learning_rate": 0.0011708646429827517,
+      "loss": 1.0668,
+      "step": 7416
+    },
+    {
+      "epoch": 0.5704945773402046,
+      "learning_rate": 0.0011705110223662162,
+      "loss": 1.3328,
+      "step": 7417
+    },
+    {
+      "epoch": 0.570571494500423,
+      "learning_rate": 0.0011701574209888857,
+      "loss": 1.094,
+      "step": 7418
+    },
+    {
+      "epoch": 0.5706484116606415,
+      "learning_rate": 0.0011698038388714069,
+      "loss": 1.6162,
+      "step": 7419
+    },
+    {
+      "epoch": 0.5707253288208599,
+      "learning_rate": 0.001169450276034426,
+      "loss": 1.0043,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5708022459810784,
+      "learning_rate": 0.0011690967324985882,
+      "loss": 1.1951,
+      "step": 7421
+    },
+    {
+      "epoch": 0.5708791631412968,
+      "learning_rate": 0.0011687432082845368,
+      "loss": 0.7274,
+      "step": 7422
+    },
+    {
+      "epoch": 0.5709560803015152,
+      "learning_rate": 0.0011683897034129155,
+      "loss": 0.8897,
+      "step": 7423
+    },
+    {
+      "epoch": 0.5710329974617337,
+      "learning_rate": 0.001168036217904364,
+      "loss": 1.4012,
+      "step": 7424
+    },
+    {
+      "epoch": 0.5711099146219522,
+      "learning_rate": 0.0011676827517795242,
+      "loss": 1.2045,
+      "step": 7425
+    },
+    {
+      "epoch": 0.5711868317821706,
+      "learning_rate": 0.001167329305059035,
+      "loss": 1.7016,
+      "step": 7426
+    },
+    {
+      "epoch": 0.5712637489423891,
+      "learning_rate": 0.0011669758777635337,
+      "loss": 1.2601,
+      "step": 7427
+    },
+    {
+      "epoch": 0.5713406661026075,
+      "learning_rate": 0.0011666224699136587,
+      "loss": 0.7885,
+      "step": 7428
+    },
+    {
+      "epoch": 0.571417583262826,
+      "learning_rate": 0.0011662690815300446,
+      "loss": 1.3878,
+      "step": 7429
+    },
+    {
+      "epoch": 0.5714945004230444,
+      "learning_rate": 0.001165915712633327,
+      "loss": 1.3918,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5715714175832628,
+      "learning_rate": 0.001165562363244139,
+      "loss": 1.343,
+      "step": 7431
+    },
+    {
+      "epoch": 0.5716483347434813,
+      "learning_rate": 0.0011652090333831128,
+      "loss": 1.037,
+      "step": 7432
+    },
+    {
+      "epoch": 0.5717252519036997,
+      "learning_rate": 0.0011648557230708805,
+      "loss": 1.3406,
+      "step": 7433
+    },
+    {
+      "epoch": 0.5718021690639181,
+      "learning_rate": 0.0011645024323280722,
+      "loss": 1.1148,
+      "step": 7434
+    },
+    {
+      "epoch": 0.5718790862241366,
+      "learning_rate": 0.0011641491611753152,
+      "loss": 1.1338,
+      "step": 7435
+    },
+    {
+      "epoch": 0.571956003384355,
+      "learning_rate": 0.00116379590963324,
+      "loss": 1.2277,
+      "step": 7436
+    },
+    {
+      "epoch": 0.5720329205445734,
+      "learning_rate": 0.0011634426777224717,
+      "loss": 1.5736,
+      "step": 7437
+    },
+    {
+      "epoch": 0.572109837704792,
+      "learning_rate": 0.0011630894654636357,
+      "loss": 1.0534,
+      "step": 7438
+    },
+    {
+      "epoch": 0.5721867548650104,
+      "learning_rate": 0.0011627362728773574,
+      "loss": 1.2428,
+      "step": 7439
+    },
+    {
+      "epoch": 0.5722636720252289,
+      "learning_rate": 0.0011623830999842593,
+      "loss": 1.3041,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5723405891854473,
+      "learning_rate": 0.001162029946804964,
+      "loss": 1.2164,
+      "step": 7441
+    },
+    {
+      "epoch": 0.5724175063456657,
+      "learning_rate": 0.0011616768133600925,
+      "loss": 0.6077,
+      "step": 7442
+    },
+    {
+      "epoch": 0.5724944235058842,
+      "learning_rate": 0.0011613236996702641,
+      "loss": 1.0239,
+      "step": 7443
+    },
+    {
+      "epoch": 0.5725713406661026,
+      "learning_rate": 0.0011609706057560982,
+      "loss": 1.3251,
+      "step": 7444
+    },
+    {
+      "epoch": 0.572648257826321,
+      "learning_rate": 0.001160617531638212,
+      "loss": 0.7548,
+      "step": 7445
+    },
+    {
+      "epoch": 0.5727251749865395,
+      "learning_rate": 0.0011602644773372212,
+      "loss": 1.1027,
+      "step": 7446
+    },
+    {
+      "epoch": 0.5728020921467579,
+      "learning_rate": 0.0011599114428737425,
+      "loss": 1.1049,
+      "step": 7447
+    },
+    {
+      "epoch": 0.5728790093069764,
+      "learning_rate": 0.0011595584282683882,
+      "loss": 1.0308,
+      "step": 7448
+    },
+    {
+      "epoch": 0.5729559264671948,
+      "learning_rate": 0.0011592054335417728,
+      "loss": 1.1353,
+      "step": 7449
+    },
+    {
+      "epoch": 0.5730328436274132,
+      "learning_rate": 0.001158852458714507,
+      "loss": 0.9771,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5731097607876318,
+      "learning_rate": 0.0011584995038072016,
+      "loss": 1.2061,
+      "step": 7451
+    },
+    {
+      "epoch": 0.5731866779478502,
+      "learning_rate": 0.0011581465688404661,
+      "loss": 1.3396,
+      "step": 7452
+    },
+    {
+      "epoch": 0.5732635951080686,
+      "learning_rate": 0.001157793653834909,
+      "loss": 1.2358,
+      "step": 7453
+    },
+    {
+      "epoch": 0.5733405122682871,
+      "learning_rate": 0.0011574407588111366,
+      "loss": 1.0596,
+      "step": 7454
+    },
+    {
+      "epoch": 0.5734174294285055,
+      "learning_rate": 0.0011570878837897558,
+      "loss": 1.2003,
+      "step": 7455
+    },
+    {
+      "epoch": 0.5734943465887239,
+      "learning_rate": 0.0011567350287913709,
+      "loss": 1.3258,
+      "step": 7456
+    },
+    {
+      "epoch": 0.5735712637489424,
+      "learning_rate": 0.0011563821938365844,
+      "loss": 1.1323,
+      "step": 7457
+    },
+    {
+      "epoch": 0.5736481809091608,
+      "learning_rate": 0.001156029378946001,
+      "loss": 0.9305,
+      "step": 7458
+    },
+    {
+      "epoch": 0.5737250980693793,
+      "learning_rate": 0.0011556765841402195,
+      "loss": 0.9148,
+      "step": 7459
+    },
+    {
+      "epoch": 0.5738020152295977,
+      "learning_rate": 0.001155323809439842,
+      "loss": 0.9186,
+      "step": 7460
+    },
+    {
+      "epoch": 0.5738789323898161,
+      "learning_rate": 0.0011549710548654664,
+      "loss": 1.4991,
+      "step": 7461
+    },
+    {
+      "epoch": 0.5739558495500346,
+      "learning_rate": 0.0011546183204376899,
+      "loss": 1.5976,
+      "step": 7462
+    },
+    {
+      "epoch": 0.574032766710253,
+      "learning_rate": 0.00115426560617711,
+      "loss": 1.1626,
+      "step": 7463
+    },
+    {
+      "epoch": 0.5741096838704715,
+      "learning_rate": 0.0011539129121043218,
+      "loss": 0.9012,
+      "step": 7464
+    },
+    {
+      "epoch": 0.57418660103069,
+      "learning_rate": 0.001153560238239919,
+      "loss": 1.1351,
+      "step": 7465
+    },
+    {
+      "epoch": 0.5742635181909084,
+      "learning_rate": 0.0011532075846044953,
+      "loss": 1.2098,
+      "step": 7466
+    },
+    {
+      "epoch": 0.5743404353511269,
+      "learning_rate": 0.0011528549512186414,
+      "loss": 1.013,
+      "step": 7467
+    },
+    {
+      "epoch": 0.5744173525113453,
+      "learning_rate": 0.0011525023381029496,
+      "loss": 1.0897,
+      "step": 7468
+    },
+    {
+      "epoch": 0.5744942696715637,
+      "learning_rate": 0.0011521497452780085,
+      "loss": 0.9073,
+      "step": 7469
+    },
+    {
+      "epoch": 0.5745711868317822,
+      "learning_rate": 0.0011517971727644052,
+      "loss": 1.2426,
+      "step": 7470
+    },
+    {
+      "epoch": 0.5746481039920006,
+      "learning_rate": 0.001151444620582729,
+      "loss": 1.2048,
+      "step": 7471
+    },
+    {
+      "epoch": 0.574725021152219,
+      "learning_rate": 0.0011510920887535644,
+      "loss": 1.2618,
+      "step": 7472
+    },
+    {
+      "epoch": 0.5748019383124375,
+      "learning_rate": 0.0011507395772974958,
+      "loss": 1.235,
+      "step": 7473
+    },
+    {
+      "epoch": 0.5748788554726559,
+      "learning_rate": 0.0011503870862351077,
+      "loss": 0.6297,
+      "step": 7474
+    },
+    {
+      "epoch": 0.5749557726328743,
+      "learning_rate": 0.001150034615586982,
+      "loss": 1.1965,
+      "step": 7475
+    },
+    {
+      "epoch": 0.5750326897930929,
+      "learning_rate": 0.0011496821653736995,
+      "loss": 1.3758,
+      "step": 7476
+    },
+    {
+      "epoch": 0.5751096069533113,
+      "learning_rate": 0.001149329735615841,
+      "loss": 1.236,
+      "step": 7477
+    },
+    {
+      "epoch": 0.5751865241135298,
+      "learning_rate": 0.001148977326333984,
+      "loss": 1.1907,
+      "step": 7478
+    },
+    {
+      "epoch": 0.5752634412737482,
+      "learning_rate": 0.0011486249375487073,
+      "loss": 1.1245,
+      "step": 7479
+    },
+    {
+      "epoch": 0.5753403584339666,
+      "learning_rate": 0.001148272569280587,
+      "loss": 0.888,
+      "step": 7480
+    },
+    {
+      "epoch": 0.5754172755941851,
+      "learning_rate": 0.0011479202215501968,
+      "loss": 1.2184,
+      "step": 7481
+    },
+    {
+      "epoch": 0.5754941927544035,
+      "learning_rate": 0.001147567894378113,
+      "loss": 1.2947,
+      "step": 7482
+    },
+    {
+      "epoch": 0.5755711099146219,
+      "learning_rate": 0.0011472155877849067,
+      "loss": 1.3256,
+      "step": 7483
+    },
+    {
+      "epoch": 0.5756480270748404,
+      "learning_rate": 0.0011468633017911495,
+      "loss": 1.3511,
+      "step": 7484
+    },
+    {
+      "epoch": 0.5757249442350588,
+      "learning_rate": 0.0011465110364174127,
+      "loss": 0.7183,
+      "step": 7485
+    },
+    {
+      "epoch": 0.5758018613952773,
+      "learning_rate": 0.0011461587916842644,
+      "loss": 1.0434,
+      "step": 7486
+    },
+    {
+      "epoch": 0.5758787785554957,
+      "learning_rate": 0.0011458065676122732,
+      "loss": 1.032,
+      "step": 7487
+    },
+    {
+      "epoch": 0.5759556957157141,
+      "learning_rate": 0.0011454543642220058,
+      "loss": 1.4464,
+      "step": 7488
+    },
+    {
+      "epoch": 0.5760326128759327,
+      "learning_rate": 0.0011451021815340272,
+      "loss": 1.2936,
+      "step": 7489
+    },
+    {
+      "epoch": 0.5761095300361511,
+      "learning_rate": 0.0011447500195689024,
+      "loss": 1.0794,
+      "step": 7490
+    },
+    {
+      "epoch": 0.5761864471963695,
+      "learning_rate": 0.0011443978783471946,
+      "loss": 1.2515,
+      "step": 7491
+    },
+    {
+      "epoch": 0.576263364356588,
+      "learning_rate": 0.0011440457578894645,
+      "loss": 1.0259,
+      "step": 7492
+    },
+    {
+      "epoch": 0.5763402815168064,
+      "learning_rate": 0.0011436936582162741,
+      "loss": 0.9607,
+      "step": 7493
+    },
+    {
+      "epoch": 0.5764171986770248,
+      "learning_rate": 0.0011433415793481823,
+      "loss": 1.8073,
+      "step": 7494
+    },
+    {
+      "epoch": 0.5764941158372433,
+      "learning_rate": 0.001142989521305747,
+      "loss": 0.8091,
+      "step": 7495
+    },
+    {
+      "epoch": 0.5765710329974617,
+      "learning_rate": 0.0011426374841095256,
+      "loss": 1.2961,
+      "step": 7496
+    },
+    {
+      "epoch": 0.5766479501576802,
+      "learning_rate": 0.001142285467780074,
+      "loss": 1.3273,
+      "step": 7497
+    },
+    {
+      "epoch": 0.5767248673178986,
+      "learning_rate": 0.001141933472337947,
+      "loss": 1.635,
+      "step": 7498
+    },
+    {
+      "epoch": 0.576801784478117,
+      "learning_rate": 0.0011415814978036978,
+      "loss": 1.0022,
+      "step": 7499
+    },
+    {
+      "epoch": 0.5768787016383355,
+      "learning_rate": 0.001141229544197878,
+      "loss": 1.1319,
+      "step": 7500
+    },
+    {
+      "epoch": 0.576955618798554,
+      "learning_rate": 0.0011408776115410396,
+      "loss": 1.3035,
+      "step": 7501
+    },
+    {
+      "epoch": 0.5770325359587724,
+      "learning_rate": 0.0011405256998537317,
+      "loss": 1.3327,
+      "step": 7502
+    },
+    {
+      "epoch": 0.5771094531189909,
+      "learning_rate": 0.001140173809156502,
+      "loss": 1.1914,
+      "step": 7503
+    },
+    {
+      "epoch": 0.5771863702792093,
+      "learning_rate": 0.0011398219394698995,
+      "loss": 1.0573,
+      "step": 7504
+    },
+    {
+      "epoch": 0.5772632874394278,
+      "learning_rate": 0.0011394700908144684,
+      "loss": 1.3627,
+      "step": 7505
+    },
+    {
+      "epoch": 0.5773402045996462,
+      "learning_rate": 0.0011391182632107554,
+      "loss": 0.9205,
+      "step": 7506
+    },
+    {
+      "epoch": 0.5774171217598646,
+      "learning_rate": 0.001138766456679303,
+      "loss": 1.3411,
+      "step": 7507
+    },
+    {
+      "epoch": 0.5774940389200831,
+      "learning_rate": 0.001138414671240653,
+      "loss": 1.5169,
+      "step": 7508
+    },
+    {
+      "epoch": 0.5775709560803015,
+      "learning_rate": 0.0011380629069153477,
+      "loss": 1.1586,
+      "step": 7509
+    },
+    {
+      "epoch": 0.5776478732405199,
+      "learning_rate": 0.0011377111637239264,
+      "loss": 0.977,
+      "step": 7510
+    },
+    {
+      "epoch": 0.5777247904007384,
+      "learning_rate": 0.0011373594416869274,
+      "loss": 0.8814,
+      "step": 7511
+    },
+    {
+      "epoch": 0.5778017075609568,
+      "learning_rate": 0.001137007740824889,
+      "loss": 1.3461,
+      "step": 7512
+    },
+    {
+      "epoch": 0.5778786247211753,
+      "learning_rate": 0.0011366560611583466,
+      "loss": 0.9412,
+      "step": 7513
+    },
+    {
+      "epoch": 0.5779555418813938,
+      "learning_rate": 0.0011363044027078357,
+      "loss": 1.235,
+      "step": 7514
+    },
+    {
+      "epoch": 0.5780324590416122,
+      "learning_rate": 0.0011359527654938902,
+      "loss": 1.1531,
+      "step": 7515
+    },
+    {
+      "epoch": 0.5781093762018307,
+      "learning_rate": 0.0011356011495370411,
+      "loss": 1.1962,
+      "step": 7516
+    },
+    {
+      "epoch": 0.5781862933620491,
+      "learning_rate": 0.0011352495548578216,
+      "loss": 1.1847,
+      "step": 7517
+    },
+    {
+      "epoch": 0.5782632105222675,
+      "learning_rate": 0.0011348979814767603,
+      "loss": 0.9579,
+      "step": 7518
+    },
+    {
+      "epoch": 0.578340127682486,
+      "learning_rate": 0.0011345464294143862,
+      "loss": 0.8197,
+      "step": 7519
+    },
+    {
+      "epoch": 0.5784170448427044,
+      "learning_rate": 0.0011341948986912272,
+      "loss": 1.2128,
+      "step": 7520
+    },
+    {
+      "epoch": 0.5784939620029228,
+      "learning_rate": 0.0011338433893278094,
+      "loss": 1.2701,
+      "step": 7521
+    },
+    {
+      "epoch": 0.5785708791631413,
+      "learning_rate": 0.0011334919013446573,
+      "loss": 1.0184,
+      "step": 7522
+    },
+    {
+      "epoch": 0.5786477963233597,
+      "learning_rate": 0.0011331404347622953,
+      "loss": 1.5434,
+      "step": 7523
+    },
+    {
+      "epoch": 0.5787247134835782,
+      "learning_rate": 0.0011327889896012454,
+      "loss": 0.9095,
+      "step": 7524
+    },
+    {
+      "epoch": 0.5788016306437966,
+      "learning_rate": 0.0011324375658820292,
+      "loss": 1.3754,
+      "step": 7525
+    },
+    {
+      "epoch": 0.578878547804015,
+      "learning_rate": 0.0011320861636251673,
+      "loss": 1.1569,
+      "step": 7526
+    },
+    {
+      "epoch": 0.5789554649642336,
+      "learning_rate": 0.0011317347828511764,
+      "loss": 1.0138,
+      "step": 7527
+    },
+    {
+      "epoch": 0.579032382124452,
+      "learning_rate": 0.0011313834235805766,
+      "loss": 1.212,
+      "step": 7528
+    },
+    {
+      "epoch": 0.5791092992846704,
+      "learning_rate": 0.0011310320858338822,
+      "loss": 1.4247,
+      "step": 7529
+    },
+    {
+      "epoch": 0.5791862164448889,
+      "learning_rate": 0.0011306807696316089,
+      "loss": 1.0551,
+      "step": 7530
+    },
+    {
+      "epoch": 0.5792631336051073,
+      "learning_rate": 0.0011303294749942705,
+      "loss": 1.189,
+      "step": 7531
+    },
+    {
+      "epoch": 0.5793400507653258,
+      "learning_rate": 0.0011299782019423789,
+      "loss": 1.4765,
+      "step": 7532
+    },
+    {
+      "epoch": 0.5794169679255442,
+      "learning_rate": 0.0011296269504964462,
+      "loss": 1.2702,
+      "step": 7533
+    },
+    {
+      "epoch": 0.5794938850857626,
+      "learning_rate": 0.0011292757206769817,
+      "loss": 1.2545,
+      "step": 7534
+    },
+    {
+      "epoch": 0.5795708022459811,
+      "learning_rate": 0.001128924512504494,
+      "loss": 1.1158,
+      "step": 7535
+    },
+    {
+      "epoch": 0.5796477194061995,
+      "learning_rate": 0.0011285733259994912,
+      "loss": 0.7285,
+      "step": 7536
+    },
+    {
+      "epoch": 0.5797246365664179,
+      "learning_rate": 0.0011282221611824792,
+      "loss": 1.5378,
+      "step": 7537
+    },
+    {
+      "epoch": 0.5798015537266364,
+      "learning_rate": 0.0011278710180739617,
+      "loss": 1.0237,
+      "step": 7538
+    },
+    {
+      "epoch": 0.5798784708868548,
+      "learning_rate": 0.0011275198966944442,
+      "loss": 1.0707,
+      "step": 7539
+    },
+    {
+      "epoch": 0.5799553880470733,
+      "learning_rate": 0.0011271687970644277,
+      "loss": 0.9255,
+      "step": 7540
+    },
+    {
+      "epoch": 0.5800323052072918,
+      "learning_rate": 0.0011268177192044135,
+      "loss": 1.1747,
+      "step": 7541
+    },
+    {
+      "epoch": 0.5801092223675102,
+      "learning_rate": 0.0011264666631349018,
+      "loss": 1.2695,
+      "step": 7542
+    },
+    {
+      "epoch": 0.5801861395277287,
+      "learning_rate": 0.0011261156288763909,
+      "loss": 1.0433,
+      "step": 7543
+    },
+    {
+      "epoch": 0.5802630566879471,
+      "learning_rate": 0.001125764616449378,
+      "loss": 0.8894,
+      "step": 7544
+    },
+    {
+      "epoch": 0.5803399738481655,
+      "learning_rate": 0.0011254136258743595,
+      "loss": 1.4501,
+      "step": 7545
+    },
+    {
+      "epoch": 0.580416891008384,
+      "learning_rate": 0.0011250626571718293,
+      "loss": 1.0785,
+      "step": 7546
+    },
+    {
+      "epoch": 0.5804938081686024,
+      "learning_rate": 0.0011247117103622816,
+      "loss": 1.2745,
+      "step": 7547
+    },
+    {
+      "epoch": 0.5805707253288208,
+      "learning_rate": 0.0011243607854662086,
+      "loss": 1.278,
+      "step": 7548
+    },
+    {
+      "epoch": 0.5806476424890393,
+      "learning_rate": 0.0011240098825040998,
+      "loss": 1.1555,
+      "step": 7549
+    },
+    {
+      "epoch": 0.5807245596492577,
+      "learning_rate": 0.001123659001496447,
+      "loss": 1.2342,
+      "step": 7550
+    },
+    {
+      "epoch": 0.5808014768094762,
+      "learning_rate": 0.001123308142463736,
+      "loss": 0.9779,
+      "step": 7551
+    },
+    {
+      "epoch": 0.5808783939696947,
+      "learning_rate": 0.0011229573054264564,
+      "loss": 1.3377,
+      "step": 7552
+    },
+    {
+      "epoch": 0.580955311129913,
+      "learning_rate": 0.0011226064904050924,
+      "loss": 0.8546,
+      "step": 7553
+    },
+    {
+      "epoch": 0.5810322282901316,
+      "learning_rate": 0.0011222556974201281,
+      "loss": 1.1539,
+      "step": 7554
+    },
+    {
+      "epoch": 0.58110914545035,
+      "learning_rate": 0.0011219049264920478,
+      "loss": 0.819,
+      "step": 7555
+    },
+    {
+      "epoch": 0.5811860626105684,
+      "learning_rate": 0.001121554177641333,
+      "loss": 1.2979,
+      "step": 7556
+    },
+    {
+      "epoch": 0.5812629797707869,
+      "learning_rate": 0.0011212034508884638,
+      "loss": 0.7625,
+      "step": 7557
+    },
+    {
+      "epoch": 0.5813398969310053,
+      "learning_rate": 0.00112085274625392,
+      "loss": 0.8487,
+      "step": 7558
+    },
+    {
+      "epoch": 0.5814168140912237,
+      "learning_rate": 0.0011205020637581795,
+      "loss": 1.195,
+      "step": 7559
+    },
+    {
+      "epoch": 0.5814937312514422,
+      "learning_rate": 0.001120151403421719,
+      "loss": 1.2928,
+      "step": 7560
+    },
+    {
+      "epoch": 0.5815706484116606,
+      "learning_rate": 0.0011198007652650143,
+      "loss": 1.268,
+      "step": 7561
+    },
+    {
+      "epoch": 0.5816475655718791,
+      "learning_rate": 0.001119450149308538,
+      "loss": 1.115,
+      "step": 7562
+    },
+    {
+      "epoch": 0.5817244827320975,
+      "learning_rate": 0.0011190995555727652,
+      "loss": 1.4845,
+      "step": 7563
+    },
+    {
+      "epoch": 0.5818013998923159,
+      "learning_rate": 0.0011187489840781663,
+      "loss": 0.9678,
+      "step": 7564
+    },
+    {
+      "epoch": 0.5818783170525345,
+      "learning_rate": 0.0011183984348452109,
+      "loss": 1.1935,
+      "step": 7565
+    },
+    {
+      "epoch": 0.5819552342127529,
+      "learning_rate": 0.0011180479078943691,
+      "loss": 1.3641,
+      "step": 7566
+    },
+    {
+      "epoch": 0.5820321513729713,
+      "learning_rate": 0.001117697403246108,
+      "loss": 1.0703,
+      "step": 7567
+    },
+    {
+      "epoch": 0.5821090685331898,
+      "learning_rate": 0.0011173469209208935,
+      "loss": 1.2778,
+      "step": 7568
+    },
+    {
+      "epoch": 0.5821859856934082,
+      "learning_rate": 0.0011169964609391914,
+      "loss": 1.0941,
+      "step": 7569
+    },
+    {
+      "epoch": 0.5822629028536267,
+      "learning_rate": 0.0011166460233214648,
+      "loss": 1.2988,
+      "step": 7570
+    },
+    {
+      "epoch": 0.5823398200138451,
+      "learning_rate": 0.0011162956080881768,
+      "loss": 1.1732,
+      "step": 7571
+    },
+    {
+      "epoch": 0.5824167371740635,
+      "learning_rate": 0.0011159452152597882,
+      "loss": 1.1327,
+      "step": 7572
+    },
+    {
+      "epoch": 0.582493654334282,
+      "learning_rate": 0.0011155948448567585,
+      "loss": 1.192,
+      "step": 7573
+    },
+    {
+      "epoch": 0.5825705714945004,
+      "learning_rate": 0.0011152444968995465,
+      "loss": 1.1793,
+      "step": 7574
+    },
+    {
+      "epoch": 0.5826474886547188,
+      "learning_rate": 0.0011148941714086097,
+      "loss": 0.7788,
+      "step": 7575
+    },
+    {
+      "epoch": 0.5827244058149373,
+      "learning_rate": 0.0011145438684044028,
+      "loss": 1.1694,
+      "step": 7576
+    },
+    {
+      "epoch": 0.5828013229751557,
+      "learning_rate": 0.0011141935879073817,
+      "loss": 0.776,
+      "step": 7577
+    },
+    {
+      "epoch": 0.5828782401353741,
+      "learning_rate": 0.0011138433299379988,
+      "loss": 1.3314,
+      "step": 7578
+    },
+    {
+      "epoch": 0.5829551572955927,
+      "learning_rate": 0.0011134930945167064,
+      "loss": 1.2636,
+      "step": 7579
+    },
+    {
+      "epoch": 0.5830320744558111,
+      "learning_rate": 0.001113142881663955,
+      "loss": 1.3117,
+      "step": 7580
+    },
+    {
+      "epoch": 0.5831089916160296,
+      "learning_rate": 0.0011127926914001936,
+      "loss": 1.4092,
+      "step": 7581
+    },
+    {
+      "epoch": 0.583185908776248,
+      "learning_rate": 0.001112442523745871,
+      "loss": 1.1105,
+      "step": 7582
+    },
+    {
+      "epoch": 0.5832628259364664,
+      "learning_rate": 0.0011120923787214332,
+      "loss": 1.3215,
+      "step": 7583
+    },
+    {
+      "epoch": 0.5833397430966849,
+      "learning_rate": 0.0011117422563473253,
+      "loss": 1.4006,
+      "step": 7584
+    },
+    {
+      "epoch": 0.5834166602569033,
+      "learning_rate": 0.0011113921566439922,
+      "loss": 1.3776,
+      "step": 7585
+    },
+    {
+      "epoch": 0.5834935774171217,
+      "learning_rate": 0.0011110420796318764,
+      "loss": 1.3827,
+      "step": 7586
+    },
+    {
+      "epoch": 0.5835704945773402,
+      "learning_rate": 0.0011106920253314179,
+      "loss": 1.1859,
+      "step": 7587
+    },
+    {
+      "epoch": 0.5836474117375586,
+      "learning_rate": 0.0011103419937630585,
+      "loss": 1.1144,
+      "step": 7588
+    },
+    {
+      "epoch": 0.5837243288977771,
+      "learning_rate": 0.0011099919849472358,
+      "loss": 1.0833,
+      "step": 7589
+    },
+    {
+      "epoch": 0.5838012460579955,
+      "learning_rate": 0.001109641998904388,
+      "loss": 1.1369,
+      "step": 7590
+    },
+    {
+      "epoch": 0.583878163218214,
+      "learning_rate": 0.0011092920356549504,
+      "loss": 0.9468,
+      "step": 7591
+    },
+    {
+      "epoch": 0.5839550803784325,
+      "learning_rate": 0.0011089420952193577,
+      "loss": 0.9523,
+      "step": 7592
+    },
+    {
+      "epoch": 0.5840319975386509,
+      "learning_rate": 0.0011085921776180442,
+      "loss": 0.9987,
+      "step": 7593
+    },
+    {
+      "epoch": 0.5841089146988693,
+      "learning_rate": 0.0011082422828714413,
+      "loss": 1.3263,
+      "step": 7594
+    },
+    {
+      "epoch": 0.5841858318590878,
+      "learning_rate": 0.0011078924109999791,
+      "loss": 0.7874,
+      "step": 7595
+    },
+    {
+      "epoch": 0.5842627490193062,
+      "learning_rate": 0.0011075425620240885,
+      "loss": 0.7541,
+      "step": 7596
+    },
+    {
+      "epoch": 0.5843396661795246,
+      "learning_rate": 0.0011071927359641968,
+      "loss": 1.4845,
+      "step": 7597
+    },
+    {
+      "epoch": 0.5844165833397431,
+      "learning_rate": 0.0011068429328407296,
+      "loss": 1.0559,
+      "step": 7598
+    },
+    {
+      "epoch": 0.5844935004999615,
+      "learning_rate": 0.0011064931526741143,
+      "loss": 1.1676,
+      "step": 7599
+    },
+    {
+      "epoch": 0.58457041766018,
+      "learning_rate": 0.0011061433954847732,
+      "loss": 1.2336,
+      "step": 7600
+    },
+    {
+      "epoch": 0.5846473348203984,
+      "learning_rate": 0.0011057936612931302,
+      "loss": 1.1726,
+      "step": 7601
+    },
+    {
+      "epoch": 0.5847242519806168,
+      "learning_rate": 0.001105443950119606,
+      "loss": 1.1307,
+      "step": 7602
+    },
+    {
+      "epoch": 0.5848011691408354,
+      "learning_rate": 0.0011050942619846205,
+      "loss": 1.2726,
+      "step": 7603
+    },
+    {
+      "epoch": 0.5848780863010538,
+      "learning_rate": 0.001104744596908593,
+      "loss": 0.8969,
+      "step": 7604
+    },
+    {
+      "epoch": 0.5849550034612722,
+      "learning_rate": 0.0011043949549119404,
+      "loss": 1.1497,
+      "step": 7605
+    },
+    {
+      "epoch": 0.5850319206214907,
+      "learning_rate": 0.001104045336015078,
+      "loss": 1.1684,
+      "step": 7606
+    },
+    {
+      "epoch": 0.5851088377817091,
+      "learning_rate": 0.0011036957402384218,
+      "loss": 1.2694,
+      "step": 7607
+    },
+    {
+      "epoch": 0.5851857549419276,
+      "learning_rate": 0.0011033461676023839,
+      "loss": 1.4821,
+      "step": 7608
+    },
+    {
+      "epoch": 0.585262672102146,
+      "learning_rate": 0.001102996618127377,
+      "loss": 0.9267,
+      "step": 7609
+    },
+    {
+      "epoch": 0.5853395892623644,
+      "learning_rate": 0.0011026470918338116,
+      "loss": 1.0763,
+      "step": 7610
+    },
+    {
+      "epoch": 0.5854165064225829,
+      "learning_rate": 0.0011022975887420958,
+      "loss": 1.2333,
+      "step": 7611
+    },
+    {
+      "epoch": 0.5854934235828013,
+      "learning_rate": 0.001101948108872639,
+      "loss": 1.174,
+      "step": 7612
+    },
+    {
+      "epoch": 0.5855703407430197,
+      "learning_rate": 0.0011015986522458466,
+      "loss": 1.4906,
+      "step": 7613
+    },
+    {
+      "epoch": 0.5856472579032382,
+      "learning_rate": 0.001101249218882124,
+      "loss": 0.809,
+      "step": 7614
+    },
+    {
+      "epoch": 0.5857241750634566,
+      "learning_rate": 0.0011008998088018753,
+      "loss": 1.0597,
+      "step": 7615
+    },
+    {
+      "epoch": 0.5858010922236752,
+      "learning_rate": 0.0011005504220255025,
+      "loss": 1.1544,
+      "step": 7616
+    },
+    {
+      "epoch": 0.5858780093838936,
+      "learning_rate": 0.001100201058573407,
+      "loss": 1.1005,
+      "step": 7617
+    },
+    {
+      "epoch": 0.585954926544112,
+      "learning_rate": 0.0010998517184659885,
+      "loss": 1.0385,
+      "step": 7618
+    },
+    {
+      "epoch": 0.5860318437043305,
+      "learning_rate": 0.0010995024017236448,
+      "loss": 1.2724,
+      "step": 7619
+    },
+    {
+      "epoch": 0.5861087608645489,
+      "learning_rate": 0.0010991531083667735,
+      "loss": 1.1509,
+      "step": 7620
+    },
+    {
+      "epoch": 0.5861856780247673,
+      "learning_rate": 0.0010988038384157704,
+      "loss": 1.073,
+      "step": 7621
+    },
+    {
+      "epoch": 0.5862625951849858,
+      "learning_rate": 0.0010984545918910282,
+      "loss": 1.2951,
+      "step": 7622
+    },
+    {
+      "epoch": 0.5863395123452042,
+      "learning_rate": 0.0010981053688129418,
+      "loss": 1.0164,
+      "step": 7623
+    },
+    {
+      "epoch": 0.5864164295054226,
+      "learning_rate": 0.0010977561692019017,
+      "loss": 1.2168,
+      "step": 7624
+    },
+    {
+      "epoch": 0.5864933466656411,
+      "learning_rate": 0.0010974069930782975,
+      "loss": 1.3536,
+      "step": 7625
+    },
+    {
+      "epoch": 0.5865702638258595,
+      "learning_rate": 0.001097057840462519,
+      "loss": 1.263,
+      "step": 7626
+    },
+    {
+      "epoch": 0.586647180986078,
+      "learning_rate": 0.001096708711374953,
+      "loss": 1.0746,
+      "step": 7627
+    },
+    {
+      "epoch": 0.5867240981462964,
+      "learning_rate": 0.0010963596058359854,
+      "loss": 0.9617,
+      "step": 7628
+    },
+    {
+      "epoch": 0.5868010153065149,
+      "learning_rate": 0.0010960105238660015,
+      "loss": 1.0653,
+      "step": 7629
+    },
+    {
+      "epoch": 0.5868779324667334,
+      "learning_rate": 0.0010956614654853837,
+      "loss": 1.0892,
+      "step": 7630
+    },
+    {
+      "epoch": 0.5869548496269518,
+      "learning_rate": 0.0010953124307145146,
+      "loss": 1.2078,
+      "step": 7631
+    },
+    {
+      "epoch": 0.5870317667871702,
+      "learning_rate": 0.0010949634195737748,
+      "loss": 1.3924,
+      "step": 7632
+    },
+    {
+      "epoch": 0.5871086839473887,
+      "learning_rate": 0.0010946144320835422,
+      "loss": 1.1187,
+      "step": 7633
+    },
+    {
+      "epoch": 0.5871856011076071,
+      "learning_rate": 0.0010942654682641962,
+      "loss": 1.0561,
+      "step": 7634
+    },
+    {
+      "epoch": 0.5872625182678256,
+      "learning_rate": 0.0010939165281361116,
+      "loss": 1.2642,
+      "step": 7635
+    },
+    {
+      "epoch": 0.587339435428044,
+      "learning_rate": 0.0010935676117196649,
+      "loss": 1.2873,
+      "step": 7636
+    },
+    {
+      "epoch": 0.5874163525882624,
+      "learning_rate": 0.0010932187190352287,
+      "loss": 1.3543,
+      "step": 7637
+    },
+    {
+      "epoch": 0.5874932697484809,
+      "learning_rate": 0.0010928698501031751,
+      "loss": 1.1779,
+      "step": 7638
+    },
+    {
+      "epoch": 0.5875701869086993,
+      "learning_rate": 0.0010925210049438755,
+      "loss": 0.9547,
+      "step": 7639
+    },
+    {
+      "epoch": 0.5876471040689177,
+      "learning_rate": 0.001092172183577699,
+      "loss": 1.0573,
+      "step": 7640
+    },
+    {
+      "epoch": 0.5877240212291363,
+      "learning_rate": 0.0010918233860250136,
+      "loss": 1.068,
+      "step": 7641
+    },
+    {
+      "epoch": 0.5878009383893547,
+      "learning_rate": 0.0010914746123061865,
+      "loss": 0.7642,
+      "step": 7642
+    },
+    {
+      "epoch": 0.5878778555495731,
+      "learning_rate": 0.0010911258624415826,
+      "loss": 1.1254,
+      "step": 7643
+    },
+    {
+      "epoch": 0.5879547727097916,
+      "learning_rate": 0.001090777136451565,
+      "loss": 1.0528,
+      "step": 7644
+    },
+    {
+      "epoch": 0.58803168987001,
+      "learning_rate": 0.0010904284343564974,
+      "loss": 1.0813,
+      "step": 7645
+    },
+    {
+      "epoch": 0.5881086070302285,
+      "learning_rate": 0.0010900797561767398,
+      "loss": 0.8347,
+      "step": 7646
+    },
+    {
+      "epoch": 0.5881855241904469,
+      "learning_rate": 0.0010897311019326533,
+      "loss": 1.137,
+      "step": 7647
+    },
+    {
+      "epoch": 0.5882624413506653,
+      "learning_rate": 0.0010893824716445952,
+      "loss": 0.9631,
+      "step": 7648
+    },
+    {
+      "epoch": 0.5883393585108838,
+      "learning_rate": 0.001089033865332922,
+      "loss": 1.2,
+      "step": 7649
+    },
+    {
+      "epoch": 0.5884162756711022,
+      "learning_rate": 0.00108868528301799,
+      "loss": 1.1167,
+      "step": 7650
+    },
+    {
+      "epoch": 0.5884931928313206,
+      "learning_rate": 0.0010883367247201529,
+      "loss": 1.158,
+      "step": 7651
+    },
+    {
+      "epoch": 0.5885701099915391,
+      "learning_rate": 0.0010879881904597634,
+      "loss": 1.1843,
+      "step": 7652
+    },
+    {
+      "epoch": 0.5886470271517575,
+      "learning_rate": 0.001087639680257173,
+      "loss": 0.872,
+      "step": 7653
+    },
+    {
+      "epoch": 0.588723944311976,
+      "learning_rate": 0.0010872911941327312,
+      "loss": 1.4038,
+      "step": 7654
+    },
+    {
+      "epoch": 0.5888008614721945,
+      "learning_rate": 0.001086942732106787,
+      "loss": 1.2358,
+      "step": 7655
+    },
+    {
+      "epoch": 0.5888777786324129,
+      "learning_rate": 0.0010865942941996874,
+      "loss": 1.1987,
+      "step": 7656
+    },
+    {
+      "epoch": 0.5889546957926314,
+      "learning_rate": 0.0010862458804317768,
+      "loss": 1.3849,
+      "step": 7657
+    },
+    {
+      "epoch": 0.5890316129528498,
+      "learning_rate": 0.0010858974908234014,
+      "loss": 0.9745,
+      "step": 7658
+    },
+    {
+      "epoch": 0.5891085301130682,
+      "learning_rate": 0.0010855491253949027,
+      "loss": 1.3895,
+      "step": 7659
+    },
+    {
+      "epoch": 0.5891854472732867,
+      "learning_rate": 0.0010852007841666224,
+      "loss": 1.146,
+      "step": 7660
+    },
+    {
+      "epoch": 0.5892623644335051,
+      "learning_rate": 0.0010848524671589008,
+      "loss": 1.1424,
+      "step": 7661
+    },
+    {
+      "epoch": 0.5893392815937235,
+      "learning_rate": 0.0010845041743920763,
+      "loss": 1.1147,
+      "step": 7662
+    },
+    {
+      "epoch": 0.589416198753942,
+      "learning_rate": 0.0010841559058864857,
+      "loss": 0.9315,
+      "step": 7663
+    },
+    {
+      "epoch": 0.5894931159141604,
+      "learning_rate": 0.0010838076616624657,
+      "loss": 1.2049,
+      "step": 7664
+    },
+    {
+      "epoch": 0.5895700330743789,
+      "learning_rate": 0.0010834594417403495,
+      "loss": 0.9668,
+      "step": 7665
+    },
+    {
+      "epoch": 0.5896469502345973,
+      "learning_rate": 0.0010831112461404712,
+      "loss": 0.8758,
+      "step": 7666
+    },
+    {
+      "epoch": 0.5897238673948157,
+      "learning_rate": 0.0010827630748831618,
+      "loss": 0.9744,
+      "step": 7667
+    },
+    {
+      "epoch": 0.5898007845550343,
+      "learning_rate": 0.0010824149279887504,
+      "loss": 1.1687,
+      "step": 7668
+    },
+    {
+      "epoch": 0.5898777017152527,
+      "learning_rate": 0.0010820668054775675,
+      "loss": 1.3187,
+      "step": 7669
+    },
+    {
+      "epoch": 0.5899546188754711,
+      "learning_rate": 0.0010817187073699388,
+      "loss": 0.8032,
+      "step": 7670
+    },
+    {
+      "epoch": 0.5900315360356896,
+      "learning_rate": 0.0010813706336861909,
+      "loss": 1.0688,
+      "step": 7671
+    },
+    {
+      "epoch": 0.590108453195908,
+      "learning_rate": 0.001081022584446648,
+      "loss": 0.9207,
+      "step": 7672
+    },
+    {
+      "epoch": 0.5901853703561265,
+      "learning_rate": 0.0010806745596716328,
+      "loss": 1.4195,
+      "step": 7673
+    },
+    {
+      "epoch": 0.5902622875163449,
+      "learning_rate": 0.0010803265593814673,
+      "loss": 1.1843,
+      "step": 7674
+    },
+    {
+      "epoch": 0.5903392046765633,
+      "learning_rate": 0.0010799785835964714,
+      "loss": 1.6493,
+      "step": 7675
+    },
+    {
+      "epoch": 0.5904161218367818,
+      "learning_rate": 0.0010796306323369634,
+      "loss": 1.2095,
+      "step": 7676
+    },
+    {
+      "epoch": 0.5904930389970002,
+      "learning_rate": 0.0010792827056232614,
+      "loss": 1.1495,
+      "step": 7677
+    },
+    {
+      "epoch": 0.5905699561572186,
+      "learning_rate": 0.0010789348034756809,
+      "loss": 1.1788,
+      "step": 7678
+    },
+    {
+      "epoch": 0.5906468733174371,
+      "learning_rate": 0.0010785869259145348,
+      "loss": 1.1645,
+      "step": 7679
+    },
+    {
+      "epoch": 0.5907237904776556,
+      "learning_rate": 0.0010782390729601385,
+      "loss": 1.136,
+      "step": 7680
+    },
+    {
+      "epoch": 0.590800707637874,
+      "learning_rate": 0.0010778912446328022,
+      "loss": 0.9883,
+      "step": 7681
+    },
+    {
+      "epoch": 0.5908776247980925,
+      "learning_rate": 0.0010775434409528353,
+      "loss": 1.2706,
+      "step": 7682
+    },
+    {
+      "epoch": 0.5909545419583109,
+      "learning_rate": 0.0010771956619405477,
+      "loss": 1.3735,
+      "step": 7683
+    },
+    {
+      "epoch": 0.5910314591185294,
+      "learning_rate": 0.0010768479076162457,
+      "loss": 1.0717,
+      "step": 7684
+    },
+    {
+      "epoch": 0.5911083762787478,
+      "learning_rate": 0.0010765001780002356,
+      "loss": 1.087,
+      "step": 7685
+    },
+    {
+      "epoch": 0.5911852934389662,
+      "learning_rate": 0.0010761524731128214,
+      "loss": 1.3048,
+      "step": 7686
+    },
+    {
+      "epoch": 0.5912622105991847,
+      "learning_rate": 0.0010758047929743059,
+      "loss": 1.2326,
+      "step": 7687
+    },
+    {
+      "epoch": 0.5913391277594031,
+      "learning_rate": 0.001075457137604991,
+      "loss": 0.9792,
+      "step": 7688
+    },
+    {
+      "epoch": 0.5914160449196215,
+      "learning_rate": 0.0010751095070251763,
+      "loss": 0.9589,
+      "step": 7689
+    },
+    {
+      "epoch": 0.59149296207984,
+      "learning_rate": 0.0010747619012551595,
+      "loss": 1.0688,
+      "step": 7690
+    },
+    {
+      "epoch": 0.5915698792400584,
+      "learning_rate": 0.0010744143203152395,
+      "loss": 1.0496,
+      "step": 7691
+    },
+    {
+      "epoch": 0.591646796400277,
+      "learning_rate": 0.0010740667642257097,
+      "loss": 1.4762,
+      "step": 7692
+    },
+    {
+      "epoch": 0.5917237135604954,
+      "learning_rate": 0.0010737192330068665,
+      "loss": 1.1987,
+      "step": 7693
+    },
+    {
+      "epoch": 0.5918006307207138,
+      "learning_rate": 0.0010733717266790013,
+      "loss": 0.9884,
+      "step": 7694
+    },
+    {
+      "epoch": 0.5918775478809323,
+      "learning_rate": 0.0010730242452624051,
+      "loss": 1.2176,
+      "step": 7695
+    },
+    {
+      "epoch": 0.5919544650411507,
+      "learning_rate": 0.0010726767887773688,
+      "loss": 1.1729,
+      "step": 7696
+    },
+    {
+      "epoch": 0.5920313822013691,
+      "learning_rate": 0.0010723293572441799,
+      "loss": 1.1773,
+      "step": 7697
+    },
+    {
+      "epoch": 0.5921082993615876,
+      "learning_rate": 0.0010719819506831252,
+      "loss": 1.1422,
+      "step": 7698
+    },
+    {
+      "epoch": 0.592185216521806,
+      "learning_rate": 0.0010716345691144911,
+      "loss": 1.1546,
+      "step": 7699
+    },
+    {
+      "epoch": 0.5922621336820244,
+      "learning_rate": 0.0010712872125585608,
+      "loss": 0.9597,
+      "step": 7700
+    },
+    {
+      "epoch": 0.5923390508422429,
+      "learning_rate": 0.0010709398810356163,
+      "loss": 1.353,
+      "step": 7701
+    },
+    {
+      "epoch": 0.5924159680024613,
+      "learning_rate": 0.0010705925745659404,
+      "loss": 1.2339,
+      "step": 7702
+    },
+    {
+      "epoch": 0.5924928851626798,
+      "learning_rate": 0.0010702452931698105,
+      "loss": 1.377,
+      "step": 7703
+    },
+    {
+      "epoch": 0.5925698023228982,
+      "learning_rate": 0.0010698980368675066,
+      "loss": 1.0627,
+      "step": 7704
+    },
+    {
+      "epoch": 0.5926467194831166,
+      "learning_rate": 0.0010695508056793044,
+      "loss": 1.0616,
+      "step": 7705
+    },
+    {
+      "epoch": 0.5927236366433352,
+      "learning_rate": 0.001069203599625479,
+      "loss": 1.4626,
+      "step": 7706
+    },
+    {
+      "epoch": 0.5928005538035536,
+      "learning_rate": 0.0010688564187263043,
+      "loss": 1.0914,
+      "step": 7707
+    },
+    {
+      "epoch": 0.592877470963772,
+      "learning_rate": 0.001068509263002053,
+      "loss": 1.2574,
+      "step": 7708
+    },
+    {
+      "epoch": 0.5929543881239905,
+      "learning_rate": 0.0010681621324729951,
+      "loss": 1.2338,
+      "step": 7709
+    },
+    {
+      "epoch": 0.5930313052842089,
+      "learning_rate": 0.0010678150271594008,
+      "loss": 1.4186,
+      "step": 7710
+    },
+    {
+      "epoch": 0.5931082224444274,
+      "learning_rate": 0.001067467947081537,
+      "loss": 1.1059,
+      "step": 7711
+    },
+    {
+      "epoch": 0.5931851396046458,
+      "learning_rate": 0.001067120892259671,
+      "loss": 1.0338,
+      "step": 7712
+    },
+    {
+      "epoch": 0.5932620567648642,
+      "learning_rate": 0.001066773862714067,
+      "loss": 0.9603,
+      "step": 7713
+    },
+    {
+      "epoch": 0.5933389739250827,
+      "learning_rate": 0.0010664268584649886,
+      "loss": 1.025,
+      "step": 7714
+    },
+    {
+      "epoch": 0.5934158910853011,
+      "learning_rate": 0.0010660798795326984,
+      "loss": 1.0605,
+      "step": 7715
+    },
+    {
+      "epoch": 0.5934928082455195,
+      "learning_rate": 0.0010657329259374558,
+      "loss": 0.9222,
+      "step": 7716
+    },
+    {
+      "epoch": 0.593569725405738,
+      "learning_rate": 0.00106538599769952,
+      "loss": 1.4801,
+      "step": 7717
+    },
+    {
+      "epoch": 0.5936466425659565,
+      "learning_rate": 0.001065039094839149,
+      "loss": 1.4249,
+      "step": 7718
+    },
+    {
+      "epoch": 0.5937235597261749,
+      "learning_rate": 0.0010646922173765986,
+      "loss": 1.2505,
+      "step": 7719
+    },
+    {
+      "epoch": 0.5938004768863934,
+      "learning_rate": 0.0010643453653321228,
+      "loss": 1.09,
+      "step": 7720
+    },
+    {
+      "epoch": 0.5938773940466118,
+      "learning_rate": 0.0010639985387259757,
+      "loss": 0.8344,
+      "step": 7721
+    },
+    {
+      "epoch": 0.5939543112068303,
+      "learning_rate": 0.001063651737578408,
+      "loss": 0.9064,
+      "step": 7722
+    },
+    {
+      "epoch": 0.5940312283670487,
+      "learning_rate": 0.0010633049619096703,
+      "loss": 1.3103,
+      "step": 7723
+    },
+    {
+      "epoch": 0.5941081455272671,
+      "learning_rate": 0.001062958211740011,
+      "loss": 1.344,
+      "step": 7724
+    },
+    {
+      "epoch": 0.5941850626874856,
+      "learning_rate": 0.001062611487089677,
+      "loss": 1.3596,
+      "step": 7725
+    },
+    {
+      "epoch": 0.594261979847704,
+      "learning_rate": 0.0010622647879789144,
+      "loss": 1.2105,
+      "step": 7726
+    },
+    {
+      "epoch": 0.5943388970079224,
+      "learning_rate": 0.0010619181144279673,
+      "loss": 1.4586,
+      "step": 7727
+    },
+    {
+      "epoch": 0.5944158141681409,
+      "learning_rate": 0.0010615714664570773,
+      "loss": 1.382,
+      "step": 7728
+    },
+    {
+      "epoch": 0.5944927313283593,
+      "learning_rate": 0.0010612248440864866,
+      "loss": 1.0237,
+      "step": 7729
+    },
+    {
+      "epoch": 0.5945696484885779,
+      "learning_rate": 0.0010608782473364344,
+      "loss": 1.4369,
+      "step": 7730
+    },
+    {
+      "epoch": 0.5946465656487963,
+      "learning_rate": 0.001060531676227159,
+      "loss": 1.1992,
+      "step": 7731
+    },
+    {
+      "epoch": 0.5947234828090147,
+      "learning_rate": 0.0010601851307788974,
+      "loss": 1.3042,
+      "step": 7732
+    },
+    {
+      "epoch": 0.5948003999692332,
+      "learning_rate": 0.001059838611011884,
+      "loss": 0.9944,
+      "step": 7733
+    },
+    {
+      "epoch": 0.5948773171294516,
+      "learning_rate": 0.001059492116946353,
+      "loss": 1.1956,
+      "step": 7734
+    },
+    {
+      "epoch": 0.59495423428967,
+      "learning_rate": 0.0010591456486025363,
+      "loss": 1.5136,
+      "step": 7735
+    },
+    {
+      "epoch": 0.5950311514498885,
+      "learning_rate": 0.0010587992060006645,
+      "loss": 1.3086,
+      "step": 7736
+    },
+    {
+      "epoch": 0.5951080686101069,
+      "learning_rate": 0.0010584527891609671,
+      "loss": 0.9448,
+      "step": 7737
+    },
+    {
+      "epoch": 0.5951849857703254,
+      "learning_rate": 0.0010581063981036712,
+      "loss": 1.2859,
+      "step": 7738
+    },
+    {
+      "epoch": 0.5952619029305438,
+      "learning_rate": 0.0010577600328490037,
+      "loss": 1.4452,
+      "step": 7739
+    },
+    {
+      "epoch": 0.5953388200907622,
+      "learning_rate": 0.001057413693417189,
+      "loss": 1.2173,
+      "step": 7740
+    },
+    {
+      "epoch": 0.5954157372509807,
+      "learning_rate": 0.0010570673798284492,
+      "loss": 1.3203,
+      "step": 7741
+    },
+    {
+      "epoch": 0.5954926544111991,
+      "learning_rate": 0.0010567210921030071,
+      "loss": 1.053,
+      "step": 7742
+    },
+    {
+      "epoch": 0.5955695715714175,
+      "learning_rate": 0.0010563748302610824,
+      "loss": 1.629,
+      "step": 7743
+    },
+    {
+      "epoch": 0.5956464887316361,
+      "learning_rate": 0.0010560285943228935,
+      "loss": 1.1092,
+      "step": 7744
+    },
+    {
+      "epoch": 0.5957234058918545,
+      "learning_rate": 0.001055682384308658,
+      "loss": 1.1787,
+      "step": 7745
+    },
+    {
+      "epoch": 0.5958003230520729,
+      "learning_rate": 0.0010553362002385912,
+      "loss": 1.0458,
+      "step": 7746
+    },
+    {
+      "epoch": 0.5958772402122914,
+      "learning_rate": 0.0010549900421329065,
+      "loss": 1.1286,
+      "step": 7747
+    },
+    {
+      "epoch": 0.5959541573725098,
+      "learning_rate": 0.0010546439100118173,
+      "loss": 1.1391,
+      "step": 7748
+    },
+    {
+      "epoch": 0.5960310745327283,
+      "learning_rate": 0.0010542978038955342,
+      "loss": 1.2542,
+      "step": 7749
+    },
+    {
+      "epoch": 0.5961079916929467,
+      "learning_rate": 0.0010539517238042672,
+      "loss": 1.3272,
+      "step": 7750
+    },
+    {
+      "epoch": 0.5961849088531651,
+      "learning_rate": 0.0010536056697582241,
+      "loss": 1.249,
+      "step": 7751
+    },
+    {
+      "epoch": 0.5962618260133836,
+      "learning_rate": 0.0010532596417776102,
+      "loss": 1.1186,
+      "step": 7752
+    },
+    {
+      "epoch": 0.596338743173602,
+      "learning_rate": 0.0010529136398826325,
+      "loss": 0.9633,
+      "step": 7753
+    },
+    {
+      "epoch": 0.5964156603338204,
+      "learning_rate": 0.001052567664093493,
+      "loss": 1.0394,
+      "step": 7754
+    },
+    {
+      "epoch": 0.596492577494039,
+      "learning_rate": 0.0010522217144303934,
+      "loss": 0.9104,
+      "step": 7755
+    },
+    {
+      "epoch": 0.5965694946542573,
+      "learning_rate": 0.0010518757909135348,
+      "loss": 1.4556,
+      "step": 7756
+    },
+    {
+      "epoch": 0.5966464118144759,
+      "learning_rate": 0.0010515298935631157,
+      "loss": 1.1109,
+      "step": 7757
+    },
+    {
+      "epoch": 0.5967233289746943,
+      "learning_rate": 0.0010511840223993338,
+      "loss": 0.9677,
+      "step": 7758
+    },
+    {
+      "epoch": 0.5968002461349127,
+      "learning_rate": 0.0010508381774423847,
+      "loss": 1.2739,
+      "step": 7759
+    },
+    {
+      "epoch": 0.5968771632951312,
+      "learning_rate": 0.001050492358712462,
+      "loss": 1.031,
+      "step": 7760
+    },
+    {
+      "epoch": 0.5969540804553496,
+      "learning_rate": 0.0010501465662297591,
+      "loss": 1.2035,
+      "step": 7761
+    },
+    {
+      "epoch": 0.597030997615568,
+      "learning_rate": 0.0010498008000144677,
+      "loss": 0.935,
+      "step": 7762
+    },
+    {
+      "epoch": 0.5971079147757865,
+      "learning_rate": 0.0010494550600867755,
+      "loss": 0.9552,
+      "step": 7763
+    },
+    {
+      "epoch": 0.5971848319360049,
+      "learning_rate": 0.001049109346466873,
+      "loss": 1.1457,
+      "step": 7764
+    },
+    {
+      "epoch": 0.5972617490962233,
+      "learning_rate": 0.0010487636591749454,
+      "loss": 1.3244,
+      "step": 7765
+    },
+    {
+      "epoch": 0.5973386662564418,
+      "learning_rate": 0.0010484179982311775,
+      "loss": 0.9031,
+      "step": 7766
+    },
+    {
+      "epoch": 0.5974155834166602,
+      "learning_rate": 0.0010480723636557539,
+      "loss": 1.2468,
+      "step": 7767
+    },
+    {
+      "epoch": 0.5974925005768787,
+      "learning_rate": 0.0010477267554688556,
+      "loss": 1.4144,
+      "step": 7768
+    },
+    {
+      "epoch": 0.5975694177370972,
+      "learning_rate": 0.0010473811736906634,
+      "loss": 1.0551,
+      "step": 7769
+    },
+    {
+      "epoch": 0.5976463348973156,
+      "learning_rate": 0.0010470356183413566,
+      "loss": 1.4656,
+      "step": 7770
+    },
+    {
+      "epoch": 0.5977232520575341,
+      "learning_rate": 0.0010466900894411117,
+      "loss": 1.3411,
+      "step": 7771
+    },
+    {
+      "epoch": 0.5978001692177525,
+      "learning_rate": 0.001046344587010105,
+      "loss": 1.3289,
+      "step": 7772
+    },
+    {
+      "epoch": 0.5978770863779709,
+      "learning_rate": 0.0010459991110685109,
+      "loss": 1.1409,
+      "step": 7773
+    },
+    {
+      "epoch": 0.5979540035381894,
+      "learning_rate": 0.0010456536616365013,
+      "loss": 1.211,
+      "step": 7774
+    },
+    {
+      "epoch": 0.5980309206984078,
+      "learning_rate": 0.0010453082387342487,
+      "loss": 1.1654,
+      "step": 7775
+    },
+    {
+      "epoch": 0.5981078378586263,
+      "learning_rate": 0.0010449628423819207,
+      "loss": 1.2061,
+      "step": 7776
+    },
+    {
+      "epoch": 0.5981847550188447,
+      "learning_rate": 0.001044617472599688,
+      "loss": 1.2899,
+      "step": 7777
+    },
+    {
+      "epoch": 0.5982616721790631,
+      "learning_rate": 0.001044272129407715,
+      "loss": 1.429,
+      "step": 7778
+    },
+    {
+      "epoch": 0.5983385893392816,
+      "learning_rate": 0.001043926812826167,
+      "loss": 1.2076,
+      "step": 7779
+    },
+    {
+      "epoch": 0.5984155064995,
+      "learning_rate": 0.0010435815228752082,
+      "loss": 1.0541,
+      "step": 7780
+    },
+    {
+      "epoch": 0.5984924236597184,
+      "learning_rate": 0.0010432362595749996,
+      "loss": 1.0445,
+      "step": 7781
+    },
+    {
+      "epoch": 0.598569340819937,
+      "learning_rate": 0.0010428910229457018,
+      "loss": 1.3896,
+      "step": 7782
+    },
+    {
+      "epoch": 0.5986462579801554,
+      "learning_rate": 0.0010425458130074738,
+      "loss": 1.277,
+      "step": 7783
+    },
+    {
+      "epoch": 0.5987231751403738,
+      "learning_rate": 0.0010422006297804729,
+      "loss": 0.8614,
+      "step": 7784
+    },
+    {
+      "epoch": 0.5988000923005923,
+      "learning_rate": 0.0010418554732848535,
+      "loss": 1.2623,
+      "step": 7785
+    },
+    {
+      "epoch": 0.5988770094608107,
+      "learning_rate": 0.0010415103435407712,
+      "loss": 0.9991,
+      "step": 7786
+    },
+    {
+      "epoch": 0.5989539266210292,
+      "learning_rate": 0.001041165240568377,
+      "loss": 1.8123,
+      "step": 7787
+    },
+    {
+      "epoch": 0.5990308437812476,
+      "learning_rate": 0.0010408201643878235,
+      "loss": 0.9383,
+      "step": 7788
+    },
+    {
+      "epoch": 0.599107760941466,
+      "learning_rate": 0.0010404751150192586,
+      "loss": 1.3833,
+      "step": 7789
+    },
+    {
+      "epoch": 0.5991846781016845,
+      "learning_rate": 0.0010401300924828306,
+      "loss": 1.1599,
+      "step": 7790
+    },
+    {
+      "epoch": 0.5992615952619029,
+      "learning_rate": 0.0010397850967986861,
+      "loss": 1.0424,
+      "step": 7791
+    },
+    {
+      "epoch": 0.5993385124221213,
+      "learning_rate": 0.001039440127986969,
+      "loss": 1.0356,
+      "step": 7792
+    },
+    {
+      "epoch": 0.5994154295823398,
+      "learning_rate": 0.001039095186067823,
+      "loss": 0.8999,
+      "step": 7793
+    },
+    {
+      "epoch": 0.5994923467425582,
+      "learning_rate": 0.0010387502710613894,
+      "loss": 0.7223,
+      "step": 7794
+    },
+    {
+      "epoch": 0.5995692639027768,
+      "learning_rate": 0.001038405382987808,
+      "loss": 0.7494,
+      "step": 7795
+    },
+    {
+      "epoch": 0.5996461810629952,
+      "learning_rate": 0.0010380605218672178,
+      "loss": 1.0897,
+      "step": 7796
+    },
+    {
+      "epoch": 0.5997230982232136,
+      "learning_rate": 0.0010377156877197553,
+      "loss": 0.9918,
+      "step": 7797
+    },
+    {
+      "epoch": 0.5998000153834321,
+      "learning_rate": 0.0010373708805655546,
+      "loss": 1.4431,
+      "step": 7798
+    },
+    {
+      "epoch": 0.5998769325436505,
+      "learning_rate": 0.0010370261004247513,
+      "loss": 0.9219,
+      "step": 7799
+    },
+    {
+      "epoch": 0.5999538497038689,
+      "learning_rate": 0.0010366813473174762,
+      "loss": 1.0544,
+      "step": 7800
+    },
+    {
+      "epoch": 0.6000307668640874,
+      "learning_rate": 0.00103633662126386,
+      "loss": 1.3724,
+      "step": 7801
+    },
+    {
+      "epoch": 0.6001076840243058,
+      "learning_rate": 0.0010359919222840317,
+      "loss": 1.1616,
+      "step": 7802
+    },
+    {
+      "epoch": 0.6001846011845242,
+      "learning_rate": 0.0010356472503981187,
+      "loss": 0.9862,
+      "step": 7803
+    },
+    {
+      "epoch": 0.6002615183447427,
+      "learning_rate": 0.0010353026056262463,
+      "loss": 1.2182,
+      "step": 7804
+    },
+    {
+      "epoch": 0.6003384355049611,
+      "learning_rate": 0.0010349579879885395,
+      "loss": 1.2488,
+      "step": 7805
+    },
+    {
+      "epoch": 0.6004153526651796,
+      "learning_rate": 0.0010346133975051204,
+      "loss": 1.319,
+      "step": 7806
+    },
+    {
+      "epoch": 0.600492269825398,
+      "learning_rate": 0.0010342688341961098,
+      "loss": 1.0387,
+      "step": 7807
+    },
+    {
+      "epoch": 0.6005691869856165,
+      "learning_rate": 0.001033924298081628,
+      "loss": 0.9835,
+      "step": 7808
+    },
+    {
+      "epoch": 0.600646104145835,
+      "learning_rate": 0.001033579789181791,
+      "loss": 0.9648,
+      "step": 7809
+    },
+    {
+      "epoch": 0.6007230213060534,
+      "learning_rate": 0.0010332353075167174,
+      "loss": 1.1639,
+      "step": 7810
+    },
+    {
+      "epoch": 0.6007999384662718,
+      "learning_rate": 0.0010328908531065202,
+      "loss": 1.1934,
+      "step": 7811
+    },
+    {
+      "epoch": 0.6008768556264903,
+      "learning_rate": 0.0010325464259713125,
+      "loss": 0.9779,
+      "step": 7812
+    },
+    {
+      "epoch": 0.6009537727867087,
+      "learning_rate": 0.0010322020261312066,
+      "loss": 1.212,
+      "step": 7813
+    },
+    {
+      "epoch": 0.6010306899469272,
+      "learning_rate": 0.0010318576536063116,
+      "loss": 0.8679,
+      "step": 7814
+    },
+    {
+      "epoch": 0.6011076071071456,
+      "learning_rate": 0.0010315133084167363,
+      "loss": 0.994,
+      "step": 7815
+    },
+    {
+      "epoch": 0.601184524267364,
+      "learning_rate": 0.0010311689905825872,
+      "loss": 1.0423,
+      "step": 7816
+    },
+    {
+      "epoch": 0.6012614414275825,
+      "learning_rate": 0.0010308247001239693,
+      "loss": 0.9352,
+      "step": 7817
+    },
+    {
+      "epoch": 0.6013383585878009,
+      "learning_rate": 0.001030480437060986,
+      "loss": 1.1421,
+      "step": 7818
+    },
+    {
+      "epoch": 0.6014152757480193,
+      "learning_rate": 0.00103013620141374,
+      "loss": 0.9548,
+      "step": 7819
+    },
+    {
+      "epoch": 0.6014921929082379,
+      "learning_rate": 0.0010297919932023296,
+      "loss": 1.0947,
+      "step": 7820
+    },
+    {
+      "epoch": 0.6015691100684563,
+      "learning_rate": 0.0010294478124468558,
+      "loss": 1.029,
+      "step": 7821
+    },
+    {
+      "epoch": 0.6016460272286747,
+      "learning_rate": 0.0010291036591674144,
+      "loss": 0.8608,
+      "step": 7822
+    },
+    {
+      "epoch": 0.6017229443888932,
+      "learning_rate": 0.0010287595333841007,
+      "loss": 1.2513,
+      "step": 7823
+    },
+    {
+      "epoch": 0.6017998615491116,
+      "learning_rate": 0.0010284154351170095,
+      "loss": 1.3662,
+      "step": 7824
+    },
+    {
+      "epoch": 0.6018767787093301,
+      "learning_rate": 0.0010280713643862319,
+      "loss": 0.9338,
+      "step": 7825
+    },
+    {
+      "epoch": 0.6019536958695485,
+      "learning_rate": 0.0010277273212118596,
+      "loss": 0.8625,
+      "step": 7826
+    },
+    {
+      "epoch": 0.6020306130297669,
+      "learning_rate": 0.0010273833056139812,
+      "loss": 1.2494,
+      "step": 7827
+    },
+    {
+      "epoch": 0.6021075301899854,
+      "learning_rate": 0.001027039317612684,
+      "loss": 0.8106,
+      "step": 7828
+    },
+    {
+      "epoch": 0.6021844473502038,
+      "learning_rate": 0.001026695357228054,
+      "loss": 1.3656,
+      "step": 7829
+    },
+    {
+      "epoch": 0.6022613645104222,
+      "learning_rate": 0.0010263514244801756,
+      "loss": 0.9023,
+      "step": 7830
+    },
+    {
+      "epoch": 0.6023382816706407,
+      "learning_rate": 0.0010260075193891307,
+      "loss": 0.9513,
+      "step": 7831
+    },
+    {
+      "epoch": 0.6024151988308591,
+      "learning_rate": 0.0010256636419750014,
+      "loss": 1.0575,
+      "step": 7832
+    },
+    {
+      "epoch": 0.6024921159910777,
+      "learning_rate": 0.0010253197922578652,
+      "loss": 0.9623,
+      "step": 7833
+    },
+    {
+      "epoch": 0.6025690331512961,
+      "learning_rate": 0.0010249759702578023,
+      "loss": 1.0004,
+      "step": 7834
+    },
+    {
+      "epoch": 0.6026459503115145,
+      "learning_rate": 0.001024632175994887,
+      "loss": 0.9848,
+      "step": 7835
+    },
+    {
+      "epoch": 0.602722867471733,
+      "learning_rate": 0.0010242884094891942,
+      "loss": 1.2173,
+      "step": 7836
+    },
+    {
+      "epoch": 0.6027997846319514,
+      "learning_rate": 0.0010239446707607973,
+      "loss": 1.2593,
+      "step": 7837
+    },
+    {
+      "epoch": 0.6028767017921698,
+      "learning_rate": 0.0010236009598297668,
+      "loss": 0.8232,
+      "step": 7838
+    },
+    {
+      "epoch": 0.6029536189523883,
+      "learning_rate": 0.0010232572767161728,
+      "loss": 1.0004,
+      "step": 7839
+    },
+    {
+      "epoch": 0.6030305361126067,
+      "learning_rate": 0.0010229136214400837,
+      "loss": 1.1045,
+      "step": 7840
+    },
+    {
+      "epoch": 0.6031074532728252,
+      "learning_rate": 0.001022569994021565,
+      "loss": 1.2416,
+      "step": 7841
+    },
+    {
+      "epoch": 0.6031843704330436,
+      "learning_rate": 0.001022226394480682,
+      "loss": 0.8683,
+      "step": 7842
+    },
+    {
+      "epoch": 0.603261287593262,
+      "learning_rate": 0.0010218828228374984,
+      "loss": 1.2691,
+      "step": 7843
+    },
+    {
+      "epoch": 0.6033382047534805,
+      "learning_rate": 0.001021539279112074,
+      "loss": 1.0257,
+      "step": 7844
+    },
+    {
+      "epoch": 0.603415121913699,
+      "learning_rate": 0.0010211957633244708,
+      "loss": 1.4517,
+      "step": 7845
+    },
+    {
+      "epoch": 0.6034920390739174,
+      "learning_rate": 0.0010208522754947453,
+      "loss": 0.9462,
+      "step": 7846
+    },
+    {
+      "epoch": 0.6035689562341359,
+      "learning_rate": 0.0010205088156429547,
+      "loss": 1.1834,
+      "step": 7847
+    },
+    {
+      "epoch": 0.6036458733943543,
+      "learning_rate": 0.0010201653837891545,
+      "loss": 1.2009,
+      "step": 7848
+    },
+    {
+      "epoch": 0.6037227905545727,
+      "learning_rate": 0.0010198219799533973,
+      "loss": 1.239,
+      "step": 7849
+    },
+    {
+      "epoch": 0.6037997077147912,
+      "learning_rate": 0.001019478604155735,
+      "loss": 1.2887,
+      "step": 7850
+    },
+    {
+      "epoch": 0.6038766248750096,
+      "learning_rate": 0.0010191352564162181,
+      "loss": 1.0743,
+      "step": 7851
+    },
+    {
+      "epoch": 0.6039535420352281,
+      "learning_rate": 0.0010187919367548944,
+      "loss": 0.8909,
+      "step": 7852
+    },
+    {
+      "epoch": 0.6040304591954465,
+      "learning_rate": 0.0010184486451918114,
+      "loss": 1.0651,
+      "step": 7853
+    },
+    {
+      "epoch": 0.6041073763556649,
+      "learning_rate": 0.0010181053817470143,
+      "loss": 0.8757,
+      "step": 7854
+    },
+    {
+      "epoch": 0.6041842935158834,
+      "learning_rate": 0.001017762146440545,
+      "loss": 1.1608,
+      "step": 7855
+    },
+    {
+      "epoch": 0.6042612106761018,
+      "learning_rate": 0.0010174189392924474,
+      "loss": 1.1209,
+      "step": 7856
+    },
+    {
+      "epoch": 0.6043381278363202,
+      "learning_rate": 0.0010170757603227606,
+      "loss": 1.1479,
+      "step": 7857
+    },
+    {
+      "epoch": 0.6044150449965388,
+      "learning_rate": 0.0010167326095515234,
+      "loss": 0.9983,
+      "step": 7858
+    },
+    {
+      "epoch": 0.6044919621567572,
+      "learning_rate": 0.0010163894869987728,
+      "loss": 1.2428,
+      "step": 7859
+    },
+    {
+      "epoch": 0.6045688793169757,
+      "learning_rate": 0.0010160463926845438,
+      "loss": 1.0819,
+      "step": 7860
+    },
+    {
+      "epoch": 0.6046457964771941,
+      "learning_rate": 0.0010157033266288708,
+      "loss": 1.0805,
+      "step": 7861
+    },
+    {
+      "epoch": 0.6047227136374125,
+      "learning_rate": 0.001015360288851785,
+      "loss": 1.3644,
+      "step": 7862
+    },
+    {
+      "epoch": 0.604799630797631,
+      "learning_rate": 0.001015017279373317,
+      "loss": 1.123,
+      "step": 7863
+    },
+    {
+      "epoch": 0.6048765479578494,
+      "learning_rate": 0.0010146742982134957,
+      "loss": 1.2639,
+      "step": 7864
+    },
+    {
+      "epoch": 0.6049534651180678,
+      "learning_rate": 0.0010143313453923478,
+      "loss": 1.0517,
+      "step": 7865
+    },
+    {
+      "epoch": 0.6050303822782863,
+      "learning_rate": 0.0010139884209298985,
+      "loss": 1.2348,
+      "step": 7866
+    },
+    {
+      "epoch": 0.6051072994385047,
+      "learning_rate": 0.0010136455248461725,
+      "loss": 1.0953,
+      "step": 7867
+    },
+    {
+      "epoch": 0.6051842165987231,
+      "learning_rate": 0.001013302657161191,
+      "loss": 0.6503,
+      "step": 7868
+    },
+    {
+      "epoch": 0.6052611337589416,
+      "learning_rate": 0.001012959817894974,
+      "loss": 1.3275,
+      "step": 7869
+    },
+    {
+      "epoch": 0.60533805091916,
+      "learning_rate": 0.0010126170070675414,
+      "loss": 1.2339,
+      "step": 7870
+    },
+    {
+      "epoch": 0.6054149680793786,
+      "learning_rate": 0.001012274224698909,
+      "loss": 1.1524,
+      "step": 7871
+    },
+    {
+      "epoch": 0.605491885239597,
+      "learning_rate": 0.0010119314708090935,
+      "loss": 0.8756,
+      "step": 7872
+    },
+    {
+      "epoch": 0.6055688023998154,
+      "learning_rate": 0.001011588745418108,
+      "loss": 1.1831,
+      "step": 7873
+    },
+    {
+      "epoch": 0.6056457195600339,
+      "learning_rate": 0.0010112460485459645,
+      "loss": 1.2186,
+      "step": 7874
+    },
+    {
+      "epoch": 0.6057226367202523,
+      "learning_rate": 0.0010109033802126737,
+      "loss": 1.281,
+      "step": 7875
+    },
+    {
+      "epoch": 0.6057995538804707,
+      "learning_rate": 0.0010105607404382443,
+      "loss": 1.1512,
+      "step": 7876
+    },
+    {
+      "epoch": 0.6058764710406892,
+      "learning_rate": 0.0010102181292426832,
+      "loss": 0.9668,
+      "step": 7877
+    },
+    {
+      "epoch": 0.6059533882009076,
+      "learning_rate": 0.001009875546645996,
+      "loss": 1.2598,
+      "step": 7878
+    },
+    {
+      "epoch": 0.6060303053611261,
+      "learning_rate": 0.0010095329926681864,
+      "loss": 1.3061,
+      "step": 7879
+    },
+    {
+      "epoch": 0.6061072225213445,
+      "learning_rate": 0.001009190467329257,
+      "loss": 1.2282,
+      "step": 7880
+    },
+    {
+      "epoch": 0.6061841396815629,
+      "learning_rate": 0.0010088479706492075,
+      "loss": 1.0917,
+      "step": 7881
+    },
+    {
+      "epoch": 0.6062610568417814,
+      "learning_rate": 0.0010085055026480365,
+      "loss": 1.3351,
+      "step": 7882
+    },
+    {
+      "epoch": 0.6063379740019998,
+      "learning_rate": 0.0010081630633457415,
+      "loss": 1.4174,
+      "step": 7883
+    },
+    {
+      "epoch": 0.6064148911622183,
+      "learning_rate": 0.0010078206527623181,
+      "loss": 0.8827,
+      "step": 7884
+    },
+    {
+      "epoch": 0.6064918083224368,
+      "learning_rate": 0.0010074782709177594,
+      "loss": 1.1687,
+      "step": 7885
+    },
+    {
+      "epoch": 0.6065687254826552,
+      "learning_rate": 0.001007135917832058,
+      "loss": 1.2557,
+      "step": 7886
+    },
+    {
+      "epoch": 0.6066456426428736,
+      "learning_rate": 0.001006793593525204,
+      "loss": 1.2157,
+      "step": 7887
+    },
+    {
+      "epoch": 0.6067225598030921,
+      "learning_rate": 0.0010064512980171856,
+      "loss": 1.1743,
+      "step": 7888
+    },
+    {
+      "epoch": 0.6067994769633105,
+      "learning_rate": 0.001006109031327991,
+      "loss": 1.2562,
+      "step": 7889
+    },
+    {
+      "epoch": 0.606876394123529,
+      "learning_rate": 0.001005766793477604,
+      "loss": 1.0796,
+      "step": 7890
+    },
+    {
+      "epoch": 0.6069533112837474,
+      "learning_rate": 0.0010054245844860095,
+      "loss": 1.1306,
+      "step": 7891
+    },
+    {
+      "epoch": 0.6070302284439658,
+      "learning_rate": 0.0010050824043731893,
+      "loss": 1.0124,
+      "step": 7892
+    },
+    {
+      "epoch": 0.6071071456041843,
+      "learning_rate": 0.0010047402531591224,
+      "loss": 1.2791,
+      "step": 7893
+    },
+    {
+      "epoch": 0.6071840627644027,
+      "learning_rate": 0.001004398130863789,
+      "loss": 1.3058,
+      "step": 7894
+    },
+    {
+      "epoch": 0.6072609799246211,
+      "learning_rate": 0.0010040560375071653,
+      "loss": 1.0634,
+      "step": 7895
+    },
+    {
+      "epoch": 0.6073378970848397,
+      "learning_rate": 0.0010037139731092258,
+      "loss": 1.2333,
+      "step": 7896
+    },
+    {
+      "epoch": 0.607414814245058,
+      "learning_rate": 0.001003371937689945,
+      "loss": 1.1136,
+      "step": 7897
+    },
+    {
+      "epoch": 0.6074917314052766,
+      "learning_rate": 0.0010030299312692942,
+      "loss": 1.0772,
+      "step": 7898
+    },
+    {
+      "epoch": 0.607568648565495,
+      "learning_rate": 0.0010026879538672438,
+      "loss": 1.3217,
+      "step": 7899
+    },
+    {
+      "epoch": 0.6076455657257134,
+      "learning_rate": 0.001002346005503762,
+      "loss": 0.9078,
+      "step": 7900
+    },
+    {
+      "epoch": 0.6077224828859319,
+      "learning_rate": 0.0010020040861988156,
+      "loss": 1.3689,
+      "step": 7901
+    },
+    {
+      "epoch": 0.6077994000461503,
+      "learning_rate": 0.0010016621959723696,
+      "loss": 1.2904,
+      "step": 7902
+    },
+    {
+      "epoch": 0.6078763172063687,
+      "learning_rate": 0.0010013203348443877,
+      "loss": 1.0072,
+      "step": 7903
+    },
+    {
+      "epoch": 0.6079532343665872,
+      "learning_rate": 0.0010009785028348302,
+      "loss": 0.9837,
+      "step": 7904
+    },
+    {
+      "epoch": 0.6080301515268056,
+      "learning_rate": 0.001000636699963659,
+      "loss": 1.3249,
+      "step": 7905
+    },
+    {
+      "epoch": 0.608107068687024,
+      "learning_rate": 0.001000294926250831,
+      "loss": 1.1906,
+      "step": 7906
+    },
+    {
+      "epoch": 0.6081839858472425,
+      "learning_rate": 0.0009999531817163026,
+      "loss": 1.3993,
+      "step": 7907
+    },
+    {
+      "epoch": 0.6082609030074609,
+      "learning_rate": 0.0009996114663800295,
+      "loss": 1.2132,
+      "step": 7908
+    },
+    {
+      "epoch": 0.6083378201676795,
+      "learning_rate": 0.0009992697802619639,
+      "loss": 0.8712,
+      "step": 7909
+    },
+    {
+      "epoch": 0.6084147373278979,
+      "learning_rate": 0.000998928123382058,
+      "loss": 1.2964,
+      "step": 7910
+    },
+    {
+      "epoch": 0.6084916544881163,
+      "learning_rate": 0.0009985864957602615,
+      "loss": 1.3279,
+      "step": 7911
+    },
+    {
+      "epoch": 0.6085685716483348,
+      "learning_rate": 0.0009982448974165216,
+      "loss": 1.19,
+      "step": 7912
+    },
+    {
+      "epoch": 0.6086454888085532,
+      "learning_rate": 0.0009979033283707853,
+      "loss": 1.4645,
+      "step": 7913
+    },
+    {
+      "epoch": 0.6087224059687716,
+      "learning_rate": 0.0009975617886429972,
+      "loss": 1.7533,
+      "step": 7914
+    },
+    {
+      "epoch": 0.6087993231289901,
+      "learning_rate": 0.000997220278253099,
+      "loss": 0.7365,
+      "step": 7915
+    },
+    {
+      "epoch": 0.6088762402892085,
+      "learning_rate": 0.0009968787972210336,
+      "loss": 1.1548,
+      "step": 7916
+    },
+    {
+      "epoch": 0.608953157449427,
+      "learning_rate": 0.0009965373455667389,
+      "loss": 0.9278,
+      "step": 7917
+    },
+    {
+      "epoch": 0.6090300746096454,
+      "learning_rate": 0.0009961959233101539,
+      "loss": 0.9917,
+      "step": 7918
+    },
+    {
+      "epoch": 0.6091069917698638,
+      "learning_rate": 0.0009958545304712138,
+      "loss": 1.0208,
+      "step": 7919
+    },
+    {
+      "epoch": 0.6091839089300823,
+      "learning_rate": 0.000995513167069853,
+      "loss": 1.0637,
+      "step": 7920
+    },
+    {
+      "epoch": 0.6092608260903007,
+      "learning_rate": 0.0009951718331260042,
+      "loss": 1.2475,
+      "step": 7921
+    },
+    {
+      "epoch": 0.6093377432505191,
+      "learning_rate": 0.000994830528659598,
+      "loss": 1.2193,
+      "step": 7922
+    },
+    {
+      "epoch": 0.6094146604107377,
+      "learning_rate": 0.0009944892536905637,
+      "loss": 1.1543,
+      "step": 7923
+    },
+    {
+      "epoch": 0.6094915775709561,
+      "learning_rate": 0.000994148008238829,
+      "loss": 1.5268,
+      "step": 7924
+    },
+    {
+      "epoch": 0.6095684947311745,
+      "learning_rate": 0.0009938067923243195,
+      "loss": 1.1637,
+      "step": 7925
+    },
+    {
+      "epoch": 0.609645411891393,
+      "learning_rate": 0.0009934656059669578,
+      "loss": 1.1411,
+      "step": 7926
+    },
+    {
+      "epoch": 0.6097223290516114,
+      "learning_rate": 0.0009931244491866685,
+      "loss": 1.1863,
+      "step": 7927
+    },
+    {
+      "epoch": 0.6097992462118299,
+      "learning_rate": 0.00099278332200337,
+      "loss": 0.9868,
+      "step": 7928
+    },
+    {
+      "epoch": 0.6098761633720483,
+      "learning_rate": 0.000992442224436983,
+      "loss": 1.2126,
+      "step": 7929
+    },
+    {
+      "epoch": 0.6099530805322667,
+      "learning_rate": 0.0009921011565074227,
+      "loss": 1.0634,
+      "step": 7930
+    },
+    {
+      "epoch": 0.6100299976924852,
+      "learning_rate": 0.000991760118234605,
+      "loss": 1.1871,
+      "step": 7931
+    },
+    {
+      "epoch": 0.6101069148527036,
+      "learning_rate": 0.000991419109638444,
+      "loss": 1.1458,
+      "step": 7932
+    },
+    {
+      "epoch": 0.610183832012922,
+      "learning_rate": 0.0009910781307388514,
+      "loss": 0.8921,
+      "step": 7933
+    },
+    {
+      "epoch": 0.6102607491731405,
+      "learning_rate": 0.0009907371815557368,
+      "loss": 1.3179,
+      "step": 7934
+    },
+    {
+      "epoch": 0.610337666333359,
+      "learning_rate": 0.000990396262109009,
+      "loss": 1.1721,
+      "step": 7935
+    },
+    {
+      "epoch": 0.6104145834935775,
+      "learning_rate": 0.0009900553724185742,
+      "loss": 1.1723,
+      "step": 7936
+    },
+    {
+      "epoch": 0.6104915006537959,
+      "learning_rate": 0.0009897145125043383,
+      "loss": 1.1278,
+      "step": 7937
+    },
+    {
+      "epoch": 0.6105684178140143,
+      "learning_rate": 0.000989373682386204,
+      "loss": 1.3781,
+      "step": 7938
+    },
+    {
+      "epoch": 0.6106453349742328,
+      "learning_rate": 0.0009890328820840715,
+      "loss": 1.0431,
+      "step": 7939
+    },
+    {
+      "epoch": 0.6107222521344512,
+      "learning_rate": 0.0009886921116178426,
+      "loss": 1.6225,
+      "step": 7940
+    },
+    {
+      "epoch": 0.6107991692946696,
+      "learning_rate": 0.0009883513710074138,
+      "loss": 1.5153,
+      "step": 7941
+    },
+    {
+      "epoch": 0.6108760864548881,
+      "learning_rate": 0.0009880106602726815,
+      "loss": 1.1906,
+      "step": 7942
+    },
+    {
+      "epoch": 0.6109530036151065,
+      "learning_rate": 0.0009876699794335404,
+      "loss": 1.0122,
+      "step": 7943
+    },
+    {
+      "epoch": 0.6110299207753249,
+      "learning_rate": 0.0009873293285098833,
+      "loss": 1.2095,
+      "step": 7944
+    },
+    {
+      "epoch": 0.6111068379355434,
+      "learning_rate": 0.0009869887075216006,
+      "loss": 1.3353,
+      "step": 7945
+    },
+    {
+      "epoch": 0.6111837550957618,
+      "learning_rate": 0.0009866481164885821,
+      "loss": 1.1863,
+      "step": 7946
+    },
+    {
+      "epoch": 0.6112606722559804,
+      "learning_rate": 0.000986307555430715,
+      "loss": 1.2304,
+      "step": 7947
+    },
+    {
+      "epoch": 0.6113375894161988,
+      "learning_rate": 0.0009859670243678855,
+      "loss": 0.8882,
+      "step": 7948
+    },
+    {
+      "epoch": 0.6114145065764172,
+      "learning_rate": 0.0009856265233199773,
+      "loss": 1.0656,
+      "step": 7949
+    },
+    {
+      "epoch": 0.6114914237366357,
+      "learning_rate": 0.0009852860523068714,
+      "loss": 1.01,
+      "step": 7950
+    },
+    {
+      "epoch": 0.6115683408968541,
+      "learning_rate": 0.0009849456113484505,
+      "loss": 0.9631,
+      "step": 7951
+    },
+    {
+      "epoch": 0.6116452580570725,
+      "learning_rate": 0.0009846052004645917,
+      "loss": 1.7442,
+      "step": 7952
+    },
+    {
+      "epoch": 0.611722175217291,
+      "learning_rate": 0.000984264819675172,
+      "loss": 1.1071,
+      "step": 7953
+    },
+    {
+      "epoch": 0.6117990923775094,
+      "learning_rate": 0.0009839244690000675,
+      "loss": 1.0134,
+      "step": 7954
+    },
+    {
+      "epoch": 0.6118760095377279,
+      "learning_rate": 0.0009835841484591506,
+      "loss": 1.2044,
+      "step": 7955
+    },
+    {
+      "epoch": 0.6119529266979463,
+      "learning_rate": 0.000983243858072294,
+      "loss": 1.4278,
+      "step": 7956
+    },
+    {
+      "epoch": 0.6120298438581647,
+      "learning_rate": 0.0009829035978593669,
+      "loss": 1.6517,
+      "step": 7957
+    },
+    {
+      "epoch": 0.6121067610183832,
+      "learning_rate": 0.000982563367840237,
+      "loss": 1.0621,
+      "step": 7958
+    },
+    {
+      "epoch": 0.6121836781786016,
+      "learning_rate": 0.000982223168034772,
+      "loss": 1.0888,
+      "step": 7959
+    },
+    {
+      "epoch": 0.61226059533882,
+      "learning_rate": 0.0009818829984628362,
+      "loss": 1.0146,
+      "step": 7960
+    },
+    {
+      "epoch": 0.6123375124990386,
+      "learning_rate": 0.000981542859144291,
+      "loss": 0.7239,
+      "step": 7961
+    },
+    {
+      "epoch": 0.612414429659257,
+      "learning_rate": 0.0009812027500989997,
+      "loss": 1.0811,
+      "step": 7962
+    },
+    {
+      "epoch": 0.6124913468194755,
+      "learning_rate": 0.0009808626713468194,
+      "loss": 0.8919,
+      "step": 7963
+    },
+    {
+      "epoch": 0.6125682639796939,
+      "learning_rate": 0.0009805226229076096,
+      "loss": 1.3856,
+      "step": 7964
+    },
+    {
+      "epoch": 0.6126451811399123,
+      "learning_rate": 0.000980182604801225,
+      "loss": 0.9222,
+      "step": 7965
+    },
+    {
+      "epoch": 0.6127220983001308,
+      "learning_rate": 0.0009798426170475197,
+      "loss": 1.085,
+      "step": 7966
+    },
+    {
+      "epoch": 0.6127990154603492,
+      "learning_rate": 0.0009795026596663462,
+      "loss": 1.1534,
+      "step": 7967
+    },
+    {
+      "epoch": 0.6128759326205676,
+      "learning_rate": 0.000979162732677555,
+      "loss": 1.3604,
+      "step": 7968
+    },
+    {
+      "epoch": 0.6129528497807861,
+      "learning_rate": 0.000978822836100994,
+      "loss": 1.1482,
+      "step": 7969
+    },
+    {
+      "epoch": 0.6130297669410045,
+      "learning_rate": 0.0009784829699565115,
+      "loss": 1.1764,
+      "step": 7970
+    },
+    {
+      "epoch": 0.6131066841012229,
+      "learning_rate": 0.000978143134263952,
+      "loss": 1.4959,
+      "step": 7971
+    },
+    {
+      "epoch": 0.6131836012614414,
+      "learning_rate": 0.000977803329043158,
+      "loss": 1.2298,
+      "step": 7972
+    },
+    {
+      "epoch": 0.6132605184216599,
+      "learning_rate": 0.0009774635543139726,
+      "loss": 0.8295,
+      "step": 7973
+    },
+    {
+      "epoch": 0.6133374355818784,
+      "learning_rate": 0.000977123810096234,
+      "loss": 1.0659,
+      "step": 7974
+    },
+    {
+      "epoch": 0.6134143527420968,
+      "learning_rate": 0.0009767840964097817,
+      "loss": 1.3188,
+      "step": 7975
+    },
+    {
+      "epoch": 0.6134912699023152,
+      "learning_rate": 0.0009764444132744513,
+      "loss": 1.2851,
+      "step": 7976
+    },
+    {
+      "epoch": 0.6135681870625337,
+      "learning_rate": 0.0009761047607100767,
+      "loss": 0.8915,
+      "step": 7977
+    },
+    {
+      "epoch": 0.6136451042227521,
+      "learning_rate": 0.0009757651387364916,
+      "loss": 0.7588,
+      "step": 7978
+    },
+    {
+      "epoch": 0.6137220213829705,
+      "learning_rate": 0.0009754255473735263,
+      "loss": 1.1247,
+      "step": 7979
+    },
+    {
+      "epoch": 0.613798938543189,
+      "learning_rate": 0.0009750859866410097,
+      "loss": 1.2587,
+      "step": 7980
+    },
+    {
+      "epoch": 0.6138758557034074,
+      "learning_rate": 0.0009747464565587698,
+      "loss": 1.2248,
+      "step": 7981
+    },
+    {
+      "epoch": 0.6139527728636259,
+      "learning_rate": 0.0009744069571466314,
+      "loss": 1.265,
+      "step": 7982
+    },
+    {
+      "epoch": 0.6140296900238443,
+      "learning_rate": 0.0009740674884244186,
+      "loss": 1.3845,
+      "step": 7983
+    },
+    {
+      "epoch": 0.6141066071840627,
+      "learning_rate": 0.0009737280504119539,
+      "loss": 1.1153,
+      "step": 7984
+    },
+    {
+      "epoch": 0.6141835243442813,
+      "learning_rate": 0.0009733886431290557,
+      "loss": 1.1525,
+      "step": 7985
+    },
+    {
+      "epoch": 0.6142604415044997,
+      "learning_rate": 0.0009730492665955443,
+      "loss": 1.2444,
+      "step": 7986
+    },
+    {
+      "epoch": 0.6143373586647181,
+      "learning_rate": 0.0009727099208312354,
+      "loss": 1.5033,
+      "step": 7987
+    },
+    {
+      "epoch": 0.6144142758249366,
+      "learning_rate": 0.0009723706058559431,
+      "loss": 1.3061,
+      "step": 7988
+    },
+    {
+      "epoch": 0.614491192985155,
+      "learning_rate": 0.0009720313216894813,
+      "loss": 1.4493,
+      "step": 7989
+    },
+    {
+      "epoch": 0.6145681101453734,
+      "learning_rate": 0.0009716920683516612,
+      "loss": 1.0055,
+      "step": 7990
+    },
+    {
+      "epoch": 0.6146450273055919,
+      "learning_rate": 0.0009713528458622911,
+      "loss": 1.2471,
+      "step": 7991
+    },
+    {
+      "epoch": 0.6147219444658103,
+      "learning_rate": 0.0009710136542411798,
+      "loss": 1.2029,
+      "step": 7992
+    },
+    {
+      "epoch": 0.6147988616260288,
+      "learning_rate": 0.0009706744935081323,
+      "loss": 0.9112,
+      "step": 7993
+    },
+    {
+      "epoch": 0.6148757787862472,
+      "learning_rate": 0.000970335363682953,
+      "loss": 1.1652,
+      "step": 7994
+    },
+    {
+      "epoch": 0.6149526959464656,
+      "learning_rate": 0.0009699962647854441,
+      "loss": 0.9244,
+      "step": 7995
+    },
+    {
+      "epoch": 0.6150296131066841,
+      "learning_rate": 0.0009696571968354048,
+      "loss": 1.2673,
+      "step": 7996
+    },
+    {
+      "epoch": 0.6151065302669025,
+      "learning_rate": 0.0009693181598526353,
+      "loss": 1.3341,
+      "step": 7997
+    },
+    {
+      "epoch": 0.6151834474271209,
+      "learning_rate": 0.0009689791538569313,
+      "loss": 1.091,
+      "step": 7998
+    },
+    {
+      "epoch": 0.6152603645873395,
+      "learning_rate": 0.0009686401788680875,
+      "loss": 1.5006,
+      "step": 7999
+    },
+    {
+      "epoch": 0.6153372817475579,
+      "learning_rate": 0.0009683012349058979,
+      "loss": 1.2326,
+      "step": 8000
+    },
+    {
+      "epoch": 0.6154141989077764,
+      "learning_rate": 0.0009679623219901529,
+      "loss": 1.1959,
+      "step": 8001
+    },
+    {
+      "epoch": 0.6154911160679948,
+      "learning_rate": 0.0009676234401406427,
+      "loss": 1.3407,
+      "step": 8002
+    },
+    {
+      "epoch": 0.6155680332282132,
+      "learning_rate": 0.0009672845893771546,
+      "loss": 1.1276,
+      "step": 8003
+    },
+    {
+      "epoch": 0.6156449503884317,
+      "learning_rate": 0.0009669457697194744,
+      "loss": 1.2509,
+      "step": 8004
+    },
+    {
+      "epoch": 0.6157218675486501,
+      "learning_rate": 0.0009666069811873863,
+      "loss": 1.461,
+      "step": 8005
+    },
+    {
+      "epoch": 0.6157987847088685,
+      "learning_rate": 0.000966268223800673,
+      "loss": 1.1217,
+      "step": 8006
+    },
+    {
+      "epoch": 0.615875701869087,
+      "learning_rate": 0.0009659294975791134,
+      "loss": 0.864,
+      "step": 8007
+    },
+    {
+      "epoch": 0.6159526190293054,
+      "learning_rate": 0.0009655908025424879,
+      "loss": 1.0173,
+      "step": 8008
+    },
+    {
+      "epoch": 0.6160295361895238,
+      "learning_rate": 0.0009652521387105719,
+      "loss": 1.4209,
+      "step": 8009
+    },
+    {
+      "epoch": 0.6161064533497423,
+      "learning_rate": 0.0009649135061031409,
+      "loss": 1.2029,
+      "step": 8010
+    },
+    {
+      "epoch": 0.6161833705099607,
+      "learning_rate": 0.0009645749047399679,
+      "loss": 1.157,
+      "step": 8011
+    },
+    {
+      "epoch": 0.6162602876701793,
+      "learning_rate": 0.000964236334640824,
+      "loss": 1.4025,
+      "step": 8012
+    },
+    {
+      "epoch": 0.6163372048303977,
+      "learning_rate": 0.0009638977958254792,
+      "loss": 1.1947,
+      "step": 8013
+    },
+    {
+      "epoch": 0.6164141219906161,
+      "learning_rate": 0.0009635592883137008,
+      "loss": 0.9063,
+      "step": 8014
+    },
+    {
+      "epoch": 0.6164910391508346,
+      "learning_rate": 0.0009632208121252544,
+      "loss": 1.0217,
+      "step": 8015
+    },
+    {
+      "epoch": 0.616567956311053,
+      "learning_rate": 0.0009628823672799045,
+      "loss": 0.8972,
+      "step": 8016
+    },
+    {
+      "epoch": 0.6166448734712714,
+      "learning_rate": 0.0009625439537974129,
+      "loss": 0.9665,
+      "step": 8017
+    },
+    {
+      "epoch": 0.6167217906314899,
+      "learning_rate": 0.0009622055716975396,
+      "loss": 1.0596,
+      "step": 8018
+    },
+    {
+      "epoch": 0.6167987077917083,
+      "learning_rate": 0.0009618672210000439,
+      "loss": 0.9571,
+      "step": 8019
+    },
+    {
+      "epoch": 0.6168756249519268,
+      "learning_rate": 0.0009615289017246816,
+      "loss": 1.1336,
+      "step": 8020
+    },
+    {
+      "epoch": 0.6169525421121452,
+      "learning_rate": 0.0009611906138912084,
+      "loss": 1.3222,
+      "step": 8021
+    },
+    {
+      "epoch": 0.6170294592723636,
+      "learning_rate": 0.0009608523575193765,
+      "loss": 1.5397,
+      "step": 8022
+    },
+    {
+      "epoch": 0.6171063764325821,
+      "learning_rate": 0.0009605141326289372,
+      "loss": 1.0812,
+      "step": 8023
+    },
+    {
+      "epoch": 0.6171832935928006,
+      "learning_rate": 0.0009601759392396401,
+      "loss": 1.4692,
+      "step": 8024
+    },
+    {
+      "epoch": 0.617260210753019,
+      "learning_rate": 0.0009598377773712327,
+      "loss": 0.9726,
+      "step": 8025
+    },
+    {
+      "epoch": 0.6173371279132375,
+      "learning_rate": 0.00095949964704346,
+      "loss": 0.8921,
+      "step": 8026
+    },
+    {
+      "epoch": 0.6174140450734559,
+      "learning_rate": 0.0009591615482760664,
+      "loss": 1.3792,
+      "step": 8027
+    },
+    {
+      "epoch": 0.6174909622336743,
+      "learning_rate": 0.0009588234810887937,
+      "loss": 1.0445,
+      "step": 8028
+    },
+    {
+      "epoch": 0.6175678793938928,
+      "learning_rate": 0.0009584854455013817,
+      "loss": 0.7489,
+      "step": 8029
+    },
+    {
+      "epoch": 0.6176447965541112,
+      "learning_rate": 0.000958147441533569,
+      "loss": 1.0757,
+      "step": 8030
+    },
+    {
+      "epoch": 0.6177217137143297,
+      "learning_rate": 0.0009578094692050918,
+      "loss": 1.0493,
+      "step": 8031
+    },
+    {
+      "epoch": 0.6177986308745481,
+      "learning_rate": 0.0009574715285356849,
+      "loss": 1.0729,
+      "step": 8032
+    },
+    {
+      "epoch": 0.6178755480347665,
+      "learning_rate": 0.0009571336195450813,
+      "loss": 1.4625,
+      "step": 8033
+    },
+    {
+      "epoch": 0.617952465194985,
+      "learning_rate": 0.0009567957422530109,
+      "loss": 1.0553,
+      "step": 8034
+    },
+    {
+      "epoch": 0.6180293823552034,
+      "learning_rate": 0.0009564578966792035,
+      "loss": 1.4732,
+      "step": 8035
+    },
+    {
+      "epoch": 0.6181062995154218,
+      "learning_rate": 0.0009561200828433858,
+      "loss": 1.2348,
+      "step": 8036
+    },
+    {
+      "epoch": 0.6181832166756404,
+      "learning_rate": 0.000955782300765283,
+      "loss": 1.2797,
+      "step": 8037
+    },
+    {
+      "epoch": 0.6182601338358588,
+      "learning_rate": 0.0009554445504646193,
+      "loss": 0.8568,
+      "step": 8038
+    },
+    {
+      "epoch": 0.6183370509960773,
+      "learning_rate": 0.0009551068319611153,
+      "loss": 1.1488,
+      "step": 8039
+    },
+    {
+      "epoch": 0.6184139681562957,
+      "learning_rate": 0.0009547691452744917,
+      "loss": 1.1749,
+      "step": 8040
+    },
+    {
+      "epoch": 0.6184908853165141,
+      "learning_rate": 0.0009544314904244658,
+      "loss": 1.3081,
+      "step": 8041
+    },
+    {
+      "epoch": 0.6185678024767326,
+      "learning_rate": 0.0009540938674307536,
+      "loss": 1.0842,
+      "step": 8042
+    },
+    {
+      "epoch": 0.618644719636951,
+      "learning_rate": 0.0009537562763130697,
+      "loss": 0.9682,
+      "step": 8043
+    },
+    {
+      "epoch": 0.6187216367971694,
+      "learning_rate": 0.0009534187170911263,
+      "loss": 0.7129,
+      "step": 8044
+    },
+    {
+      "epoch": 0.6187985539573879,
+      "learning_rate": 0.0009530811897846328,
+      "loss": 1.0388,
+      "step": 8045
+    },
+    {
+      "epoch": 0.6188754711176063,
+      "learning_rate": 0.0009527436944132996,
+      "loss": 1.1447,
+      "step": 8046
+    },
+    {
+      "epoch": 0.6189523882778247,
+      "learning_rate": 0.0009524062309968321,
+      "loss": 0.8394,
+      "step": 8047
+    },
+    {
+      "epoch": 0.6190293054380432,
+      "learning_rate": 0.000952068799554935,
+      "loss": 1.229,
+      "step": 8048
+    },
+    {
+      "epoch": 0.6191062225982616,
+      "learning_rate": 0.0009517314001073121,
+      "loss": 1.0077,
+      "step": 8049
+    },
+    {
+      "epoch": 0.6191831397584802,
+      "learning_rate": 0.000951394032673664,
+      "loss": 1.2297,
+      "step": 8050
+    },
+    {
+      "epoch": 0.6192600569186986,
+      "learning_rate": 0.0009510566972736901,
+      "loss": 1.1122,
+      "step": 8051
+    },
+    {
+      "epoch": 0.619336974078917,
+      "learning_rate": 0.000950719393927088,
+      "loss": 0.9952,
+      "step": 8052
+    },
+    {
+      "epoch": 0.6194138912391355,
+      "learning_rate": 0.0009503821226535525,
+      "loss": 0.8947,
+      "step": 8053
+    },
+    {
+      "epoch": 0.6194908083993539,
+      "learning_rate": 0.000950044883472778,
+      "loss": 1.2,
+      "step": 8054
+    },
+    {
+      "epoch": 0.6195677255595723,
+      "learning_rate": 0.0009497076764044561,
+      "loss": 1.0742,
+      "step": 8055
+    },
+    {
+      "epoch": 0.6196446427197908,
+      "learning_rate": 0.0009493705014682757,
+      "loss": 1.3496,
+      "step": 8056
+    },
+    {
+      "epoch": 0.6197215598800092,
+      "learning_rate": 0.0009490333586839266,
+      "loss": 1.0115,
+      "step": 8057
+    },
+    {
+      "epoch": 0.6197984770402277,
+      "learning_rate": 0.0009486962480710927,
+      "loss": 1.3346,
+      "step": 8058
+    },
+    {
+      "epoch": 0.6198753942004461,
+      "learning_rate": 0.0009483591696494606,
+      "loss": 0.9056,
+      "step": 8059
+    },
+    {
+      "epoch": 0.6199523113606645,
+      "learning_rate": 0.0009480221234387108,
+      "loss": 1.3529,
+      "step": 8060
+    },
+    {
+      "epoch": 0.620029228520883,
+      "learning_rate": 0.0009476851094585246,
+      "loss": 1.3914,
+      "step": 8061
+    },
+    {
+      "epoch": 0.6201061456811015,
+      "learning_rate": 0.0009473481277285805,
+      "loss": 1.0591,
+      "step": 8062
+    },
+    {
+      "epoch": 0.6201830628413199,
+      "learning_rate": 0.0009470111782685554,
+      "loss": 1.2498,
+      "step": 8063
+    },
+    {
+      "epoch": 0.6202599800015384,
+      "learning_rate": 0.0009466742610981235,
+      "loss": 1.0873,
+      "step": 8064
+    },
+    {
+      "epoch": 0.6203368971617568,
+      "learning_rate": 0.0009463373762369584,
+      "loss": 1.4817,
+      "step": 8065
+    },
+    {
+      "epoch": 0.6204138143219753,
+      "learning_rate": 0.0009460005237047306,
+      "loss": 1.1098,
+      "step": 8066
+    },
+    {
+      "epoch": 0.6204907314821937,
+      "learning_rate": 0.00094566370352111,
+      "loss": 1.1081,
+      "step": 8067
+    },
+    {
+      "epoch": 0.6205676486424121,
+      "learning_rate": 0.0009453269157057637,
+      "loss": 1.1561,
+      "step": 8068
+    },
+    {
+      "epoch": 0.6206445658026306,
+      "learning_rate": 0.0009449901602783558,
+      "loss": 0.9915,
+      "step": 8069
+    },
+    {
+      "epoch": 0.620721482962849,
+      "learning_rate": 0.000944653437258552,
+      "loss": 1.3577,
+      "step": 8070
+    },
+    {
+      "epoch": 0.6207984001230674,
+      "learning_rate": 0.0009443167466660124,
+      "loss": 1.1501,
+      "step": 8071
+    },
+    {
+      "epoch": 0.6208753172832859,
+      "learning_rate": 0.0009439800885203968,
+      "loss": 1.0286,
+      "step": 8072
+    },
+    {
+      "epoch": 0.6209522344435043,
+      "learning_rate": 0.0009436434628413636,
+      "loss": 1.3252,
+      "step": 8073
+    },
+    {
+      "epoch": 0.6210291516037227,
+      "learning_rate": 0.0009433068696485684,
+      "loss": 1.2765,
+      "step": 8074
+    },
+    {
+      "epoch": 0.6211060687639413,
+      "learning_rate": 0.0009429703089616649,
+      "loss": 1.0048,
+      "step": 8075
+    },
+    {
+      "epoch": 0.6211829859241597,
+      "learning_rate": 0.0009426337808003058,
+      "loss": 1.0308,
+      "step": 8076
+    },
+    {
+      "epoch": 0.6212599030843782,
+      "learning_rate": 0.0009422972851841411,
+      "loss": 1.1436,
+      "step": 8077
+    },
+    {
+      "epoch": 0.6213368202445966,
+      "learning_rate": 0.0009419608221328191,
+      "loss": 1.4121,
+      "step": 8078
+    },
+    {
+      "epoch": 0.621413737404815,
+      "learning_rate": 0.0009416243916659869,
+      "loss": 1.1917,
+      "step": 8079
+    },
+    {
+      "epoch": 0.6214906545650335,
+      "learning_rate": 0.0009412879938032872,
+      "loss": 1.535,
+      "step": 8080
+    },
+    {
+      "epoch": 0.6215675717252519,
+      "learning_rate": 0.000940951628564365,
+      "loss": 1.042,
+      "step": 8081
+    },
+    {
+      "epoch": 0.6216444888854703,
+      "learning_rate": 0.0009406152959688593,
+      "loss": 0.7926,
+      "step": 8082
+    },
+    {
+      "epoch": 0.6217214060456888,
+      "learning_rate": 0.0009402789960364093,
+      "loss": 0.8164,
+      "step": 8083
+    },
+    {
+      "epoch": 0.6217983232059072,
+      "learning_rate": 0.0009399427287866524,
+      "loss": 1.0363,
+      "step": 8084
+    },
+    {
+      "epoch": 0.6218752403661257,
+      "learning_rate": 0.0009396064942392228,
+      "loss": 0.9432,
+      "step": 8085
+    },
+    {
+      "epoch": 0.6219521575263441,
+      "learning_rate": 0.0009392702924137543,
+      "loss": 1.1427,
+      "step": 8086
+    },
+    {
+      "epoch": 0.6220290746865625,
+      "learning_rate": 0.0009389341233298779,
+      "loss": 1.3593,
+      "step": 8087
+    },
+    {
+      "epoch": 0.6221059918467811,
+      "learning_rate": 0.0009385979870072225,
+      "loss": 1.2726,
+      "step": 8088
+    },
+    {
+      "epoch": 0.6221829090069995,
+      "learning_rate": 0.000938261883465416,
+      "loss": 1.041,
+      "step": 8089
+    },
+    {
+      "epoch": 0.6222598261672179,
+      "learning_rate": 0.0009379258127240838,
+      "loss": 0.8908,
+      "step": 8090
+    },
+    {
+      "epoch": 0.6223367433274364,
+      "learning_rate": 0.0009375897748028481,
+      "loss": 1.0971,
+      "step": 8091
+    },
+    {
+      "epoch": 0.6224136604876548,
+      "learning_rate": 0.0009372537697213328,
+      "loss": 0.9342,
+      "step": 8092
+    },
+    {
+      "epoch": 0.6224905776478732,
+      "learning_rate": 0.0009369177974991558,
+      "loss": 1.1111,
+      "step": 8093
+    },
+    {
+      "epoch": 0.6225674948080917,
+      "learning_rate": 0.0009365818581559354,
+      "loss": 1.1114,
+      "step": 8094
+    },
+    {
+      "epoch": 0.6226444119683101,
+      "learning_rate": 0.0009362459517112875,
+      "loss": 1.1003,
+      "step": 8095
+    },
+    {
+      "epoch": 0.6227213291285286,
+      "learning_rate": 0.0009359100781848259,
+      "loss": 0.9977,
+      "step": 8096
+    },
+    {
+      "epoch": 0.622798246288747,
+      "learning_rate": 0.000935574237596163,
+      "loss": 0.8794,
+      "step": 8097
+    },
+    {
+      "epoch": 0.6228751634489654,
+      "learning_rate": 0.0009352384299649086,
+      "loss": 1.4536,
+      "step": 8098
+    },
+    {
+      "epoch": 0.622952080609184,
+      "learning_rate": 0.0009349026553106707,
+      "loss": 1.2938,
+      "step": 8099
+    },
+    {
+      "epoch": 0.6230289977694023,
+      "learning_rate": 0.0009345669136530556,
+      "loss": 0.8505,
+      "step": 8100
+    },
+    {
+      "epoch": 0.6231059149296208,
+      "learning_rate": 0.0009342312050116683,
+      "loss": 0.7771,
+      "step": 8101
+    },
+    {
+      "epoch": 0.6231828320898393,
+      "learning_rate": 0.0009338955294061095,
+      "loss": 0.9755,
+      "step": 8102
+    },
+    {
+      "epoch": 0.6232597492500577,
+      "learning_rate": 0.0009335598868559817,
+      "loss": 0.7584,
+      "step": 8103
+    },
+    {
+      "epoch": 0.6233366664102762,
+      "learning_rate": 0.0009332242773808815,
+      "loss": 1.0345,
+      "step": 8104
+    },
+    {
+      "epoch": 0.6234135835704946,
+      "learning_rate": 0.0009328887010004075,
+      "loss": 1.3528,
+      "step": 8105
+    },
+    {
+      "epoch": 0.623490500730713,
+      "learning_rate": 0.0009325531577341527,
+      "loss": 1.3813,
+      "step": 8106
+    },
+    {
+      "epoch": 0.6235674178909315,
+      "learning_rate": 0.0009322176476017103,
+      "loss": 1.2704,
+      "step": 8107
+    },
+    {
+      "epoch": 0.6236443350511499,
+      "learning_rate": 0.0009318821706226715,
+      "loss": 1.2574,
+      "step": 8108
+    },
+    {
+      "epoch": 0.6237212522113683,
+      "learning_rate": 0.0009315467268166247,
+      "loss": 0.9548,
+      "step": 8109
+    },
+    {
+      "epoch": 0.6237981693715868,
+      "learning_rate": 0.0009312113162031566,
+      "loss": 1.5025,
+      "step": 8110
+    },
+    {
+      "epoch": 0.6238750865318052,
+      "learning_rate": 0.0009308759388018528,
+      "loss": 0.971,
+      "step": 8111
+    },
+    {
+      "epoch": 0.6239520036920236,
+      "learning_rate": 0.0009305405946322966,
+      "loss": 1.3512,
+      "step": 8112
+    },
+    {
+      "epoch": 0.6240289208522422,
+      "learning_rate": 0.0009302052837140676,
+      "loss": 1.215,
+      "step": 8113
+    },
+    {
+      "epoch": 0.6241058380124606,
+      "learning_rate": 0.0009298700060667468,
+      "loss": 1.0269,
+      "step": 8114
+    },
+    {
+      "epoch": 0.6241827551726791,
+      "learning_rate": 0.0009295347617099094,
+      "loss": 1.1533,
+      "step": 8115
+    },
+    {
+      "epoch": 0.6242596723328975,
+      "learning_rate": 0.0009291995506631328,
+      "loss": 1.1212,
+      "step": 8116
+    },
+    {
+      "epoch": 0.6243365894931159,
+      "learning_rate": 0.0009288643729459891,
+      "loss": 1.1638,
+      "step": 8117
+    },
+    {
+      "epoch": 0.6244135066533344,
+      "learning_rate": 0.0009285292285780496,
+      "loss": 1.1108,
+      "step": 8118
+    },
+    {
+      "epoch": 0.6244904238135528,
+      "learning_rate": 0.0009281941175788844,
+      "loss": 1.2896,
+      "step": 8119
+    },
+    {
+      "epoch": 0.6245673409737712,
+      "learning_rate": 0.0009278590399680605,
+      "loss": 1.2024,
+      "step": 8120
+    },
+    {
+      "epoch": 0.6246442581339897,
+      "learning_rate": 0.0009275239957651431,
+      "loss": 1.2816,
+      "step": 8121
+    },
+    {
+      "epoch": 0.6247211752942081,
+      "learning_rate": 0.0009271889849896966,
+      "loss": 1.5103,
+      "step": 8122
+    },
+    {
+      "epoch": 0.6247980924544266,
+      "learning_rate": 0.000926854007661282,
+      "loss": 1.2158,
+      "step": 8123
+    },
+    {
+      "epoch": 0.624875009614645,
+      "learning_rate": 0.0009265190637994594,
+      "loss": 0.923,
+      "step": 8124
+    },
+    {
+      "epoch": 0.6249519267748634,
+      "learning_rate": 0.0009261841534237867,
+      "loss": 0.9093,
+      "step": 8125
+    },
+    {
+      "epoch": 0.625028843935082,
+      "learning_rate": 0.0009258492765538183,
+      "loss": 1.359,
+      "step": 8126
+    },
+    {
+      "epoch": 0.6251057610953004,
+      "learning_rate": 0.0009255144332091101,
+      "loss": 0.9265,
+      "step": 8127
+    },
+    {
+      "epoch": 0.6251826782555188,
+      "learning_rate": 0.0009251796234092123,
+      "loss": 1.3336,
+      "step": 8128
+    },
+    {
+      "epoch": 0.6252595954157373,
+      "learning_rate": 0.0009248448471736753,
+      "loss": 1.2168,
+      "step": 8129
+    },
+    {
+      "epoch": 0.6253365125759557,
+      "learning_rate": 0.0009245101045220472,
+      "loss": 1.0361,
+      "step": 8130
+    },
+    {
+      "epoch": 0.6254134297361741,
+      "learning_rate": 0.000924175395473874,
+      "loss": 0.9382,
+      "step": 8131
+    },
+    {
+      "epoch": 0.6254903468963926,
+      "learning_rate": 0.0009238407200486992,
+      "loss": 1.1063,
+      "step": 8132
+    },
+    {
+      "epoch": 0.625567264056611,
+      "learning_rate": 0.0009235060782660655,
+      "loss": 1.5559,
+      "step": 8133
+    },
+    {
+      "epoch": 0.6256441812168295,
+      "learning_rate": 0.0009231714701455123,
+      "loss": 1.0572,
+      "step": 8134
+    },
+    {
+      "epoch": 0.6257210983770479,
+      "learning_rate": 0.0009228368957065788,
+      "loss": 1.2197,
+      "step": 8135
+    },
+    {
+      "epoch": 0.6257980155372663,
+      "learning_rate": 0.0009225023549688005,
+      "loss": 1.1472,
+      "step": 8136
+    },
+    {
+      "epoch": 0.6258749326974848,
+      "learning_rate": 0.0009221678479517106,
+      "loss": 1.4049,
+      "step": 8137
+    },
+    {
+      "epoch": 0.6259518498577032,
+      "learning_rate": 0.0009218333746748432,
+      "loss": 1.0489,
+      "step": 8138
+    },
+    {
+      "epoch": 0.6260287670179217,
+      "learning_rate": 0.0009214989351577274,
+      "loss": 1.3332,
+      "step": 8139
+    },
+    {
+      "epoch": 0.6261056841781402,
+      "learning_rate": 0.0009211645294198912,
+      "loss": 1.7117,
+      "step": 8140
+    },
+    {
+      "epoch": 0.6261826013383586,
+      "learning_rate": 0.0009208301574808618,
+      "loss": 1.2354,
+      "step": 8141
+    },
+    {
+      "epoch": 0.6262595184985771,
+      "learning_rate": 0.0009204958193601627,
+      "loss": 1.0934,
+      "step": 8142
+    },
+    {
+      "epoch": 0.6263364356587955,
+      "learning_rate": 0.0009201615150773169,
+      "loss": 1.2804,
+      "step": 8143
+    },
+    {
+      "epoch": 0.6264133528190139,
+      "learning_rate": 0.0009198272446518449,
+      "loss": 0.9547,
+      "step": 8144
+    },
+    {
+      "epoch": 0.6264902699792324,
+      "learning_rate": 0.0009194930081032641,
+      "loss": 0.735,
+      "step": 8145
+    },
+    {
+      "epoch": 0.6265671871394508,
+      "learning_rate": 0.000919158805451092,
+      "loss": 1.0617,
+      "step": 8146
+    },
+    {
+      "epoch": 0.6266441042996692,
+      "learning_rate": 0.0009188246367148426,
+      "loss": 1.3298,
+      "step": 8147
+    },
+    {
+      "epoch": 0.6267210214598877,
+      "learning_rate": 0.0009184905019140277,
+      "loss": 1.2708,
+      "step": 8148
+    },
+    {
+      "epoch": 0.6267979386201061,
+      "learning_rate": 0.0009181564010681593,
+      "loss": 0.9733,
+      "step": 8149
+    },
+    {
+      "epoch": 0.6268748557803245,
+      "learning_rate": 0.0009178223341967447,
+      "loss": 1.524,
+      "step": 8150
+    },
+    {
+      "epoch": 0.626951772940543,
+      "learning_rate": 0.0009174883013192904,
+      "loss": 1.04,
+      "step": 8151
+    },
+    {
+      "epoch": 0.6270286901007615,
+      "learning_rate": 0.0009171543024553016,
+      "loss": 0.8358,
+      "step": 8152
+    },
+    {
+      "epoch": 0.62710560726098,
+      "learning_rate": 0.0009168203376242803,
+      "loss": 1.2077,
+      "step": 8153
+    },
+    {
+      "epoch": 0.6271825244211984,
+      "learning_rate": 0.0009164864068457275,
+      "loss": 1.3779,
+      "step": 8154
+    },
+    {
+      "epoch": 0.6272594415814168,
+      "learning_rate": 0.0009161525101391416,
+      "loss": 0.8769,
+      "step": 8155
+    },
+    {
+      "epoch": 0.6273363587416353,
+      "learning_rate": 0.0009158186475240187,
+      "loss": 1.0898,
+      "step": 8156
+    },
+    {
+      "epoch": 0.6274132759018537,
+      "learning_rate": 0.0009154848190198539,
+      "loss": 1.0027,
+      "step": 8157
+    },
+    {
+      "epoch": 0.6274901930620721,
+      "learning_rate": 0.0009151510246461403,
+      "loss": 1.0532,
+      "step": 8158
+    },
+    {
+      "epoch": 0.6275671102222906,
+      "learning_rate": 0.0009148172644223669,
+      "loss": 1.1443,
+      "step": 8159
+    },
+    {
+      "epoch": 0.627644027382509,
+      "learning_rate": 0.000914483538368024,
+      "loss": 1.1674,
+      "step": 8160
+    },
+    {
+      "epoch": 0.6277209445427275,
+      "learning_rate": 0.0009141498465025968,
+      "loss": 1.0831,
+      "step": 8161
+    },
+    {
+      "epoch": 0.6277978617029459,
+      "learning_rate": 0.0009138161888455716,
+      "loss": 1.1636,
+      "step": 8162
+    },
+    {
+      "epoch": 0.6278747788631643,
+      "learning_rate": 0.0009134825654164295,
+      "loss": 0.9123,
+      "step": 8163
+    },
+    {
+      "epoch": 0.6279516960233829,
+      "learning_rate": 0.0009131489762346515,
+      "loss": 1.1613,
+      "step": 8164
+    },
+    {
+      "epoch": 0.6280286131836013,
+      "learning_rate": 0.0009128154213197168,
+      "loss": 1.4738,
+      "step": 8165
+    },
+    {
+      "epoch": 0.6281055303438197,
+      "learning_rate": 0.0009124819006911015,
+      "loss": 1.2864,
+      "step": 8166
+    },
+    {
+      "epoch": 0.6281824475040382,
+      "learning_rate": 0.0009121484143682799,
+      "loss": 1.0098,
+      "step": 8167
+    },
+    {
+      "epoch": 0.6282593646642566,
+      "learning_rate": 0.0009118149623707254,
+      "loss": 1.2836,
+      "step": 8168
+    },
+    {
+      "epoch": 0.6283362818244751,
+      "learning_rate": 0.000911481544717908,
+      "loss": 0.5749,
+      "step": 8169
+    },
+    {
+      "epoch": 0.6284131989846935,
+      "learning_rate": 0.0009111481614292968,
+      "loss": 1.1781,
+      "step": 8170
+    },
+    {
+      "epoch": 0.6284901161449119,
+      "learning_rate": 0.0009108148125243584,
+      "loss": 0.7363,
+      "step": 8171
+    },
+    {
+      "epoch": 0.6285670333051304,
+      "learning_rate": 0.0009104814980225562,
+      "loss": 1.1923,
+      "step": 8172
+    },
+    {
+      "epoch": 0.6286439504653488,
+      "learning_rate": 0.0009101482179433548,
+      "loss": 1.1633,
+      "step": 8173
+    },
+    {
+      "epoch": 0.6287208676255672,
+      "learning_rate": 0.0009098149723062132,
+      "loss": 1.5104,
+      "step": 8174
+    },
+    {
+      "epoch": 0.6287977847857857,
+      "learning_rate": 0.0009094817611305902,
+      "loss": 1.0493,
+      "step": 8175
+    },
+    {
+      "epoch": 0.6288747019460041,
+      "learning_rate": 0.0009091485844359428,
+      "loss": 0.8758,
+      "step": 8176
+    },
+    {
+      "epoch": 0.6289516191062225,
+      "learning_rate": 0.0009088154422417255,
+      "loss": 1.1665,
+      "step": 8177
+    },
+    {
+      "epoch": 0.6290285362664411,
+      "learning_rate": 0.0009084823345673903,
+      "loss": 0.8523,
+      "step": 8178
+    },
+    {
+      "epoch": 0.6291054534266595,
+      "learning_rate": 0.0009081492614323882,
+      "loss": 1.1758,
+      "step": 8179
+    },
+    {
+      "epoch": 0.629182370586878,
+      "learning_rate": 0.0009078162228561674,
+      "loss": 0.9003,
+      "step": 8180
+    },
+    {
+      "epoch": 0.6292592877470964,
+      "learning_rate": 0.0009074832188581749,
+      "loss": 0.8993,
+      "step": 8181
+    },
+    {
+      "epoch": 0.6293362049073148,
+      "learning_rate": 0.0009071502494578545,
+      "loss": 1.4683,
+      "step": 8182
+    },
+    {
+      "epoch": 0.6294131220675333,
+      "learning_rate": 0.0009068173146746488,
+      "loss": 0.7721,
+      "step": 8183
+    },
+    {
+      "epoch": 0.6294900392277517,
+      "learning_rate": 0.0009064844145279985,
+      "loss": 1.0183,
+      "step": 8184
+    },
+    {
+      "epoch": 0.6295669563879701,
+      "learning_rate": 0.0009061515490373421,
+      "loss": 1.184,
+      "step": 8185
+    },
+    {
+      "epoch": 0.6296438735481886,
+      "learning_rate": 0.0009058187182221148,
+      "loss": 1.1737,
+      "step": 8186
+    },
+    {
+      "epoch": 0.629720790708407,
+      "learning_rate": 0.0009054859221017524,
+      "loss": 1.2641,
+      "step": 8187
+    },
+    {
+      "epoch": 0.6297977078686255,
+      "learning_rate": 0.0009051531606956863,
+      "loss": 1.1508,
+      "step": 8188
+    },
+    {
+      "epoch": 0.629874625028844,
+      "learning_rate": 0.0009048204340233474,
+      "loss": 1.3649,
+      "step": 8189
+    },
+    {
+      "epoch": 0.6299515421890624,
+      "learning_rate": 0.0009044877421041633,
+      "loss": 1.2373,
+      "step": 8190
+    },
+    {
+      "epoch": 0.6300284593492809,
+      "learning_rate": 0.0009041550849575605,
+      "loss": 1.1942,
+      "step": 8191
+    },
+    {
+      "epoch": 0.6301053765094993,
+      "learning_rate": 0.0009038224626029636,
+      "loss": 0.8716,
+      "step": 8192
+    },
+    {
+      "epoch": 0.6301822936697177,
+      "learning_rate": 0.0009034898750597945,
+      "loss": 1.1884,
+      "step": 8193
+    },
+    {
+      "epoch": 0.6302592108299362,
+      "learning_rate": 0.0009031573223474729,
+      "loss": 0.9751,
+      "step": 8194
+    },
+    {
+      "epoch": 0.6303361279901546,
+      "learning_rate": 0.0009028248044854175,
+      "loss": 1.0067,
+      "step": 8195
+    },
+    {
+      "epoch": 0.630413045150373,
+      "learning_rate": 0.0009024923214930443,
+      "loss": 0.8566,
+      "step": 8196
+    },
+    {
+      "epoch": 0.6304899623105915,
+      "learning_rate": 0.0009021598733897664,
+      "loss": 1.1764,
+      "step": 8197
+    },
+    {
+      "epoch": 0.6305668794708099,
+      "learning_rate": 0.0009018274601949975,
+      "loss": 1.1409,
+      "step": 8198
+    },
+    {
+      "epoch": 0.6306437966310284,
+      "learning_rate": 0.0009014950819281459,
+      "loss": 1.0688,
+      "step": 8199
+    },
+    {
+      "epoch": 0.6307207137912468,
+      "learning_rate": 0.0009011627386086206,
+      "loss": 0.8363,
+      "step": 8200
+    },
+    {
+      "epoch": 0.6307976309514652,
+      "learning_rate": 0.0009008304302558269,
+      "loss": 0.9203,
+      "step": 8201
+    },
+    {
+      "epoch": 0.6308745481116838,
+      "learning_rate": 0.0009004981568891688,
+      "loss": 1.0089,
+      "step": 8202
+    },
+    {
+      "epoch": 0.6309514652719022,
+      "learning_rate": 0.0009001659185280482,
+      "loss": 1.5066,
+      "step": 8203
+    },
+    {
+      "epoch": 0.6310283824321206,
+      "learning_rate": 0.0008998337151918649,
+      "loss": 1.5113,
+      "step": 8204
+    },
+    {
+      "epoch": 0.6311052995923391,
+      "learning_rate": 0.000899501546900016,
+      "loss": 1.0865,
+      "step": 8205
+    },
+    {
+      "epoch": 0.6311822167525575,
+      "learning_rate": 0.000899169413671898,
+      "loss": 1.0892,
+      "step": 8206
+    },
+    {
+      "epoch": 0.631259133912776,
+      "learning_rate": 0.0008988373155269037,
+      "loss": 1.3598,
+      "step": 8207
+    },
+    {
+      "epoch": 0.6313360510729944,
+      "learning_rate": 0.0008985052524844254,
+      "loss": 1.3512,
+      "step": 8208
+    },
+    {
+      "epoch": 0.6314129682332128,
+      "learning_rate": 0.0008981732245638528,
+      "loss": 1.119,
+      "step": 8209
+    },
+    {
+      "epoch": 0.6314898853934313,
+      "learning_rate": 0.0008978412317845717,
+      "loss": 0.883,
+      "step": 8210
+    },
+    {
+      "epoch": 0.6315668025536497,
+      "learning_rate": 0.0008975092741659694,
+      "loss": 1.031,
+      "step": 8211
+    },
+    {
+      "epoch": 0.6316437197138681,
+      "learning_rate": 0.0008971773517274284,
+      "loss": 0.9025,
+      "step": 8212
+    },
+    {
+      "epoch": 0.6317206368740866,
+      "learning_rate": 0.0008968454644883296,
+      "loss": 1.5185,
+      "step": 8213
+    },
+    {
+      "epoch": 0.631797554034305,
+      "learning_rate": 0.0008965136124680533,
+      "loss": 1.0785,
+      "step": 8214
+    },
+    {
+      "epoch": 0.6318744711945234,
+      "learning_rate": 0.0008961817956859759,
+      "loss": 1.4106,
+      "step": 8215
+    },
+    {
+      "epoch": 0.631951388354742,
+      "learning_rate": 0.0008958500141614725,
+      "loss": 0.8974,
+      "step": 8216
+    },
+    {
+      "epoch": 0.6320283055149604,
+      "learning_rate": 0.0008955182679139165,
+      "loss": 1.1247,
+      "step": 8217
+    },
+    {
+      "epoch": 0.6321052226751789,
+      "learning_rate": 0.0008951865569626788,
+      "loss": 1.3157,
+      "step": 8218
+    },
+    {
+      "epoch": 0.6321821398353973,
+      "learning_rate": 0.0008948548813271285,
+      "loss": 0.957,
+      "step": 8219
+    },
+    {
+      "epoch": 0.6322590569956157,
+      "learning_rate": 0.0008945232410266328,
+      "loss": 1.1534,
+      "step": 8220
+    },
+    {
+      "epoch": 0.6323359741558342,
+      "learning_rate": 0.0008941916360805551,
+      "loss": 1.1409,
+      "step": 8221
+    },
+    {
+      "epoch": 0.6324128913160526,
+      "learning_rate": 0.00089386006650826,
+      "loss": 1.2061,
+      "step": 8222
+    },
+    {
+      "epoch": 0.632489808476271,
+      "learning_rate": 0.0008935285323291069,
+      "loss": 1.0451,
+      "step": 8223
+    },
+    {
+      "epoch": 0.6325667256364895,
+      "learning_rate": 0.0008931970335624548,
+      "loss": 1.1052,
+      "step": 8224
+    },
+    {
+      "epoch": 0.6326436427967079,
+      "learning_rate": 0.0008928655702276605,
+      "loss": 1.0595,
+      "step": 8225
+    },
+    {
+      "epoch": 0.6327205599569264,
+      "learning_rate": 0.0008925341423440779,
+      "loss": 1.0323,
+      "step": 8226
+    },
+    {
+      "epoch": 0.6327974771171448,
+      "learning_rate": 0.0008922027499310602,
+      "loss": 0.9534,
+      "step": 8227
+    },
+    {
+      "epoch": 0.6328743942773633,
+      "learning_rate": 0.0008918713930079574,
+      "loss": 0.8965,
+      "step": 8228
+    },
+    {
+      "epoch": 0.6329513114375818,
+      "learning_rate": 0.0008915400715941174,
+      "loss": 1.2773,
+      "step": 8229
+    },
+    {
+      "epoch": 0.6330282285978002,
+      "learning_rate": 0.0008912087857088868,
+      "loss": 1.133,
+      "step": 8230
+    },
+    {
+      "epoch": 0.6331051457580186,
+      "learning_rate": 0.0008908775353716103,
+      "loss": 1.3332,
+      "step": 8231
+    },
+    {
+      "epoch": 0.6331820629182371,
+      "learning_rate": 0.0008905463206016283,
+      "loss": 1.1906,
+      "step": 8232
+    },
+    {
+      "epoch": 0.6332589800784555,
+      "learning_rate": 0.0008902151414182828,
+      "loss": 1.9048,
+      "step": 8233
+    },
+    {
+      "epoch": 0.6333358972386739,
+      "learning_rate": 0.0008898839978409103,
+      "loss": 1.2236,
+      "step": 8234
+    },
+    {
+      "epoch": 0.6334128143988924,
+      "learning_rate": 0.0008895528898888467,
+      "loss": 0.9474,
+      "step": 8235
+    },
+    {
+      "epoch": 0.6334897315591108,
+      "learning_rate": 0.0008892218175814265,
+      "loss": 1.0123,
+      "step": 8236
+    },
+    {
+      "epoch": 0.6335666487193293,
+      "learning_rate": 0.0008888907809379804,
+      "loss": 1.0698,
+      "step": 8237
+    },
+    {
+      "epoch": 0.6336435658795477,
+      "learning_rate": 0.0008885597799778389,
+      "loss": 1.343,
+      "step": 8238
+    },
+    {
+      "epoch": 0.6337204830397661,
+      "learning_rate": 0.0008882288147203289,
+      "loss": 1.156,
+      "step": 8239
+    },
+    {
+      "epoch": 0.6337974001999847,
+      "learning_rate": 0.0008878978851847759,
+      "loss": 1.2981,
+      "step": 8240
+    },
+    {
+      "epoch": 0.633874317360203,
+      "learning_rate": 0.0008875669913905034,
+      "loss": 1.0384,
+      "step": 8241
+    },
+    {
+      "epoch": 0.6339512345204215,
+      "learning_rate": 0.0008872361333568332,
+      "loss": 0.9956,
+      "step": 8242
+    },
+    {
+      "epoch": 0.63402815168064,
+      "learning_rate": 0.0008869053111030825,
+      "loss": 1.084,
+      "step": 8243
+    },
+    {
+      "epoch": 0.6341050688408584,
+      "learning_rate": 0.0008865745246485708,
+      "loss": 0.9186,
+      "step": 8244
+    },
+    {
+      "epoch": 0.6341819860010769,
+      "learning_rate": 0.0008862437740126107,
+      "loss": 0.9003,
+      "step": 8245
+    },
+    {
+      "epoch": 0.6342589031612953,
+      "learning_rate": 0.0008859130592145175,
+      "loss": 1.1746,
+      "step": 8246
+    },
+    {
+      "epoch": 0.6343358203215137,
+      "learning_rate": 0.0008855823802736001,
+      "loss": 1.2765,
+      "step": 8247
+    },
+    {
+      "epoch": 0.6344127374817322,
+      "learning_rate": 0.0008852517372091679,
+      "loss": 1.2865,
+      "step": 8248
+    },
+    {
+      "epoch": 0.6344896546419506,
+      "learning_rate": 0.0008849211300405278,
+      "loss": 1.1054,
+      "step": 8249
+    },
+    {
+      "epoch": 0.634566571802169,
+      "learning_rate": 0.0008845905587869835,
+      "loss": 1.122,
+      "step": 8250
+    },
+    {
+      "epoch": 0.6346434889623875,
+      "learning_rate": 0.0008842600234678379,
+      "loss": 0.8618,
+      "step": 8251
+    },
+    {
+      "epoch": 0.6347204061226059,
+      "learning_rate": 0.0008839295241023915,
+      "loss": 1.2543,
+      "step": 8252
+    },
+    {
+      "epoch": 0.6347973232828243,
+      "learning_rate": 0.0008835990607099423,
+      "loss": 1.0198,
+      "step": 8253
+    },
+    {
+      "epoch": 0.6348742404430429,
+      "learning_rate": 0.0008832686333097857,
+      "loss": 1.2164,
+      "step": 8254
+    },
+    {
+      "epoch": 0.6349511576032613,
+      "learning_rate": 0.0008829382419212171,
+      "loss": 1.1041,
+      "step": 8255
+    },
+    {
+      "epoch": 0.6350280747634798,
+      "learning_rate": 0.0008826078865635267,
+      "loss": 0.9281,
+      "step": 8256
+    },
+    {
+      "epoch": 0.6351049919236982,
+      "learning_rate": 0.0008822775672560064,
+      "loss": 0.991,
+      "step": 8257
+    },
+    {
+      "epoch": 0.6351819090839166,
+      "learning_rate": 0.0008819472840179422,
+      "loss": 0.8556,
+      "step": 8258
+    },
+    {
+      "epoch": 0.6352588262441351,
+      "learning_rate": 0.0008816170368686199,
+      "loss": 1.1821,
+      "step": 8259
+    },
+    {
+      "epoch": 0.6353357434043535,
+      "learning_rate": 0.0008812868258273236,
+      "loss": 1.1493,
+      "step": 8260
+    },
+    {
+      "epoch": 0.6354126605645719,
+      "learning_rate": 0.0008809566509133343,
+      "loss": 0.9212,
+      "step": 8261
+    },
+    {
+      "epoch": 0.6354895777247904,
+      "learning_rate": 0.000880626512145931,
+      "loss": 1.503,
+      "step": 8262
+    },
+    {
+      "epoch": 0.6355664948850088,
+      "learning_rate": 0.0008802964095443917,
+      "loss": 1.1168,
+      "step": 8263
+    },
+    {
+      "epoch": 0.6356434120452273,
+      "learning_rate": 0.0008799663431279903,
+      "loss": 1.1888,
+      "step": 8264
+    },
+    {
+      "epoch": 0.6357203292054457,
+      "learning_rate": 0.0008796363129160009,
+      "loss": 1.1403,
+      "step": 8265
+    },
+    {
+      "epoch": 0.6357972463656641,
+      "learning_rate": 0.0008793063189276937,
+      "loss": 1.4732,
+      "step": 8266
+    },
+    {
+      "epoch": 0.6358741635258827,
+      "learning_rate": 0.0008789763611823366,
+      "loss": 1.4032,
+      "step": 8267
+    },
+    {
+      "epoch": 0.6359510806861011,
+      "learning_rate": 0.000878646439699198,
+      "loss": 1.2138,
+      "step": 8268
+    },
+    {
+      "epoch": 0.6360279978463195,
+      "learning_rate": 0.000878316554497541,
+      "loss": 1.1688,
+      "step": 8269
+    },
+    {
+      "epoch": 0.636104915006538,
+      "learning_rate": 0.0008779867055966279,
+      "loss": 1.0932,
+      "step": 8270
+    },
+    {
+      "epoch": 0.6361818321667564,
+      "learning_rate": 0.0008776568930157198,
+      "loss": 1.4304,
+      "step": 8271
+    },
+    {
+      "epoch": 0.6362587493269748,
+      "learning_rate": 0.0008773271167740741,
+      "loss": 1.0514,
+      "step": 8272
+    },
+    {
+      "epoch": 0.6363356664871933,
+      "learning_rate": 0.0008769973768909467,
+      "loss": 0.9654,
+      "step": 8273
+    },
+    {
+      "epoch": 0.6364125836474117,
+      "learning_rate": 0.0008766676733855921,
+      "loss": 0.9579,
+      "step": 8274
+    },
+    {
+      "epoch": 0.6364895008076302,
+      "learning_rate": 0.0008763380062772614,
+      "loss": 0.7052,
+      "step": 8275
+    },
+    {
+      "epoch": 0.6365664179678486,
+      "learning_rate": 0.0008760083755852048,
+      "loss": 0.9583,
+      "step": 8276
+    },
+    {
+      "epoch": 0.636643335128067,
+      "learning_rate": 0.0008756787813286695,
+      "loss": 1.5324,
+      "step": 8277
+    },
+    {
+      "epoch": 0.6367202522882855,
+      "learning_rate": 0.0008753492235269007,
+      "loss": 1.0465,
+      "step": 8278
+    },
+    {
+      "epoch": 0.636797169448504,
+      "learning_rate": 0.0008750197021991417,
+      "loss": 0.9068,
+      "step": 8279
+    },
+    {
+      "epoch": 0.6368740866087224,
+      "learning_rate": 0.0008746902173646337,
+      "loss": 0.9917,
+      "step": 8280
+    },
+    {
+      "epoch": 0.6369510037689409,
+      "learning_rate": 0.000874360769042615,
+      "loss": 0.9032,
+      "step": 8281
+    },
+    {
+      "epoch": 0.6370279209291593,
+      "learning_rate": 0.0008740313572523236,
+      "loss": 1.0142,
+      "step": 8282
+    },
+    {
+      "epoch": 0.6371048380893778,
+      "learning_rate": 0.0008737019820129937,
+      "loss": 1.4506,
+      "step": 8283
+    },
+    {
+      "epoch": 0.6371817552495962,
+      "learning_rate": 0.0008733726433438577,
+      "loss": 1.026,
+      "step": 8284
+    },
+    {
+      "epoch": 0.6372586724098146,
+      "learning_rate": 0.0008730433412641461,
+      "loss": 0.8937,
+      "step": 8285
+    },
+    {
+      "epoch": 0.6373355895700331,
+      "learning_rate": 0.0008727140757930866,
+      "loss": 1.0528,
+      "step": 8286
+    },
+    {
+      "epoch": 0.6374125067302515,
+      "learning_rate": 0.0008723848469499072,
+      "loss": 1.2463,
+      "step": 8287
+    },
+    {
+      "epoch": 0.6374894238904699,
+      "learning_rate": 0.0008720556547538299,
+      "loss": 1.297,
+      "step": 8288
+    },
+    {
+      "epoch": 0.6375663410506884,
+      "learning_rate": 0.000871726499224077,
+      "loss": 1.1878,
+      "step": 8289
+    },
+    {
+      "epoch": 0.6376432582109068,
+      "learning_rate": 0.000871397380379869,
+      "loss": 1.2336,
+      "step": 8290
+    },
+    {
+      "epoch": 0.6377201753711254,
+      "learning_rate": 0.0008710682982404233,
+      "loss": 1.1063,
+      "step": 8291
+    },
+    {
+      "epoch": 0.6377970925313438,
+      "learning_rate": 0.0008707392528249548,
+      "loss": 0.9645,
+      "step": 8292
+    },
+    {
+      "epoch": 0.6378740096915622,
+      "learning_rate": 0.0008704102441526772,
+      "loss": 0.6542,
+      "step": 8293
+    },
+    {
+      "epoch": 0.6379509268517807,
+      "learning_rate": 0.0008700812722428011,
+      "loss": 1.0287,
+      "step": 8294
+    },
+    {
+      "epoch": 0.6380278440119991,
+      "learning_rate": 0.0008697523371145367,
+      "loss": 1.1559,
+      "step": 8295
+    },
+    {
+      "epoch": 0.6381047611722175,
+      "learning_rate": 0.0008694234387870906,
+      "loss": 1.142,
+      "step": 8296
+    },
+    {
+      "epoch": 0.638181678332436,
+      "learning_rate": 0.0008690945772796657,
+      "loss": 1.09,
+      "step": 8297
+    },
+    {
+      "epoch": 0.6382585954926544,
+      "learning_rate": 0.0008687657526114668,
+      "loss": 1.2456,
+      "step": 8298
+    },
+    {
+      "epoch": 0.6383355126528728,
+      "learning_rate": 0.0008684369648016935,
+      "loss": 1.1513,
+      "step": 8299
+    },
+    {
+      "epoch": 0.6384124298130913,
+      "learning_rate": 0.0008681082138695441,
+      "loss": 1.1071,
+      "step": 8300
+    },
+    {
+      "epoch": 0.6384893469733097,
+      "learning_rate": 0.0008677794998342144,
+      "loss": 0.9978,
+      "step": 8301
+    },
+    {
+      "epoch": 0.6385662641335282,
+      "learning_rate": 0.0008674508227148981,
+      "loss": 0.801,
+      "step": 8302
+    },
+    {
+      "epoch": 0.6386431812937466,
+      "learning_rate": 0.000867122182530788,
+      "loss": 1.0011,
+      "step": 8303
+    },
+    {
+      "epoch": 0.638720098453965,
+      "learning_rate": 0.0008667935793010733,
+      "loss": 1.017,
+      "step": 8304
+    },
+    {
+      "epoch": 0.6387970156141836,
+      "learning_rate": 0.0008664650130449414,
+      "loss": 1.3035,
+      "step": 8305
+    },
+    {
+      "epoch": 0.638873932774402,
+      "learning_rate": 0.0008661364837815777,
+      "loss": 1.1366,
+      "step": 8306
+    },
+    {
+      "epoch": 0.6389508499346204,
+      "learning_rate": 0.0008658079915301652,
+      "loss": 1.0112,
+      "step": 8307
+    },
+    {
+      "epoch": 0.6390277670948389,
+      "learning_rate": 0.0008654795363098845,
+      "loss": 1.348,
+      "step": 8308
+    },
+    {
+      "epoch": 0.6391046842550573,
+      "learning_rate": 0.0008651511181399163,
+      "loss": 1.2637,
+      "step": 8309
+    },
+    {
+      "epoch": 0.6391816014152758,
+      "learning_rate": 0.0008648227370394344,
+      "loss": 1.4518,
+      "step": 8310
+    },
+    {
+      "epoch": 0.6392585185754942,
+      "learning_rate": 0.0008644943930276157,
+      "loss": 0.9448,
+      "step": 8311
+    },
+    {
+      "epoch": 0.6393354357357126,
+      "learning_rate": 0.0008641660861236315,
+      "loss": 0.8564,
+      "step": 8312
+    },
+    {
+      "epoch": 0.6394123528959311,
+      "learning_rate": 0.0008638378163466524,
+      "loss": 1.0912,
+      "step": 8313
+    },
+    {
+      "epoch": 0.6394892700561495,
+      "learning_rate": 0.0008635095837158461,
+      "loss": 1.1062,
+      "step": 8314
+    },
+    {
+      "epoch": 0.6395661872163679,
+      "learning_rate": 0.0008631813882503785,
+      "loss": 1.1459,
+      "step": 8315
+    },
+    {
+      "epoch": 0.6396431043765864,
+      "learning_rate": 0.0008628532299694129,
+      "loss": 1.1181,
+      "step": 8316
+    },
+    {
+      "epoch": 0.6397200215368049,
+      "learning_rate": 0.0008625251088921114,
+      "loss": 0.8185,
+      "step": 8317
+    },
+    {
+      "epoch": 0.6397969386970233,
+      "learning_rate": 0.0008621970250376342,
+      "loss": 1.2591,
+      "step": 8318
+    },
+    {
+      "epoch": 0.6398738558572418,
+      "learning_rate": 0.0008618689784251362,
+      "loss": 1.5552,
+      "step": 8319
+    },
+    {
+      "epoch": 0.6399507730174602,
+      "learning_rate": 0.0008615409690737739,
+      "loss": 0.9654,
+      "step": 8320
+    },
+    {
+      "epoch": 0.6400276901776787,
+      "learning_rate": 0.0008612129970026991,
+      "loss": 0.9854,
+      "step": 8321
+    },
+    {
+      "epoch": 0.6401046073378971,
+      "learning_rate": 0.0008608850622310646,
+      "loss": 1.2055,
+      "step": 8322
+    },
+    {
+      "epoch": 0.6401815244981155,
+      "learning_rate": 0.0008605571647780168,
+      "loss": 1.1933,
+      "step": 8323
+    },
+    {
+      "epoch": 0.640258441658334,
+      "learning_rate": 0.000860229304662702,
+      "loss": 1.2742,
+      "step": 8324
+    },
+    {
+      "epoch": 0.6403353588185524,
+      "learning_rate": 0.0008599014819042653,
+      "loss": 1.0467,
+      "step": 8325
+    },
+    {
+      "epoch": 0.6404122759787708,
+      "learning_rate": 0.0008595736965218483,
+      "loss": 1.3372,
+      "step": 8326
+    },
+    {
+      "epoch": 0.6404891931389893,
+      "learning_rate": 0.0008592459485345905,
+      "loss": 0.9231,
+      "step": 8327
+    },
+    {
+      "epoch": 0.6405661102992077,
+      "learning_rate": 0.0008589182379616296,
+      "loss": 1.0623,
+      "step": 8328
+    },
+    {
+      "epoch": 0.6406430274594263,
+      "learning_rate": 0.0008585905648221004,
+      "loss": 0.6794,
+      "step": 8329
+    },
+    {
+      "epoch": 0.6407199446196447,
+      "learning_rate": 0.0008582629291351369,
+      "loss": 1.6995,
+      "step": 8330
+    },
+    {
+      "epoch": 0.6407968617798631,
+      "learning_rate": 0.0008579353309198705,
+      "loss": 0.9766,
+      "step": 8331
+    },
+    {
+      "epoch": 0.6408737789400816,
+      "learning_rate": 0.0008576077701954282,
+      "loss": 1.282,
+      "step": 8332
+    },
+    {
+      "epoch": 0.6409506961003,
+      "learning_rate": 0.0008572802469809381,
+      "loss": 1.0178,
+      "step": 8333
+    },
+    {
+      "epoch": 0.6410276132605184,
+      "learning_rate": 0.0008569527612955245,
+      "loss": 1.1893,
+      "step": 8334
+    },
+    {
+      "epoch": 0.6411045304207369,
+      "learning_rate": 0.000856625313158309,
+      "loss": 1.3076,
+      "step": 8335
+    },
+    {
+      "epoch": 0.6411814475809553,
+      "learning_rate": 0.0008562979025884121,
+      "loss": 1.2457,
+      "step": 8336
+    },
+    {
+      "epoch": 0.6412583647411737,
+      "learning_rate": 0.0008559705296049515,
+      "loss": 1.2937,
+      "step": 8337
+    },
+    {
+      "epoch": 0.6413352819013922,
+      "learning_rate": 0.0008556431942270422,
+      "loss": 1.1312,
+      "step": 8338
+    },
+    {
+      "epoch": 0.6414121990616106,
+      "learning_rate": 0.000855315896473799,
+      "loss": 1.4275,
+      "step": 8339
+    },
+    {
+      "epoch": 0.6414891162218291,
+      "learning_rate": 0.0008549886363643326,
+      "loss": 0.9869,
+      "step": 8340
+    },
+    {
+      "epoch": 0.6415660333820475,
+      "learning_rate": 0.0008546614139177518,
+      "loss": 1.3316,
+      "step": 8341
+    },
+    {
+      "epoch": 0.6416429505422659,
+      "learning_rate": 0.0008543342291531636,
+      "loss": 0.914,
+      "step": 8342
+    },
+    {
+      "epoch": 0.6417198677024845,
+      "learning_rate": 0.0008540070820896721,
+      "loss": 1.2597,
+      "step": 8343
+    },
+    {
+      "epoch": 0.6417967848627029,
+      "learning_rate": 0.0008536799727463815,
+      "loss": 1.0967,
+      "step": 8344
+    },
+    {
+      "epoch": 0.6418737020229213,
+      "learning_rate": 0.0008533529011423902,
+      "loss": 1.4358,
+      "step": 8345
+    },
+    {
+      "epoch": 0.6419506191831398,
+      "learning_rate": 0.0008530258672967963,
+      "loss": 1.2265,
+      "step": 8346
+    },
+    {
+      "epoch": 0.6420275363433582,
+      "learning_rate": 0.000852698871228697,
+      "loss": 1.0294,
+      "step": 8347
+    },
+    {
+      "epoch": 0.6421044535035767,
+      "learning_rate": 0.0008523719129571852,
+      "loss": 0.8969,
+      "step": 8348
+    },
+    {
+      "epoch": 0.6421813706637951,
+      "learning_rate": 0.0008520449925013524,
+      "loss": 1.1363,
+      "step": 8349
+    },
+    {
+      "epoch": 0.6422582878240135,
+      "learning_rate": 0.0008517181098802875,
+      "loss": 0.9958,
+      "step": 8350
+    },
+    {
+      "epoch": 0.642335204984232,
+      "learning_rate": 0.0008513912651130774,
+      "loss": 0.9373,
+      "step": 8351
+    },
+    {
+      "epoch": 0.6424121221444504,
+      "learning_rate": 0.0008510644582188078,
+      "loss": 1.4266,
+      "step": 8352
+    },
+    {
+      "epoch": 0.6424890393046688,
+      "learning_rate": 0.0008507376892165616,
+      "loss": 1.178,
+      "step": 8353
+    },
+    {
+      "epoch": 0.6425659564648873,
+      "learning_rate": 0.000850410958125417,
+      "loss": 1.2896,
+      "step": 8354
+    },
+    {
+      "epoch": 0.6426428736251057,
+      "learning_rate": 0.0008500842649644541,
+      "loss": 1.1034,
+      "step": 8355
+    },
+    {
+      "epoch": 0.6427197907853242,
+      "learning_rate": 0.0008497576097527483,
+      "loss": 1.2321,
+      "step": 8356
+    },
+    {
+      "epoch": 0.6427967079455427,
+      "learning_rate": 0.0008494309925093735,
+      "loss": 1.1474,
+      "step": 8357
+    },
+    {
+      "epoch": 0.6428736251057611,
+      "learning_rate": 0.0008491044132534008,
+      "loss": 1.0116,
+      "step": 8358
+    },
+    {
+      "epoch": 0.6429505422659796,
+      "learning_rate": 0.0008487778720038993,
+      "loss": 1.0924,
+      "step": 8359
+    },
+    {
+      "epoch": 0.643027459426198,
+      "learning_rate": 0.0008484513687799372,
+      "loss": 1.1276,
+      "step": 8360
+    },
+    {
+      "epoch": 0.6431043765864164,
+      "learning_rate": 0.0008481249036005786,
+      "loss": 0.8738,
+      "step": 8361
+    },
+    {
+      "epoch": 0.6431812937466349,
+      "learning_rate": 0.0008477984764848864,
+      "loss": 1.0035,
+      "step": 8362
+    },
+    {
+      "epoch": 0.6432582109068533,
+      "learning_rate": 0.0008474720874519208,
+      "loss": 1.0124,
+      "step": 8363
+    },
+    {
+      "epoch": 0.6433351280670717,
+      "learning_rate": 0.00084714573652074,
+      "loss": 1.0515,
+      "step": 8364
+    },
+    {
+      "epoch": 0.6434120452272902,
+      "learning_rate": 0.0008468194237103994,
+      "loss": 1.1871,
+      "step": 8365
+    },
+    {
+      "epoch": 0.6434889623875086,
+      "learning_rate": 0.0008464931490399546,
+      "loss": 1.1569,
+      "step": 8366
+    },
+    {
+      "epoch": 0.6435658795477271,
+      "learning_rate": 0.0008461669125284548,
+      "loss": 1.3059,
+      "step": 8367
+    },
+    {
+      "epoch": 0.6436427967079456,
+      "learning_rate": 0.000845840714194951,
+      "loss": 1.0381,
+      "step": 8368
+    },
+    {
+      "epoch": 0.643719713868164,
+      "learning_rate": 0.0008455145540584894,
+      "loss": 1.3407,
+      "step": 8369
+    },
+    {
+      "epoch": 0.6437966310283825,
+      "learning_rate": 0.0008451884321381151,
+      "loss": 1.3718,
+      "step": 8370
+    },
+    {
+      "epoch": 0.6438735481886009,
+      "learning_rate": 0.0008448623484528708,
+      "loss": 1.267,
+      "step": 8371
+    },
+    {
+      "epoch": 0.6439504653488193,
+      "learning_rate": 0.000844536303021797,
+      "loss": 0.8983,
+      "step": 8372
+    },
+    {
+      "epoch": 0.6440273825090378,
+      "learning_rate": 0.0008442102958639305,
+      "loss": 1.2889,
+      "step": 8373
+    },
+    {
+      "epoch": 0.6441042996692562,
+      "learning_rate": 0.0008438843269983092,
+      "loss": 0.8949,
+      "step": 8374
+    },
+    {
+      "epoch": 0.6441812168294746,
+      "learning_rate": 0.0008435583964439663,
+      "loss": 0.898,
+      "step": 8375
+    },
+    {
+      "epoch": 0.6442581339896931,
+      "learning_rate": 0.0008432325042199317,
+      "loss": 1.0528,
+      "step": 8376
+    },
+    {
+      "epoch": 0.6443350511499115,
+      "learning_rate": 0.0008429066503452359,
+      "loss": 0.7661,
+      "step": 8377
+    },
+    {
+      "epoch": 0.64441196831013,
+      "learning_rate": 0.0008425808348389054,
+      "loss": 1.3702,
+      "step": 8378
+    },
+    {
+      "epoch": 0.6444888854703484,
+      "learning_rate": 0.000842255057719966,
+      "loss": 0.8219,
+      "step": 8379
+    },
+    {
+      "epoch": 0.6445658026305668,
+      "learning_rate": 0.0008419293190074389,
+      "loss": 1.3809,
+      "step": 8380
+    },
+    {
+      "epoch": 0.6446427197907854,
+      "learning_rate": 0.000841603618720344,
+      "loss": 1.9077,
+      "step": 8381
+    },
+    {
+      "epoch": 0.6447196369510038,
+      "learning_rate": 0.0008412779568777005,
+      "loss": 0.8228,
+      "step": 8382
+    },
+    {
+      "epoch": 0.6447965541112222,
+      "learning_rate": 0.0008409523334985238,
+      "loss": 1.405,
+      "step": 8383
+    },
+    {
+      "epoch": 0.6448734712714407,
+      "learning_rate": 0.0008406267486018268,
+      "loss": 1.0319,
+      "step": 8384
+    },
+    {
+      "epoch": 0.6449503884316591,
+      "learning_rate": 0.0008403012022066215,
+      "loss": 1.0342,
+      "step": 8385
+    },
+    {
+      "epoch": 0.6450273055918776,
+      "learning_rate": 0.0008399756943319157,
+      "loss": 1.2664,
+      "step": 8386
+    },
+    {
+      "epoch": 0.645104222752096,
+      "learning_rate": 0.0008396502249967177,
+      "loss": 0.888,
+      "step": 8387
+    },
+    {
+      "epoch": 0.6451811399123144,
+      "learning_rate": 0.000839324794220032,
+      "loss": 1.3294,
+      "step": 8388
+    },
+    {
+      "epoch": 0.6452580570725329,
+      "learning_rate": 0.0008389994020208587,
+      "loss": 1.1977,
+      "step": 8389
+    },
+    {
+      "epoch": 0.6453349742327513,
+      "learning_rate": 0.0008386740484181998,
+      "loss": 1.0861,
+      "step": 8390
+    },
+    {
+      "epoch": 0.6454118913929697,
+      "learning_rate": 0.0008383487334310523,
+      "loss": 1.0666,
+      "step": 8391
+    },
+    {
+      "epoch": 0.6454888085531882,
+      "learning_rate": 0.000838023457078412,
+      "loss": 0.9194,
+      "step": 8392
+    },
+    {
+      "epoch": 0.6455657257134066,
+      "learning_rate": 0.0008376982193792718,
+      "loss": 1.4254,
+      "step": 8393
+    },
+    {
+      "epoch": 0.6456426428736252,
+      "learning_rate": 0.0008373730203526226,
+      "loss": 1.1746,
+      "step": 8394
+    },
+    {
+      "epoch": 0.6457195600338436,
+      "learning_rate": 0.0008370478600174535,
+      "loss": 0.8879,
+      "step": 8395
+    },
+    {
+      "epoch": 0.645796477194062,
+      "learning_rate": 0.0008367227383927508,
+      "loss": 1.0596,
+      "step": 8396
+    },
+    {
+      "epoch": 0.6458733943542805,
+      "learning_rate": 0.0008363976554974987,
+      "loss": 1.005,
+      "step": 8397
+    },
+    {
+      "epoch": 0.6459503115144989,
+      "learning_rate": 0.0008360726113506792,
+      "loss": 1.2018,
+      "step": 8398
+    },
+    {
+      "epoch": 0.6460272286747173,
+      "learning_rate": 0.0008357476059712718,
+      "loss": 1.1786,
+      "step": 8399
+    },
+    {
+      "epoch": 0.6461041458349358,
+      "learning_rate": 0.0008354226393782534,
+      "loss": 1.2608,
+      "step": 8400
+    },
+    {
+      "epoch": 0.6461810629951542,
+      "learning_rate": 0.0008350977115906007,
+      "loss": 1.0666,
+      "step": 8401
+    },
+    {
+      "epoch": 0.6462579801553726,
+      "learning_rate": 0.0008347728226272851,
+      "loss": 1.3131,
+      "step": 8402
+    },
+    {
+      "epoch": 0.6463348973155911,
+      "learning_rate": 0.000834447972507277,
+      "loss": 1.5519,
+      "step": 8403
+    },
+    {
+      "epoch": 0.6464118144758095,
+      "learning_rate": 0.0008341231612495461,
+      "loss": 1.1245,
+      "step": 8404
+    },
+    {
+      "epoch": 0.646488731636028,
+      "learning_rate": 0.000833798388873058,
+      "loss": 0.9058,
+      "step": 8405
+    },
+    {
+      "epoch": 0.6465656487962465,
+      "learning_rate": 0.0008334736553967759,
+      "loss": 1.3004,
+      "step": 8406
+    },
+    {
+      "epoch": 0.6466425659564649,
+      "learning_rate": 0.0008331489608396618,
+      "loss": 1.427,
+      "step": 8407
+    },
+    {
+      "epoch": 0.6467194831166834,
+      "learning_rate": 0.0008328243052206743,
+      "loss": 1.0121,
+      "step": 8408
+    },
+    {
+      "epoch": 0.6467964002769018,
+      "learning_rate": 0.0008324996885587714,
+      "loss": 1.2156,
+      "step": 8409
+    },
+    {
+      "epoch": 0.6468733174371202,
+      "learning_rate": 0.0008321751108729075,
+      "loss": 1.3845,
+      "step": 8410
+    },
+    {
+      "epoch": 0.6469502345973387,
+      "learning_rate": 0.0008318505721820348,
+      "loss": 0.8668,
+      "step": 8411
+    },
+    {
+      "epoch": 0.6470271517575571,
+      "learning_rate": 0.0008315260725051036,
+      "loss": 1.0659,
+      "step": 8412
+    },
+    {
+      "epoch": 0.6471040689177756,
+      "learning_rate": 0.000831201611861061,
+      "loss": 1.1882,
+      "step": 8413
+    },
+    {
+      "epoch": 0.647180986077994,
+      "learning_rate": 0.0008308771902688544,
+      "loss": 0.8096,
+      "step": 8414
+    },
+    {
+      "epoch": 0.6472579032382124,
+      "learning_rate": 0.0008305528077474255,
+      "loss": 0.9452,
+      "step": 8415
+    },
+    {
+      "epoch": 0.6473348203984309,
+      "learning_rate": 0.0008302284643157153,
+      "loss": 1.2173,
+      "step": 8416
+    },
+    {
+      "epoch": 0.6474117375586493,
+      "learning_rate": 0.0008299041599926636,
+      "loss": 0.9444,
+      "step": 8417
+    },
+    {
+      "epoch": 0.6474886547188677,
+      "learning_rate": 0.0008295798947972063,
+      "loss": 0.9073,
+      "step": 8418
+    },
+    {
+      "epoch": 0.6475655718790863,
+      "learning_rate": 0.0008292556687482778,
+      "loss": 1.0288,
+      "step": 8419
+    },
+    {
+      "epoch": 0.6476424890393047,
+      "learning_rate": 0.0008289314818648098,
+      "loss": 1.2546,
+      "step": 8420
+    },
+    {
+      "epoch": 0.6477194061995231,
+      "learning_rate": 0.0008286073341657317,
+      "loss": 1.1336,
+      "step": 8421
+    },
+    {
+      "epoch": 0.6477963233597416,
+      "learning_rate": 0.0008282832256699709,
+      "loss": 0.7539,
+      "step": 8422
+    },
+    {
+      "epoch": 0.64787324051996,
+      "learning_rate": 0.0008279591563964528,
+      "loss": 0.8988,
+      "step": 8423
+    },
+    {
+      "epoch": 0.6479501576801785,
+      "learning_rate": 0.0008276351263641002,
+      "loss": 1.029,
+      "step": 8424
+    },
+    {
+      "epoch": 0.6480270748403969,
+      "learning_rate": 0.0008273111355918332,
+      "loss": 0.9604,
+      "step": 8425
+    },
+    {
+      "epoch": 0.6481039920006153,
+      "learning_rate": 0.0008269871840985701,
+      "loss": 1.4591,
+      "step": 8426
+    },
+    {
+      "epoch": 0.6481809091608338,
+      "learning_rate": 0.000826663271903226,
+      "loss": 1.145,
+      "step": 8427
+    },
+    {
+      "epoch": 0.6482578263210522,
+      "learning_rate": 0.0008263393990247165,
+      "loss": 1.0779,
+      "step": 8428
+    },
+    {
+      "epoch": 0.6483347434812706,
+      "learning_rate": 0.000826015565481951,
+      "loss": 1.0784,
+      "step": 8429
+    },
+    {
+      "epoch": 0.6484116606414891,
+      "learning_rate": 0.0008256917712938382,
+      "loss": 1.0924,
+      "step": 8430
+    },
+    {
+      "epoch": 0.6484885778017075,
+      "learning_rate": 0.0008253680164792865,
+      "loss": 1.3725,
+      "step": 8431
+    },
+    {
+      "epoch": 0.6485654949619261,
+      "learning_rate": 0.0008250443010571994,
+      "loss": 1.3314,
+      "step": 8432
+    },
+    {
+      "epoch": 0.6486424121221445,
+      "learning_rate": 0.0008247206250464788,
+      "loss": 0.8478,
+      "step": 8433
+    },
+    {
+      "epoch": 0.6487193292823629,
+      "learning_rate": 0.0008243969884660247,
+      "loss": 1.0771,
+      "step": 8434
+    },
+    {
+      "epoch": 0.6487962464425814,
+      "learning_rate": 0.0008240733913347341,
+      "loss": 0.9964,
+      "step": 8435
+    },
+    {
+      "epoch": 0.6488731636027998,
+      "learning_rate": 0.000823749833671503,
+      "loss": 1.3977,
+      "step": 8436
+    },
+    {
+      "epoch": 0.6489500807630182,
+      "learning_rate": 0.0008234263154952245,
+      "loss": 0.9625,
+      "step": 8437
+    },
+    {
+      "epoch": 0.6490269979232367,
+      "learning_rate": 0.0008231028368247876,
+      "loss": 0.9187,
+      "step": 8438
+    },
+    {
+      "epoch": 0.6491039150834551,
+      "learning_rate": 0.0008227793976790818,
+      "loss": 1.3996,
+      "step": 8439
+    },
+    {
+      "epoch": 0.6491808322436735,
+      "learning_rate": 0.0008224559980769928,
+      "loss": 0.9809,
+      "step": 8440
+    },
+    {
+      "epoch": 0.649257749403892,
+      "learning_rate": 0.0008221326380374044,
+      "loss": 1.2785,
+      "step": 8441
+    },
+    {
+      "epoch": 0.6493346665641104,
+      "learning_rate": 0.0008218093175791976,
+      "loss": 1.2468,
+      "step": 8442
+    },
+    {
+      "epoch": 0.649411583724329,
+      "learning_rate": 0.0008214860367212509,
+      "loss": 1.1019,
+      "step": 8443
+    },
+    {
+      "epoch": 0.6494885008845473,
+      "learning_rate": 0.0008211627954824422,
+      "loss": 1.2127,
+      "step": 8444
+    },
+    {
+      "epoch": 0.6495654180447658,
+      "learning_rate": 0.0008208395938816455,
+      "loss": 1.3368,
+      "step": 8445
+    },
+    {
+      "epoch": 0.6496423352049843,
+      "learning_rate": 0.0008205164319377327,
+      "loss": 1.2483,
+      "step": 8446
+    },
+    {
+      "epoch": 0.6497192523652027,
+      "learning_rate": 0.0008201933096695735,
+      "loss": 1.2783,
+      "step": 8447
+    },
+    {
+      "epoch": 0.6497961695254211,
+      "learning_rate": 0.0008198702270960354,
+      "loss": 1.3545,
+      "step": 8448
+    },
+    {
+      "epoch": 0.6498730866856396,
+      "learning_rate": 0.000819547184235983,
+      "loss": 1.2456,
+      "step": 8449
+    },
+    {
+      "epoch": 0.649950003845858,
+      "learning_rate": 0.0008192241811082808,
+      "loss": 1.492,
+      "step": 8450
+    },
+    {
+      "epoch": 0.6500269210060765,
+      "learning_rate": 0.0008189012177317869,
+      "loss": 1.0569,
+      "step": 8451
+    },
+    {
+      "epoch": 0.6501038381662949,
+      "learning_rate": 0.0008185782941253615,
+      "loss": 1.1524,
+      "step": 8452
+    },
+    {
+      "epoch": 0.6501807553265133,
+      "learning_rate": 0.0008182554103078596,
+      "loss": 1.4448,
+      "step": 8453
+    },
+    {
+      "epoch": 0.6502576724867318,
+      "learning_rate": 0.0008179325662981347,
+      "loss": 1.024,
+      "step": 8454
+    },
+    {
+      "epoch": 0.6503345896469502,
+      "learning_rate": 0.0008176097621150382,
+      "loss": 1.0922,
+      "step": 8455
+    },
+    {
+      "epoch": 0.6504115068071686,
+      "learning_rate": 0.000817286997777419,
+      "loss": 2.074,
+      "step": 8456
+    },
+    {
+      "epoch": 0.6504884239673872,
+      "learning_rate": 0.0008169642733041226,
+      "loss": 1.0375,
+      "step": 8457
+    },
+    {
+      "epoch": 0.6505653411276056,
+      "learning_rate": 0.000816641588713995,
+      "loss": 1.3542,
+      "step": 8458
+    },
+    {
+      "epoch": 0.650642258287824,
+      "learning_rate": 0.0008163189440258777,
+      "loss": 0.7715,
+      "step": 8459
+    },
+    {
+      "epoch": 0.6507191754480425,
+      "learning_rate": 0.0008159963392586084,
+      "loss": 1.0151,
+      "step": 8460
+    },
+    {
+      "epoch": 0.6507960926082609,
+      "learning_rate": 0.0008156737744310266,
+      "loss": 1.2908,
+      "step": 8461
+    },
+    {
+      "epoch": 0.6508730097684794,
+      "learning_rate": 0.0008153512495619654,
+      "loss": 1.152,
+      "step": 8462
+    },
+    {
+      "epoch": 0.6509499269286978,
+      "learning_rate": 0.0008150287646702597,
+      "loss": 0.6677,
+      "step": 8463
+    },
+    {
+      "epoch": 0.6510268440889162,
+      "learning_rate": 0.0008147063197747376,
+      "loss": 1.2038,
+      "step": 8464
+    },
+    {
+      "epoch": 0.6511037612491347,
+      "learning_rate": 0.0008143839148942271,
+      "loss": 0.8156,
+      "step": 8465
+    },
+    {
+      "epoch": 0.6511806784093531,
+      "learning_rate": 0.000814061550047555,
+      "loss": 1.3697,
+      "step": 8466
+    },
+    {
+      "epoch": 0.6512575955695715,
+      "learning_rate": 0.0008137392252535436,
+      "loss": 0.9923,
+      "step": 8467
+    },
+    {
+      "epoch": 0.65133451272979,
+      "learning_rate": 0.0008134169405310144,
+      "loss": 1.8218,
+      "step": 8468
+    },
+    {
+      "epoch": 0.6514114298900084,
+      "learning_rate": 0.0008130946958987855,
+      "loss": 1.0249,
+      "step": 8469
+    },
+    {
+      "epoch": 0.651488347050227,
+      "learning_rate": 0.0008127724913756724,
+      "loss": 1.2627,
+      "step": 8470
+    },
+    {
+      "epoch": 0.6515652642104454,
+      "learning_rate": 0.0008124503269804905,
+      "loss": 0.9481,
+      "step": 8471
+    },
+    {
+      "epoch": 0.6516421813706638,
+      "learning_rate": 0.0008121282027320509,
+      "loss": 0.8972,
+      "step": 8472
+    },
+    {
+      "epoch": 0.6517190985308823,
+      "learning_rate": 0.0008118061186491616,
+      "loss": 1.292,
+      "step": 8473
+    },
+    {
+      "epoch": 0.6517960156911007,
+      "learning_rate": 0.0008114840747506307,
+      "loss": 1.2915,
+      "step": 8474
+    },
+    {
+      "epoch": 0.6518729328513191,
+      "learning_rate": 0.000811162071055262,
+      "loss": 1.2656,
+      "step": 8475
+    },
+    {
+      "epoch": 0.6519498500115376,
+      "learning_rate": 0.0008108401075818582,
+      "loss": 1.2514,
+      "step": 8476
+    },
+    {
+      "epoch": 0.652026767171756,
+      "learning_rate": 0.0008105181843492184,
+      "loss": 1.3778,
+      "step": 8477
+    },
+    {
+      "epoch": 0.6521036843319744,
+      "learning_rate": 0.0008101963013761407,
+      "loss": 1.355,
+      "step": 8478
+    },
+    {
+      "epoch": 0.6521806014921929,
+      "learning_rate": 0.0008098744586814192,
+      "loss": 1.2155,
+      "step": 8479
+    },
+    {
+      "epoch": 0.6522575186524113,
+      "learning_rate": 0.0008095526562838476,
+      "loss": 0.9347,
+      "step": 8480
+    },
+    {
+      "epoch": 0.6523344358126298,
+      "learning_rate": 0.0008092308942022163,
+      "loss": 1.192,
+      "step": 8481
+    },
+    {
+      "epoch": 0.6524113529728482,
+      "learning_rate": 0.000808909172455313,
+      "loss": 1.2006,
+      "step": 8482
+    },
+    {
+      "epoch": 0.6524882701330667,
+      "learning_rate": 0.0008085874910619232,
+      "loss": 0.7732,
+      "step": 8483
+    },
+    {
+      "epoch": 0.6525651872932852,
+      "learning_rate": 0.0008082658500408298,
+      "loss": 0.8882,
+      "step": 8484
+    },
+    {
+      "epoch": 0.6526421044535036,
+      "learning_rate": 0.0008079442494108155,
+      "loss": 0.8389,
+      "step": 8485
+    },
+    {
+      "epoch": 0.652719021613722,
+      "learning_rate": 0.0008076226891906573,
+      "loss": 0.9276,
+      "step": 8486
+    },
+    {
+      "epoch": 0.6527959387739405,
+      "learning_rate": 0.0008073011693991309,
+      "loss": 1.6867,
+      "step": 8487
+    },
+    {
+      "epoch": 0.6528728559341589,
+      "learning_rate": 0.0008069796900550119,
+      "loss": 1.0525,
+      "step": 8488
+    },
+    {
+      "epoch": 0.6529497730943774,
+      "learning_rate": 0.0008066582511770712,
+      "loss": 1.1896,
+      "step": 8489
+    },
+    {
+      "epoch": 0.6530266902545958,
+      "learning_rate": 0.0008063368527840777,
+      "loss": 1.534,
+      "step": 8490
+    },
+    {
+      "epoch": 0.6531036074148142,
+      "learning_rate": 0.0008060154948947982,
+      "loss": 1.292,
+      "step": 8491
+    },
+    {
+      "epoch": 0.6531805245750327,
+      "learning_rate": 0.0008056941775279967,
+      "loss": 1.1956,
+      "step": 8492
+    },
+    {
+      "epoch": 0.6532574417352511,
+      "learning_rate": 0.0008053729007024363,
+      "loss": 1.2427,
+      "step": 8493
+    },
+    {
+      "epoch": 0.6533343588954695,
+      "learning_rate": 0.0008050516644368769,
+      "loss": 1.2715,
+      "step": 8494
+    },
+    {
+      "epoch": 0.653411276055688,
+      "learning_rate": 0.0008047304687500737,
+      "loss": 1.4022,
+      "step": 8495
+    },
+    {
+      "epoch": 0.6534881932159065,
+      "learning_rate": 0.0008044093136607836,
+      "loss": 1.4004,
+      "step": 8496
+    },
+    {
+      "epoch": 0.6535651103761249,
+      "learning_rate": 0.0008040881991877588,
+      "loss": 1.005,
+      "step": 8497
+    },
+    {
+      "epoch": 0.6536420275363434,
+      "learning_rate": 0.0008037671253497492,
+      "loss": 0.9758,
+      "step": 8498
+    },
+    {
+      "epoch": 0.6537189446965618,
+      "learning_rate": 0.0008034460921655027,
+      "loss": 1.2197,
+      "step": 8499
+    },
+    {
+      "epoch": 0.6537958618567803,
+      "learning_rate": 0.0008031250996537642,
+      "loss": 1.2212,
+      "step": 8500
+    },
+    {
+      "epoch": 0.6538727790169987,
+      "learning_rate": 0.000802804147833278,
+      "loss": 1.3095,
+      "step": 8501
+    },
+    {
+      "epoch": 0.6539496961772171,
+      "learning_rate": 0.0008024832367227844,
+      "loss": 0.8511,
+      "step": 8502
+    },
+    {
+      "epoch": 0.6540266133374356,
+      "learning_rate": 0.0008021623663410215,
+      "loss": 0.9117,
+      "step": 8503
+    },
+    {
+      "epoch": 0.654103530497654,
+      "learning_rate": 0.0008018415367067251,
+      "loss": 1.19,
+      "step": 8504
+    },
+    {
+      "epoch": 0.6541804476578724,
+      "learning_rate": 0.0008015207478386291,
+      "loss": 1.0384,
+      "step": 8505
+    },
+    {
+      "epoch": 0.6542573648180909,
+      "learning_rate": 0.000801199999755464,
+      "loss": 1.12,
+      "step": 8506
+    },
+    {
+      "epoch": 0.6543342819783093,
+      "learning_rate": 0.0008008792924759602,
+      "loss": 0.9334,
+      "step": 8507
+    },
+    {
+      "epoch": 0.6544111991385279,
+      "learning_rate": 0.0008005586260188422,
+      "loss": 1.1652,
+      "step": 8508
+    },
+    {
+      "epoch": 0.6544881162987463,
+      "learning_rate": 0.0008002380004028352,
+      "loss": 0.9915,
+      "step": 8509
+    },
+    {
+      "epoch": 0.6545650334589647,
+      "learning_rate": 0.000799917415646661,
+      "loss": 1.2801,
+      "step": 8510
+    },
+    {
+      "epoch": 0.6546419506191832,
+      "learning_rate": 0.0007995968717690383,
+      "loss": 1.2619,
+      "step": 8511
+    },
+    {
+      "epoch": 0.6547188677794016,
+      "learning_rate": 0.0007992763687886843,
+      "loss": 0.9215,
+      "step": 8512
+    },
+    {
+      "epoch": 0.65479578493962,
+      "learning_rate": 0.0007989559067243133,
+      "loss": 1.0984,
+      "step": 8513
+    },
+    {
+      "epoch": 0.6548727020998385,
+      "learning_rate": 0.0007986354855946371,
+      "loss": 1.2901,
+      "step": 8514
+    },
+    {
+      "epoch": 0.6549496192600569,
+      "learning_rate": 0.0007983151054183662,
+      "loss": 1.2948,
+      "step": 8515
+    },
+    {
+      "epoch": 0.6550265364202754,
+      "learning_rate": 0.0007979947662142076,
+      "loss": 1.1017,
+      "step": 8516
+    },
+    {
+      "epoch": 0.6551034535804938,
+      "learning_rate": 0.0007976744680008665,
+      "loss": 1.2741,
+      "step": 8517
+    },
+    {
+      "epoch": 0.6551803707407122,
+      "learning_rate": 0.0007973542107970449,
+      "loss": 1.1714,
+      "step": 8518
+    },
+    {
+      "epoch": 0.6552572879009307,
+      "learning_rate": 0.0007970339946214427,
+      "loss": 1.087,
+      "step": 8519
+    },
+    {
+      "epoch": 0.6553342050611491,
+      "learning_rate": 0.0007967138194927594,
+      "loss": 1.3374,
+      "step": 8520
+    },
+    {
+      "epoch": 0.6554111222213675,
+      "learning_rate": 0.0007963936854296884,
+      "loss": 0.9367,
+      "step": 8521
+    },
+    {
+      "epoch": 0.6554880393815861,
+      "learning_rate": 0.0007960735924509229,
+      "loss": 1.2661,
+      "step": 8522
+    },
+    {
+      "epoch": 0.6555649565418045,
+      "learning_rate": 0.0007957535405751544,
+      "loss": 1.1345,
+      "step": 8523
+    },
+    {
+      "epoch": 0.6556418737020229,
+      "learning_rate": 0.0007954335298210708,
+      "loss": 1.5253,
+      "step": 8524
+    },
+    {
+      "epoch": 0.6557187908622414,
+      "learning_rate": 0.0007951135602073577,
+      "loss": 0.8936,
+      "step": 8525
+    },
+    {
+      "epoch": 0.6557957080224598,
+      "learning_rate": 0.0007947936317526982,
+      "loss": 0.944,
+      "step": 8526
+    },
+    {
+      "epoch": 0.6558726251826783,
+      "learning_rate": 0.000794473744475773,
+      "loss": 1.3554,
+      "step": 8527
+    },
+    {
+      "epoch": 0.6559495423428967,
+      "learning_rate": 0.0007941538983952617,
+      "loss": 1.6405,
+      "step": 8528
+    },
+    {
+      "epoch": 0.6560264595031151,
+      "learning_rate": 0.0007938340935298407,
+      "loss": 1.3003,
+      "step": 8529
+    },
+    {
+      "epoch": 0.6561033766633336,
+      "learning_rate": 0.0007935143298981816,
+      "loss": 1.0776,
+      "step": 8530
+    },
+    {
+      "epoch": 0.656180293823552,
+      "learning_rate": 0.0007931946075189576,
+      "loss": 0.9144,
+      "step": 8531
+    },
+    {
+      "epoch": 0.6562572109837704,
+      "learning_rate": 0.000792874926410837,
+      "loss": 1.6682,
+      "step": 8532
+    },
+    {
+      "epoch": 0.656334128143989,
+      "learning_rate": 0.0007925552865924867,
+      "loss": 1.0407,
+      "step": 8533
+    },
+    {
+      "epoch": 0.6564110453042074,
+      "learning_rate": 0.0007922356880825702,
+      "loss": 1.3451,
+      "step": 8534
+    },
+    {
+      "epoch": 0.6564879624644259,
+      "learning_rate": 0.000791916130899749,
+      "loss": 1.165,
+      "step": 8535
+    },
+    {
+      "epoch": 0.6565648796246443,
+      "learning_rate": 0.0007915966150626836,
+      "loss": 0.8988,
+      "step": 8536
+    },
+    {
+      "epoch": 0.6566417967848627,
+      "learning_rate": 0.0007912771405900299,
+      "loss": 1.065,
+      "step": 8537
+    },
+    {
+      "epoch": 0.6567187139450812,
+      "learning_rate": 0.0007909577075004427,
+      "loss": 1.1627,
+      "step": 8538
+    },
+    {
+      "epoch": 0.6567956311052996,
+      "learning_rate": 0.000790638315812574,
+      "loss": 1.3178,
+      "step": 8539
+    },
+    {
+      "epoch": 0.656872548265518,
+      "learning_rate": 0.000790318965545073,
+      "loss": 0.6202,
+      "step": 8540
+    },
+    {
+      "epoch": 0.6569494654257365,
+      "learning_rate": 0.0007899996567165869,
+      "loss": 1.1995,
+      "step": 8541
+    },
+    {
+      "epoch": 0.6570263825859549,
+      "learning_rate": 0.0007896803893457618,
+      "loss": 1.2792,
+      "step": 8542
+    },
+    {
+      "epoch": 0.6571032997461733,
+      "learning_rate": 0.0007893611634512386,
+      "loss": 1.3113,
+      "step": 8543
+    },
+    {
+      "epoch": 0.6571802169063918,
+      "learning_rate": 0.000789041979051657,
+      "loss": 1.0086,
+      "step": 8544
+    },
+    {
+      "epoch": 0.6572571340666102,
+      "learning_rate": 0.0007887228361656559,
+      "loss": 1.2499,
+      "step": 8545
+    },
+    {
+      "epoch": 0.6573340512268288,
+      "learning_rate": 0.0007884037348118696,
+      "loss": 1.4176,
+      "step": 8546
+    },
+    {
+      "epoch": 0.6574109683870472,
+      "learning_rate": 0.0007880846750089307,
+      "loss": 0.8643,
+      "step": 8547
+    },
+    {
+      "epoch": 0.6574878855472656,
+      "learning_rate": 0.0007877656567754697,
+      "loss": 1.1159,
+      "step": 8548
+    },
+    {
+      "epoch": 0.6575648027074841,
+      "learning_rate": 0.0007874466801301136,
+      "loss": 1.2298,
+      "step": 8549
+    },
+    {
+      "epoch": 0.6576417198677025,
+      "learning_rate": 0.000787127745091489,
+      "loss": 1.4595,
+      "step": 8550
+    },
+    {
+      "epoch": 0.6577186370279209,
+      "learning_rate": 0.0007868088516782184,
+      "loss": 1.1378,
+      "step": 8551
+    },
+    {
+      "epoch": 0.6577955541881394,
+      "learning_rate": 0.0007864899999089223,
+      "loss": 1.0557,
+      "step": 8552
+    },
+    {
+      "epoch": 0.6578724713483578,
+      "learning_rate": 0.0007861711898022185,
+      "loss": 1.337,
+      "step": 8553
+    },
+    {
+      "epoch": 0.6579493885085763,
+      "learning_rate": 0.0007858524213767225,
+      "loss": 1.1447,
+      "step": 8554
+    },
+    {
+      "epoch": 0.6580263056687947,
+      "learning_rate": 0.0007855336946510491,
+      "loss": 1.3924,
+      "step": 8555
+    },
+    {
+      "epoch": 0.6581032228290131,
+      "learning_rate": 0.0007852150096438069,
+      "loss": 1.3481,
+      "step": 8556
+    },
+    {
+      "epoch": 0.6581801399892316,
+      "learning_rate": 0.000784896366373605,
+      "loss": 0.9431,
+      "step": 8557
+    },
+    {
+      "epoch": 0.65825705714945,
+      "learning_rate": 0.00078457776485905,
+      "loss": 1.3165,
+      "step": 8558
+    },
+    {
+      "epoch": 0.6583339743096684,
+      "learning_rate": 0.0007842592051187448,
+      "loss": 1.1005,
+      "step": 8559
+    },
+    {
+      "epoch": 0.658410891469887,
+      "learning_rate": 0.0007839406871712908,
+      "loss": 0.8452,
+      "step": 8560
+    },
+    {
+      "epoch": 0.6584878086301054,
+      "learning_rate": 0.000783622211035286,
+      "loss": 0.9317,
+      "step": 8561
+    },
+    {
+      "epoch": 0.6585647257903238,
+      "learning_rate": 0.0007833037767293271,
+      "loss": 1.0621,
+      "step": 8562
+    },
+    {
+      "epoch": 0.6586416429505423,
+      "learning_rate": 0.000782985384272007,
+      "loss": 1.3057,
+      "step": 8563
+    },
+    {
+      "epoch": 0.6587185601107607,
+      "learning_rate": 0.0007826670336819181,
+      "loss": 0.9818,
+      "step": 8564
+    },
+    {
+      "epoch": 0.6587954772709792,
+      "learning_rate": 0.0007823487249776488,
+      "loss": 1.0354,
+      "step": 8565
+    },
+    {
+      "epoch": 0.6588723944311976,
+      "learning_rate": 0.0007820304581777851,
+      "loss": 0.7924,
+      "step": 8566
+    },
+    {
+      "epoch": 0.658949311591416,
+      "learning_rate": 0.0007817122333009114,
+      "loss": 1.5939,
+      "step": 8567
+    },
+    {
+      "epoch": 0.6590262287516345,
+      "learning_rate": 0.0007813940503656091,
+      "loss": 1.0248,
+      "step": 8568
+    },
+    {
+      "epoch": 0.6591031459118529,
+      "learning_rate": 0.0007810759093904568,
+      "loss": 1.3644,
+      "step": 8569
+    },
+    {
+      "epoch": 0.6591800630720713,
+      "learning_rate": 0.0007807578103940317,
+      "loss": 1.3638,
+      "step": 8570
+    },
+    {
+      "epoch": 0.6592569802322898,
+      "learning_rate": 0.000780439753394907,
+      "loss": 1.1535,
+      "step": 8571
+    },
+    {
+      "epoch": 0.6593338973925083,
+      "learning_rate": 0.0007801217384116553,
+      "loss": 0.7405,
+      "step": 8572
+    },
+    {
+      "epoch": 0.6594108145527268,
+      "learning_rate": 0.0007798037654628458,
+      "loss": 1.2405,
+      "step": 8573
+    },
+    {
+      "epoch": 0.6594877317129452,
+      "learning_rate": 0.0007794858345670448,
+      "loss": 1.1254,
+      "step": 8574
+    },
+    {
+      "epoch": 0.6595646488731636,
+      "learning_rate": 0.0007791679457428171,
+      "loss": 1.2448,
+      "step": 8575
+    },
+    {
+      "epoch": 0.6596415660333821,
+      "learning_rate": 0.0007788500990087234,
+      "loss": 1.2192,
+      "step": 8576
+    },
+    {
+      "epoch": 0.6597184831936005,
+      "learning_rate": 0.0007785322943833248,
+      "loss": 1.0965,
+      "step": 8577
+    },
+    {
+      "epoch": 0.6597954003538189,
+      "learning_rate": 0.0007782145318851779,
+      "loss": 1.0793,
+      "step": 8578
+    },
+    {
+      "epoch": 0.6598723175140374,
+      "learning_rate": 0.0007778968115328356,
+      "loss": 1.3679,
+      "step": 8579
+    },
+    {
+      "epoch": 0.6599492346742558,
+      "learning_rate": 0.0007775791333448516,
+      "loss": 1.194,
+      "step": 8580
+    },
+    {
+      "epoch": 0.6600261518344742,
+      "learning_rate": 0.000777261497339775,
+      "loss": 1.0274,
+      "step": 8581
+    },
+    {
+      "epoch": 0.6601030689946927,
+      "learning_rate": 0.0007769439035361527,
+      "loss": 1.1253,
+      "step": 8582
+    },
+    {
+      "epoch": 0.6601799861549111,
+      "learning_rate": 0.0007766263519525292,
+      "loss": 1.2471,
+      "step": 8583
+    },
+    {
+      "epoch": 0.6602569033151297,
+      "learning_rate": 0.0007763088426074466,
+      "loss": 1.1612,
+      "step": 8584
+    },
+    {
+      "epoch": 0.660333820475348,
+      "learning_rate": 0.0007759913755194455,
+      "loss": 1.2863,
+      "step": 8585
+    },
+    {
+      "epoch": 0.6604107376355665,
+      "learning_rate": 0.0007756739507070622,
+      "loss": 0.919,
+      "step": 8586
+    },
+    {
+      "epoch": 0.660487654795785,
+      "learning_rate": 0.0007753565681888321,
+      "loss": 1.322,
+      "step": 8587
+    },
+    {
+      "epoch": 0.6605645719560034,
+      "learning_rate": 0.0007750392279832869,
+      "loss": 0.8747,
+      "step": 8588
+    },
+    {
+      "epoch": 0.6606414891162218,
+      "learning_rate": 0.0007747219301089571,
+      "loss": 1.2222,
+      "step": 8589
+    },
+    {
+      "epoch": 0.6607184062764403,
+      "learning_rate": 0.000774404674584369,
+      "loss": 1.0856,
+      "step": 8590
+    },
+    {
+      "epoch": 0.6607953234366587,
+      "learning_rate": 0.000774087461428049,
+      "loss": 1.197,
+      "step": 8591
+    },
+    {
+      "epoch": 0.6608722405968772,
+      "learning_rate": 0.0007737702906585178,
+      "loss": 1.3886,
+      "step": 8592
+    },
+    {
+      "epoch": 0.6609491577570956,
+      "learning_rate": 0.0007734531622942965,
+      "loss": 1.0896,
+      "step": 8593
+    },
+    {
+      "epoch": 0.661026074917314,
+      "learning_rate": 0.0007731360763539024,
+      "loss": 1.1717,
+      "step": 8594
+    },
+    {
+      "epoch": 0.6611029920775325,
+      "learning_rate": 0.0007728190328558504,
+      "loss": 1.1633,
+      "step": 8595
+    },
+    {
+      "epoch": 0.6611799092377509,
+      "learning_rate": 0.0007725020318186529,
+      "loss": 1.4007,
+      "step": 8596
+    },
+    {
+      "epoch": 0.6612568263979693,
+      "learning_rate": 0.0007721850732608199,
+      "loss": 0.8127,
+      "step": 8597
+    },
+    {
+      "epoch": 0.6613337435581879,
+      "learning_rate": 0.0007718681572008583,
+      "loss": 1.1896,
+      "step": 8598
+    },
+    {
+      "epoch": 0.6614106607184063,
+      "learning_rate": 0.0007715512836572744,
+      "loss": 1.0697,
+      "step": 8599
+    },
+    {
+      "epoch": 0.6614875778786247,
+      "learning_rate": 0.0007712344526485711,
+      "loss": 1.4127,
+      "step": 8600
+    },
+    {
+      "epoch": 0.6615644950388432,
+      "learning_rate": 0.0007709176641932462,
+      "loss": 1.2476,
+      "step": 8601
+    },
+    {
+      "epoch": 0.6616414121990616,
+      "learning_rate": 0.0007706009183097993,
+      "loss": 1.0648,
+      "step": 8602
+    },
+    {
+      "epoch": 0.6617183293592801,
+      "learning_rate": 0.0007702842150167245,
+      "loss": 1.2622,
+      "step": 8603
+    },
+    {
+      "epoch": 0.6617952465194985,
+      "learning_rate": 0.000769967554332516,
+      "loss": 0.8255,
+      "step": 8604
+    },
+    {
+      "epoch": 0.6618721636797169,
+      "learning_rate": 0.0007696509362756622,
+      "loss": 1.1466,
+      "step": 8605
+    },
+    {
+      "epoch": 0.6619490808399354,
+      "learning_rate": 0.0007693343608646508,
+      "loss": 1.3083,
+      "step": 8606
+    },
+    {
+      "epoch": 0.6620259980001538,
+      "learning_rate": 0.0007690178281179682,
+      "loss": 1.3135,
+      "step": 8607
+    },
+    {
+      "epoch": 0.6621029151603722,
+      "learning_rate": 0.0007687013380540961,
+      "loss": 1.053,
+      "step": 8608
+    },
+    {
+      "epoch": 0.6621798323205907,
+      "learning_rate": 0.0007683848906915153,
+      "loss": 0.8834,
+      "step": 8609
+    },
+    {
+      "epoch": 0.6622567494808091,
+      "learning_rate": 0.0007680684860487034,
+      "loss": 1.067,
+      "step": 8610
+    },
+    {
+      "epoch": 0.6623336666410277,
+      "learning_rate": 0.0007677521241441344,
+      "loss": 1.2828,
+      "step": 8611
+    },
+    {
+      "epoch": 0.6624105838012461,
+      "learning_rate": 0.000767435804996283,
+      "loss": 1.2955,
+      "step": 8612
+    },
+    {
+      "epoch": 0.6624875009614645,
+      "learning_rate": 0.0007671195286236186,
+      "loss": 1.196,
+      "step": 8613
+    },
+    {
+      "epoch": 0.662564418121683,
+      "learning_rate": 0.0007668032950446077,
+      "loss": 1.2471,
+      "step": 8614
+    },
+    {
+      "epoch": 0.6626413352819014,
+      "learning_rate": 0.0007664871042777172,
+      "loss": 1.2154,
+      "step": 8615
+    },
+    {
+      "epoch": 0.6627182524421198,
+      "learning_rate": 0.0007661709563414091,
+      "loss": 1.3028,
+      "step": 8616
+    },
+    {
+      "epoch": 0.6627951696023383,
+      "learning_rate": 0.0007658548512541435,
+      "loss": 1.4589,
+      "step": 8617
+    },
+    {
+      "epoch": 0.6628720867625567,
+      "learning_rate": 0.0007655387890343785,
+      "loss": 1.0832,
+      "step": 8618
+    },
+    {
+      "epoch": 0.6629490039227752,
+      "learning_rate": 0.000765222769700568,
+      "loss": 1.2397,
+      "step": 8619
+    },
+    {
+      "epoch": 0.6630259210829936,
+      "learning_rate": 0.0007649067932711668,
+      "loss": 1.276,
+      "step": 8620
+    },
+    {
+      "epoch": 0.663102838243212,
+      "learning_rate": 0.0007645908597646239,
+      "loss": 1.0063,
+      "step": 8621
+    },
+    {
+      "epoch": 0.6631797554034305,
+      "learning_rate": 0.0007642749691993871,
+      "loss": 0.8444,
+      "step": 8622
+    },
+    {
+      "epoch": 0.663256672563649,
+      "learning_rate": 0.0007639591215939015,
+      "loss": 1.1295,
+      "step": 8623
+    },
+    {
+      "epoch": 0.6633335897238674,
+      "learning_rate": 0.00076364331696661,
+      "loss": 0.9346,
+      "step": 8624
+    },
+    {
+      "epoch": 0.6634105068840859,
+      "learning_rate": 0.0007633275553359519,
+      "loss": 1.3005,
+      "step": 8625
+    },
+    {
+      "epoch": 0.6634874240443043,
+      "learning_rate": 0.000763011836720367,
+      "loss": 1.2666,
+      "step": 8626
+    },
+    {
+      "epoch": 0.6635643412045227,
+      "learning_rate": 0.0007626961611382882,
+      "loss": 1.3495,
+      "step": 8627
+    },
+    {
+      "epoch": 0.6636412583647412,
+      "learning_rate": 0.0007623805286081485,
+      "loss": 1.2139,
+      "step": 8628
+    },
+    {
+      "epoch": 0.6637181755249596,
+      "learning_rate": 0.0007620649391483791,
+      "loss": 1.0073,
+      "step": 8629
+    },
+    {
+      "epoch": 0.6637950926851781,
+      "learning_rate": 0.0007617493927774069,
+      "loss": 1.8874,
+      "step": 8630
+    },
+    {
+      "epoch": 0.6638720098453965,
+      "learning_rate": 0.0007614338895136569,
+      "loss": 1.246,
+      "step": 8631
+    },
+    {
+      "epoch": 0.6639489270056149,
+      "learning_rate": 0.0007611184293755519,
+      "loss": 1.1016,
+      "step": 8632
+    },
+    {
+      "epoch": 0.6640258441658334,
+      "learning_rate": 0.0007608030123815114,
+      "loss": 0.963,
+      "step": 8633
+    },
+    {
+      "epoch": 0.6641027613260518,
+      "learning_rate": 0.0007604876385499538,
+      "loss": 0.8887,
+      "step": 8634
+    },
+    {
+      "epoch": 0.6641796784862702,
+      "learning_rate": 0.0007601723078992943,
+      "loss": 1.2444,
+      "step": 8635
+    },
+    {
+      "epoch": 0.6642565956464888,
+      "learning_rate": 0.0007598570204479435,
+      "loss": 1.2917,
+      "step": 8636
+    },
+    {
+      "epoch": 0.6643335128067072,
+      "learning_rate": 0.0007595417762143135,
+      "loss": 1.231,
+      "step": 8637
+    },
+    {
+      "epoch": 0.6644104299669257,
+      "learning_rate": 0.0007592265752168102,
+      "loss": 1.2965,
+      "step": 8638
+    },
+    {
+      "epoch": 0.6644873471271441,
+      "learning_rate": 0.0007589114174738402,
+      "loss": 0.9876,
+      "step": 8639
+    },
+    {
+      "epoch": 0.6645642642873625,
+      "learning_rate": 0.0007585963030038046,
+      "loss": 1.1763,
+      "step": 8640
+    },
+    {
+      "epoch": 0.664641181447581,
+      "learning_rate": 0.0007582812318251026,
+      "loss": 1.6656,
+      "step": 8641
+    },
+    {
+      "epoch": 0.6647180986077994,
+      "learning_rate": 0.0007579662039561332,
+      "loss": 1.1732,
+      "step": 8642
+    },
+    {
+      "epoch": 0.6647950157680178,
+      "learning_rate": 0.0007576512194152905,
+      "loss": 1.2624,
+      "step": 8643
+    },
+    {
+      "epoch": 0.6648719329282363,
+      "learning_rate": 0.0007573362782209668,
+      "loss": 0.695,
+      "step": 8644
+    },
+    {
+      "epoch": 0.6649488500884547,
+      "learning_rate": 0.0007570213803915518,
+      "loss": 1.1702,
+      "step": 8645
+    },
+    {
+      "epoch": 0.6650257672486731,
+      "learning_rate": 0.0007567065259454325,
+      "loss": 1.5539,
+      "step": 8646
+    },
+    {
+      "epoch": 0.6651026844088916,
+      "learning_rate": 0.0007563917149009935,
+      "loss": 0.9346,
+      "step": 8647
+    },
+    {
+      "epoch": 0.66517960156911,
+      "learning_rate": 0.0007560769472766181,
+      "loss": 1.0809,
+      "step": 8648
+    },
+    {
+      "epoch": 0.6652565187293286,
+      "learning_rate": 0.0007557622230906839,
+      "loss": 0.9848,
+      "step": 8649
+    },
+    {
+      "epoch": 0.665333435889547,
+      "learning_rate": 0.00075544754236157,
+      "loss": 1.0547,
+      "step": 8650
+    },
+    {
+      "epoch": 0.6654103530497654,
+      "learning_rate": 0.0007551329051076499,
+      "loss": 0.9818,
+      "step": 8651
+    },
+    {
+      "epoch": 0.6654872702099839,
+      "learning_rate": 0.0007548183113472958,
+      "loss": 1.2576,
+      "step": 8652
+    },
+    {
+      "epoch": 0.6655641873702023,
+      "learning_rate": 0.0007545037610988772,
+      "loss": 1.4696,
+      "step": 8653
+    },
+    {
+      "epoch": 0.6656411045304207,
+      "learning_rate": 0.0007541892543807606,
+      "loss": 1.1655,
+      "step": 8654
+    },
+    {
+      "epoch": 0.6657180216906392,
+      "learning_rate": 0.0007538747912113104,
+      "loss": 1.3117,
+      "step": 8655
+    },
+    {
+      "epoch": 0.6657949388508576,
+      "learning_rate": 0.0007535603716088892,
+      "loss": 1.2257,
+      "step": 8656
+    },
+    {
+      "epoch": 0.6658718560110761,
+      "learning_rate": 0.000753245995591856,
+      "loss": 0.9942,
+      "step": 8657
+    },
+    {
+      "epoch": 0.6659487731712945,
+      "learning_rate": 0.000752931663178567,
+      "loss": 1.17,
+      "step": 8658
+    },
+    {
+      "epoch": 0.6660256903315129,
+      "learning_rate": 0.000752617374387377,
+      "loss": 1.0842,
+      "step": 8659
+    },
+    {
+      "epoch": 0.6661026074917314,
+      "learning_rate": 0.0007523031292366369,
+      "loss": 1.1858,
+      "step": 8660
+    },
+    {
+      "epoch": 0.6661795246519499,
+      "learning_rate": 0.0007519889277446973,
+      "loss": 1.1992,
+      "step": 8661
+    },
+    {
+      "epoch": 0.6662564418121683,
+      "learning_rate": 0.0007516747699299034,
+      "loss": 1.3494,
+      "step": 8662
+    },
+    {
+      "epoch": 0.6663333589723868,
+      "learning_rate": 0.0007513606558105987,
+      "loss": 1.2976,
+      "step": 8663
+    },
+    {
+      "epoch": 0.6664102761326052,
+      "learning_rate": 0.0007510465854051264,
+      "loss": 0.881,
+      "step": 8664
+    },
+    {
+      "epoch": 0.6664871932928236,
+      "learning_rate": 0.0007507325587318244,
+      "loss": 1.3611,
+      "step": 8665
+    },
+    {
+      "epoch": 0.6665641104530421,
+      "learning_rate": 0.0007504185758090291,
+      "loss": 1.1902,
+      "step": 8666
+    },
+    {
+      "epoch": 0.6666410276132605,
+      "learning_rate": 0.0007501046366550746,
+      "loss": 1.2523,
+      "step": 8667
+    },
+    {
+      "epoch": 0.666717944773479,
+      "learning_rate": 0.0007497907412882909,
+      "loss": 1.0071,
+      "step": 8668
+    },
+    {
+      "epoch": 0.6667948619336974,
+      "learning_rate": 0.0007494768897270084,
+      "loss": 1.1752,
+      "step": 8669
+    },
+    {
+      "epoch": 0.6668717790939158,
+      "learning_rate": 0.0007491630819895532,
+      "loss": 1.0591,
+      "step": 8670
+    },
+    {
+      "epoch": 0.6669486962541343,
+      "learning_rate": 0.0007488493180942469,
+      "loss": 0.9658,
+      "step": 8671
+    },
+    {
+      "epoch": 0.6670256134143527,
+      "learning_rate": 0.0007485355980594124,
+      "loss": 1.0712,
+      "step": 8672
+    },
+    {
+      "epoch": 0.6671025305745711,
+      "learning_rate": 0.0007482219219033674,
+      "loss": 1.1305,
+      "step": 8673
+    },
+    {
+      "epoch": 0.6671794477347897,
+      "learning_rate": 0.0007479082896444278,
+      "loss": 1.2732,
+      "step": 8674
+    },
+    {
+      "epoch": 0.6672563648950081,
+      "learning_rate": 0.0007475947013009071,
+      "loss": 1.1556,
+      "step": 8675
+    },
+    {
+      "epoch": 0.6673332820552266,
+      "learning_rate": 0.0007472811568911155,
+      "loss": 1.1323,
+      "step": 8676
+    },
+    {
+      "epoch": 0.667410199215445,
+      "learning_rate": 0.000746967656433362,
+      "loss": 1.2265,
+      "step": 8677
+    },
+    {
+      "epoch": 0.6674871163756634,
+      "learning_rate": 0.0007466541999459517,
+      "loss": 1.0317,
+      "step": 8678
+    },
+    {
+      "epoch": 0.6675640335358819,
+      "learning_rate": 0.0007463407874471879,
+      "loss": 0.9159,
+      "step": 8679
+    },
+    {
+      "epoch": 0.6676409506961003,
+      "learning_rate": 0.000746027418955371,
+      "loss": 1.0863,
+      "step": 8680
+    },
+    {
+      "epoch": 0.6677178678563187,
+      "learning_rate": 0.0007457140944887989,
+      "loss": 0.951,
+      "step": 8681
+    },
+    {
+      "epoch": 0.6677947850165372,
+      "learning_rate": 0.0007454008140657663,
+      "loss": 0.9111,
+      "step": 8682
+    },
+    {
+      "epoch": 0.6678717021767556,
+      "learning_rate": 0.0007450875777045677,
+      "loss": 0.9591,
+      "step": 8683
+    },
+    {
+      "epoch": 0.667948619336974,
+      "learning_rate": 0.0007447743854234918,
+      "loss": 1.2639,
+      "step": 8684
+    },
+    {
+      "epoch": 0.6680255364971925,
+      "learning_rate": 0.0007444612372408258,
+      "loss": 0.9367,
+      "step": 8685
+    },
+    {
+      "epoch": 0.6681024536574109,
+      "learning_rate": 0.0007441481331748562,
+      "loss": 1.1169,
+      "step": 8686
+    },
+    {
+      "epoch": 0.6681793708176295,
+      "learning_rate": 0.000743835073243865,
+      "loss": 1.1504,
+      "step": 8687
+    },
+    {
+      "epoch": 0.6682562879778479,
+      "learning_rate": 0.0007435220574661316,
+      "loss": 1.211,
+      "step": 8688
+    },
+    {
+      "epoch": 0.6683332051380663,
+      "learning_rate": 0.0007432090858599341,
+      "loss": 1.1638,
+      "step": 8689
+    },
+    {
+      "epoch": 0.6684101222982848,
+      "learning_rate": 0.0007428961584435459,
+      "loss": 1.0729,
+      "step": 8690
+    },
+    {
+      "epoch": 0.6684870394585032,
+      "learning_rate": 0.0007425832752352407,
+      "loss": 1.3794,
+      "step": 8691
+    },
+    {
+      "epoch": 0.6685639566187216,
+      "learning_rate": 0.000742270436253288,
+      "loss": 1.3861,
+      "step": 8692
+    },
+    {
+      "epoch": 0.6686408737789401,
+      "learning_rate": 0.0007419576415159529,
+      "loss": 1.3238,
+      "step": 8693
+    },
+    {
+      "epoch": 0.6687177909391585,
+      "learning_rate": 0.0007416448910415018,
+      "loss": 1.3932,
+      "step": 8694
+    },
+    {
+      "epoch": 0.668794708099377,
+      "learning_rate": 0.0007413321848481954,
+      "loss": 1.407,
+      "step": 8695
+    },
+    {
+      "epoch": 0.6688716252595954,
+      "learning_rate": 0.0007410195229542945,
+      "loss": 1.0167,
+      "step": 8696
+    },
+    {
+      "epoch": 0.6689485424198138,
+      "learning_rate": 0.0007407069053780539,
+      "loss": 0.8776,
+      "step": 8697
+    },
+    {
+      "epoch": 0.6690254595800323,
+      "learning_rate": 0.000740394332137728,
+      "loss": 1.0069,
+      "step": 8698
+    },
+    {
+      "epoch": 0.6691023767402507,
+      "learning_rate": 0.0007400818032515692,
+      "loss": 0.7575,
+      "step": 8699
+    },
+    {
+      "epoch": 0.6691792939004692,
+      "learning_rate": 0.0007397693187378259,
+      "loss": 1.0696,
+      "step": 8700
+    },
+    {
+      "epoch": 0.6692562110606877,
+      "learning_rate": 0.0007394568786147444,
+      "loss": 1.0674,
+      "step": 8701
+    },
+    {
+      "epoch": 0.6693331282209061,
+      "learning_rate": 0.0007391444829005684,
+      "loss": 1.4395,
+      "step": 8702
+    },
+    {
+      "epoch": 0.6694100453811245,
+      "learning_rate": 0.000738832131613539,
+      "loss": 1.1392,
+      "step": 8703
+    },
+    {
+      "epoch": 0.669486962541343,
+      "learning_rate": 0.0007385198247718941,
+      "loss": 0.9501,
+      "step": 8704
+    },
+    {
+      "epoch": 0.6695638797015614,
+      "learning_rate": 0.0007382075623938712,
+      "loss": 1.457,
+      "step": 8705
+    },
+    {
+      "epoch": 0.6696407968617799,
+      "learning_rate": 0.0007378953444977017,
+      "loss": 1.1341,
+      "step": 8706
+    },
+    {
+      "epoch": 0.6697177140219983,
+      "learning_rate": 0.0007375831711016179,
+      "loss": 1.1324,
+      "step": 8707
+    },
+    {
+      "epoch": 0.6697946311822167,
+      "learning_rate": 0.0007372710422238471,
+      "loss": 1.078,
+      "step": 8708
+    },
+    {
+      "epoch": 0.6698715483424352,
+      "learning_rate": 0.0007369589578826151,
+      "loss": 1.3785,
+      "step": 8709
+    },
+    {
+      "epoch": 0.6699484655026536,
+      "learning_rate": 0.0007366469180961449,
+      "loss": 1.2788,
+      "step": 8710
+    },
+    {
+      "epoch": 0.670025382662872,
+      "learning_rate": 0.0007363349228826565,
+      "loss": 1.4992,
+      "step": 8711
+    },
+    {
+      "epoch": 0.6701022998230906,
+      "learning_rate": 0.0007360229722603675,
+      "loss": 1.3291,
+      "step": 8712
+    },
+    {
+      "epoch": 0.670179216983309,
+      "learning_rate": 0.0007357110662474938,
+      "loss": 1.2314,
+      "step": 8713
+    },
+    {
+      "epoch": 0.6702561341435275,
+      "learning_rate": 0.0007353992048622475,
+      "loss": 1.1453,
+      "step": 8714
+    },
+    {
+      "epoch": 0.6703330513037459,
+      "learning_rate": 0.0007350873881228384,
+      "loss": 1.0049,
+      "step": 8715
+    },
+    {
+      "epoch": 0.6704099684639643,
+      "learning_rate": 0.0007347756160474739,
+      "loss": 1.227,
+      "step": 8716
+    },
+    {
+      "epoch": 0.6704868856241828,
+      "learning_rate": 0.0007344638886543582,
+      "loss": 1.1893,
+      "step": 8717
+    },
+    {
+      "epoch": 0.6705638027844012,
+      "learning_rate": 0.0007341522059616944,
+      "loss": 1.3653,
+      "step": 8718
+    },
+    {
+      "epoch": 0.6706407199446196,
+      "learning_rate": 0.000733840567987682,
+      "loss": 1.4323,
+      "step": 8719
+    },
+    {
+      "epoch": 0.6707176371048381,
+      "learning_rate": 0.0007335289747505161,
+      "loss": 1.0225,
+      "step": 8720
+    },
+    {
+      "epoch": 0.6707945542650565,
+      "learning_rate": 0.0007332174262683926,
+      "loss": 0.6422,
+      "step": 8721
+    },
+    {
+      "epoch": 0.6708714714252749,
+      "learning_rate": 0.000732905922559503,
+      "loss": 0.9006,
+      "step": 8722
+    },
+    {
+      "epoch": 0.6709483885854934,
+      "learning_rate": 0.0007325944636420358,
+      "loss": 1.0171,
+      "step": 8723
+    },
+    {
+      "epoch": 0.6710253057457118,
+      "learning_rate": 0.0007322830495341779,
+      "loss": 0.9103,
+      "step": 8724
+    },
+    {
+      "epoch": 0.6711022229059304,
+      "learning_rate": 0.000731971680254112,
+      "loss": 1.1405,
+      "step": 8725
+    },
+    {
+      "epoch": 0.6711791400661488,
+      "learning_rate": 0.0007316603558200209,
+      "loss": 1.1034,
+      "step": 8726
+    },
+    {
+      "epoch": 0.6712560572263672,
+      "learning_rate": 0.0007313490762500822,
+      "loss": 1.3074,
+      "step": 8727
+    },
+    {
+      "epoch": 0.6713329743865857,
+      "learning_rate": 0.000731037841562472,
+      "loss": 0.9205,
+      "step": 8728
+    },
+    {
+      "epoch": 0.6714098915468041,
+      "learning_rate": 0.0007307266517753638,
+      "loss": 1.4002,
+      "step": 8729
+    },
+    {
+      "epoch": 0.6714868087070225,
+      "learning_rate": 0.0007304155069069279,
+      "loss": 1.4464,
+      "step": 8730
+    },
+    {
+      "epoch": 0.671563725867241,
+      "learning_rate": 0.0007301044069753322,
+      "loss": 1.2686,
+      "step": 8731
+    },
+    {
+      "epoch": 0.6716406430274594,
+      "learning_rate": 0.0007297933519987436,
+      "loss": 0.9774,
+      "step": 8732
+    },
+    {
+      "epoch": 0.6717175601876779,
+      "learning_rate": 0.0007294823419953226,
+      "loss": 1.4144,
+      "step": 8733
+    },
+    {
+      "epoch": 0.6717944773478963,
+      "learning_rate": 0.0007291713769832316,
+      "loss": 0.6688,
+      "step": 8734
+    },
+    {
+      "epoch": 0.6718713945081147,
+      "learning_rate": 0.000728860456980627,
+      "loss": 1.3451,
+      "step": 8735
+    },
+    {
+      "epoch": 0.6719483116683332,
+      "learning_rate": 0.0007285495820056639,
+      "loss": 1.0131,
+      "step": 8736
+    },
+    {
+      "epoch": 0.6720252288285516,
+      "learning_rate": 0.0007282387520764948,
+      "loss": 1.3469,
+      "step": 8737
+    },
+    {
+      "epoch": 0.67210214598877,
+      "learning_rate": 0.0007279279672112693,
+      "loss": 0.9187,
+      "step": 8738
+    },
+    {
+      "epoch": 0.6721790631489886,
+      "learning_rate": 0.0007276172274281337,
+      "loss": 1.411,
+      "step": 8739
+    },
+    {
+      "epoch": 0.672255980309207,
+      "learning_rate": 0.0007273065327452339,
+      "loss": 1.0844,
+      "step": 8740
+    },
+    {
+      "epoch": 0.6723328974694255,
+      "learning_rate": 0.0007269958831807109,
+      "loss": 1.2339,
+      "step": 8741
+    },
+    {
+      "epoch": 0.6724098146296439,
+      "learning_rate": 0.0007266852787527038,
+      "loss": 1.1215,
+      "step": 8742
+    },
+    {
+      "epoch": 0.6724867317898623,
+      "learning_rate": 0.0007263747194793493,
+      "loss": 0.9118,
+      "step": 8743
+    },
+    {
+      "epoch": 0.6725636489500808,
+      "learning_rate": 0.0007260642053787806,
+      "loss": 1.1623,
+      "step": 8744
+    },
+    {
+      "epoch": 0.6726405661102992,
+      "learning_rate": 0.0007257537364691306,
+      "loss": 1.271,
+      "step": 8745
+    },
+    {
+      "epoch": 0.6727174832705176,
+      "learning_rate": 0.0007254433127685263,
+      "loss": 1.3033,
+      "step": 8746
+    },
+    {
+      "epoch": 0.6727944004307361,
+      "learning_rate": 0.0007251329342950935,
+      "loss": 1.141,
+      "step": 8747
+    },
+    {
+      "epoch": 0.6728713175909545,
+      "learning_rate": 0.0007248226010669569,
+      "loss": 1.0358,
+      "step": 8748
+    },
+    {
+      "epoch": 0.6729482347511729,
+      "learning_rate": 0.0007245123131022366,
+      "loss": 1.2178,
+      "step": 8749
+    },
+    {
+      "epoch": 0.6730251519113915,
+      "learning_rate": 0.0007242020704190504,
+      "loss": 1.1747,
+      "step": 8750
+    },
+    {
+      "epoch": 0.6731020690716099,
+      "learning_rate": 0.0007238918730355138,
+      "loss": 1.2216,
+      "step": 8751
+    },
+    {
+      "epoch": 0.6731789862318284,
+      "learning_rate": 0.0007235817209697391,
+      "loss": 1.3866,
+      "step": 8752
+    },
+    {
+      "epoch": 0.6732559033920468,
+      "learning_rate": 0.0007232716142398374,
+      "loss": 1.4212,
+      "step": 8753
+    },
+    {
+      "epoch": 0.6733328205522652,
+      "learning_rate": 0.0007229615528639162,
+      "loss": 1.1867,
+      "step": 8754
+    },
+    {
+      "epoch": 0.6734097377124837,
+      "learning_rate": 0.0007226515368600786,
+      "loss": 1.7131,
+      "step": 8755
+    },
+    {
+      "epoch": 0.6734866548727021,
+      "learning_rate": 0.0007223415662464286,
+      "loss": 0.9497,
+      "step": 8756
+    },
+    {
+      "epoch": 0.6735635720329205,
+      "learning_rate": 0.0007220316410410647,
+      "loss": 1.1333,
+      "step": 8757
+    },
+    {
+      "epoch": 0.673640489193139,
+      "learning_rate": 0.0007217217612620842,
+      "loss": 0.8965,
+      "step": 8758
+    },
+    {
+      "epoch": 0.6737174063533574,
+      "learning_rate": 0.0007214119269275813,
+      "loss": 1.562,
+      "step": 8759
+    },
+    {
+      "epoch": 0.6737943235135759,
+      "learning_rate": 0.0007211021380556467,
+      "loss": 1.2326,
+      "step": 8760
+    },
+    {
+      "epoch": 0.6738712406737943,
+      "learning_rate": 0.0007207923946643707,
+      "loss": 1.2517,
+      "step": 8761
+    },
+    {
+      "epoch": 0.6739481578340127,
+      "learning_rate": 0.0007204826967718386,
+      "loss": 1.0405,
+      "step": 8762
+    },
+    {
+      "epoch": 0.6740250749942313,
+      "learning_rate": 0.0007201730443961345,
+      "loss": 1.126,
+      "step": 8763
+    },
+    {
+      "epoch": 0.6741019921544497,
+      "learning_rate": 0.0007198634375553389,
+      "loss": 1.2274,
+      "step": 8764
+    },
+    {
+      "epoch": 0.6741789093146681,
+      "learning_rate": 0.0007195538762675302,
+      "loss": 0.9536,
+      "step": 8765
+    },
+    {
+      "epoch": 0.6742558264748866,
+      "learning_rate": 0.0007192443605507836,
+      "loss": 0.9043,
+      "step": 8766
+    },
+    {
+      "epoch": 0.674332743635105,
+      "learning_rate": 0.0007189348904231734,
+      "loss": 1.4426,
+      "step": 8767
+    },
+    {
+      "epoch": 0.6744096607953234,
+      "learning_rate": 0.0007186254659027685,
+      "loss": 1.2582,
+      "step": 8768
+    },
+    {
+      "epoch": 0.6744865779555419,
+      "learning_rate": 0.0007183160870076365,
+      "loss": 0.8202,
+      "step": 8769
+    },
+    {
+      "epoch": 0.6745634951157603,
+      "learning_rate": 0.0007180067537558433,
+      "loss": 0.9519,
+      "step": 8770
+    },
+    {
+      "epoch": 0.6746404122759788,
+      "learning_rate": 0.0007176974661654507,
+      "loss": 1.1611,
+      "step": 8771
+    },
+    {
+      "epoch": 0.6747173294361972,
+      "learning_rate": 0.0007173882242545184,
+      "loss": 1.5816,
+      "step": 8772
+    },
+    {
+      "epoch": 0.6747942465964156,
+      "learning_rate": 0.0007170790280411033,
+      "loss": 1.1072,
+      "step": 8773
+    },
+    {
+      "epoch": 0.6748711637566341,
+      "learning_rate": 0.0007167698775432591,
+      "loss": 1.2849,
+      "step": 8774
+    },
+    {
+      "epoch": 0.6749480809168525,
+      "learning_rate": 0.0007164607727790387,
+      "loss": 1.091,
+      "step": 8775
+    },
+    {
+      "epoch": 0.675024998077071,
+      "learning_rate": 0.0007161517137664908,
+      "loss": 1.4821,
+      "step": 8776
+    },
+    {
+      "epoch": 0.6751019152372895,
+      "learning_rate": 0.0007158427005236601,
+      "loss": 0.7069,
+      "step": 8777
+    },
+    {
+      "epoch": 0.6751788323975079,
+      "learning_rate": 0.0007155337330685919,
+      "loss": 1.1757,
+      "step": 8778
+    },
+    {
+      "epoch": 0.6752557495577264,
+      "learning_rate": 0.0007152248114193261,
+      "loss": 0.9925,
+      "step": 8779
+    },
+    {
+      "epoch": 0.6753326667179448,
+      "learning_rate": 0.0007149159355939025,
+      "loss": 0.915,
+      "step": 8780
+    },
+    {
+      "epoch": 0.6754095838781632,
+      "learning_rate": 0.000714607105610355,
+      "loss": 1.007,
+      "step": 8781
+    },
+    {
+      "epoch": 0.6754865010383817,
+      "learning_rate": 0.0007142983214867167,
+      "loss": 1.2613,
+      "step": 8782
+    },
+    {
+      "epoch": 0.6755634181986001,
+      "learning_rate": 0.0007139895832410188,
+      "loss": 1.2682,
+      "step": 8783
+    },
+    {
+      "epoch": 0.6756403353588185,
+      "learning_rate": 0.0007136808908912881,
+      "loss": 1.1472,
+      "step": 8784
+    },
+    {
+      "epoch": 0.675717252519037,
+      "learning_rate": 0.00071337224445555,
+      "loss": 1.0913,
+      "step": 8785
+    },
+    {
+      "epoch": 0.6757941696792554,
+      "learning_rate": 0.0007130636439518259,
+      "loss": 1.4439,
+      "step": 8786
+    },
+    {
+      "epoch": 0.6758710868394738,
+      "learning_rate": 0.0007127550893981361,
+      "loss": 1.1878,
+      "step": 8787
+    },
+    {
+      "epoch": 0.6759480039996923,
+      "learning_rate": 0.0007124465808124963,
+      "loss": 1.1368,
+      "step": 8788
+    },
+    {
+      "epoch": 0.6760249211599108,
+      "learning_rate": 0.0007121381182129228,
+      "loss": 0.6754,
+      "step": 8789
+    },
+    {
+      "epoch": 0.6761018383201293,
+      "learning_rate": 0.0007118297016174244,
+      "loss": 1.1726,
+      "step": 8790
+    },
+    {
+      "epoch": 0.6761787554803477,
+      "learning_rate": 0.0007115213310440118,
+      "loss": 1.3301,
+      "step": 8791
+    },
+    {
+      "epoch": 0.6762556726405661,
+      "learning_rate": 0.0007112130065106901,
+      "loss": 1.1458,
+      "step": 8792
+    },
+    {
+      "epoch": 0.6763325898007846,
+      "learning_rate": 0.0007109047280354633,
+      "loss": 0.8939,
+      "step": 8793
+    },
+    {
+      "epoch": 0.676409506961003,
+      "learning_rate": 0.0007105964956363318,
+      "loss": 0.815,
+      "step": 8794
+    },
+    {
+      "epoch": 0.6764864241212214,
+      "learning_rate": 0.0007102883093312934,
+      "loss": 0.9993,
+      "step": 8795
+    },
+    {
+      "epoch": 0.6765633412814399,
+      "learning_rate": 0.0007099801691383433,
+      "loss": 1.1996,
+      "step": 8796
+    },
+    {
+      "epoch": 0.6766402584416583,
+      "learning_rate": 0.0007096720750754749,
+      "loss": 1.2154,
+      "step": 8797
+    },
+    {
+      "epoch": 0.6767171756018768,
+      "learning_rate": 0.0007093640271606777,
+      "loss": 1.2682,
+      "step": 8798
+    },
+    {
+      "epoch": 0.6767940927620952,
+      "learning_rate": 0.000709056025411939,
+      "loss": 1.1586,
+      "step": 8799
+    },
+    {
+      "epoch": 0.6768710099223136,
+      "learning_rate": 0.0007087480698472433,
+      "loss": 1.3004,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6769479270825322,
+      "learning_rate": 0.0007084401604845717,
+      "loss": 1.5114,
+      "step": 8801
+    },
+    {
+      "epoch": 0.6770248442427506,
+      "learning_rate": 0.0007081322973419054,
+      "loss": 1.2009,
+      "step": 8802
+    },
+    {
+      "epoch": 0.677101761402969,
+      "learning_rate": 0.0007078244804372187,
+      "loss": 1.0807,
+      "step": 8803
+    },
+    {
+      "epoch": 0.6771786785631875,
+      "learning_rate": 0.0007075167097884857,
+      "loss": 0.7348,
+      "step": 8804
+    },
+    {
+      "epoch": 0.6772555957234059,
+      "learning_rate": 0.0007072089854136787,
+      "loss": 0.7291,
+      "step": 8805
+    },
+    {
+      "epoch": 0.6773325128836243,
+      "learning_rate": 0.0007069013073307651,
+      "loss": 1.2954,
+      "step": 8806
+    },
+    {
+      "epoch": 0.6774094300438428,
+      "learning_rate": 0.0007065936755577107,
+      "loss": 0.7671,
+      "step": 8807
+    },
+    {
+      "epoch": 0.6774863472040612,
+      "learning_rate": 0.0007062860901124785,
+      "loss": 1.0619,
+      "step": 8808
+    },
+    {
+      "epoch": 0.6775632643642797,
+      "learning_rate": 0.0007059785510130282,
+      "loss": 1.1243,
+      "step": 8809
+    },
+    {
+      "epoch": 0.6776401815244981,
+      "learning_rate": 0.0007056710582773184,
+      "loss": 1.1076,
+      "step": 8810
+    },
+    {
+      "epoch": 0.6777170986847165,
+      "learning_rate": 0.0007053636119233037,
+      "loss": 1.1666,
+      "step": 8811
+    },
+    {
+      "epoch": 0.677794015844935,
+      "learning_rate": 0.0007050562119689347,
+      "loss": 1.1684,
+      "step": 8812
+    },
+    {
+      "epoch": 0.6778709330051534,
+      "learning_rate": 0.0007047488584321628,
+      "loss": 1.2276,
+      "step": 8813
+    },
+    {
+      "epoch": 0.6779478501653718,
+      "learning_rate": 0.0007044415513309335,
+      "loss": 1.07,
+      "step": 8814
+    },
+    {
+      "epoch": 0.6780247673255904,
+      "learning_rate": 0.0007041342906831913,
+      "loss": 1.4927,
+      "step": 8815
+    },
+    {
+      "epoch": 0.6781016844858088,
+      "learning_rate": 0.0007038270765068774,
+      "loss": 1.3837,
+      "step": 8816
+    },
+    {
+      "epoch": 0.6781786016460273,
+      "learning_rate": 0.0007035199088199296,
+      "loss": 1.326,
+      "step": 8817
+    },
+    {
+      "epoch": 0.6782555188062457,
+      "learning_rate": 0.000703212787640285,
+      "loss": 1.3295,
+      "step": 8818
+    },
+    {
+      "epoch": 0.6783324359664641,
+      "learning_rate": 0.000702905712985876,
+      "loss": 1.4315,
+      "step": 8819
+    },
+    {
+      "epoch": 0.6784093531266826,
+      "learning_rate": 0.0007025986848746333,
+      "loss": 1.2604,
+      "step": 8820
+    },
+    {
+      "epoch": 0.678486270286901,
+      "learning_rate": 0.0007022917033244843,
+      "loss": 1.132,
+      "step": 8821
+    },
+    {
+      "epoch": 0.6785631874471194,
+      "learning_rate": 0.0007019847683533541,
+      "loss": 1.0873,
+      "step": 8822
+    },
+    {
+      "epoch": 0.6786401046073379,
+      "learning_rate": 0.0007016778799791643,
+      "loss": 1.1496,
+      "step": 8823
+    },
+    {
+      "epoch": 0.6787170217675563,
+      "learning_rate": 0.0007013710382198365,
+      "loss": 0.6258,
+      "step": 8824
+    },
+    {
+      "epoch": 0.6787939389277747,
+      "learning_rate": 0.0007010642430932852,
+      "loss": 0.93,
+      "step": 8825
+    },
+    {
+      "epoch": 0.6788708560879932,
+      "learning_rate": 0.0007007574946174249,
+      "loss": 1.0889,
+      "step": 8826
+    },
+    {
+      "epoch": 0.6789477732482117,
+      "learning_rate": 0.0007004507928101681,
+      "loss": 1.2397,
+      "step": 8827
+    },
+    {
+      "epoch": 0.6790246904084302,
+      "learning_rate": 0.0007001441376894229,
+      "loss": 1.3494,
+      "step": 8828
+    },
+    {
+      "epoch": 0.6791016075686486,
+      "learning_rate": 0.0006998375292730949,
+      "loss": 1.0884,
+      "step": 8829
+    },
+    {
+      "epoch": 0.679178524728867,
+      "learning_rate": 0.0006995309675790875,
+      "loss": 1.1061,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6792554418890855,
+      "learning_rate": 0.0006992244526253005,
+      "loss": 1.322,
+      "step": 8831
+    },
+    {
+      "epoch": 0.6793323590493039,
+      "learning_rate": 0.0006989179844296328,
+      "loss": 1.5245,
+      "step": 8832
+    },
+    {
+      "epoch": 0.6794092762095223,
+      "learning_rate": 0.0006986115630099797,
+      "loss": 1.1052,
+      "step": 8833
+    },
+    {
+      "epoch": 0.6794861933697408,
+      "learning_rate": 0.0006983051883842313,
+      "loss": 0.8898,
+      "step": 8834
+    },
+    {
+      "epoch": 0.6795631105299592,
+      "learning_rate": 0.000697998860570279,
+      "loss": 1.2782,
+      "step": 8835
+    },
+    {
+      "epoch": 0.6796400276901777,
+      "learning_rate": 0.0006976925795860085,
+      "loss": 1.2881,
+      "step": 8836
+    },
+    {
+      "epoch": 0.6797169448503961,
+      "learning_rate": 0.0006973863454493055,
+      "loss": 0.9359,
+      "step": 8837
+    },
+    {
+      "epoch": 0.6797938620106145,
+      "learning_rate": 0.0006970801581780499,
+      "loss": 0.9424,
+      "step": 8838
+    },
+    {
+      "epoch": 0.679870779170833,
+      "learning_rate": 0.0006967740177901198,
+      "loss": 0.7891,
+      "step": 8839
+    },
+    {
+      "epoch": 0.6799476963310515,
+      "learning_rate": 0.0006964679243033926,
+      "loss": 1.3828,
+      "step": 8840
+    },
+    {
+      "epoch": 0.6800246134912699,
+      "learning_rate": 0.0006961618777357408,
+      "loss": 0.9813,
+      "step": 8841
+    },
+    {
+      "epoch": 0.6801015306514884,
+      "learning_rate": 0.0006958558781050348,
+      "loss": 1.0776,
+      "step": 8842
+    },
+    {
+      "epoch": 0.6801784478117068,
+      "learning_rate": 0.0006955499254291419,
+      "loss": 1.3601,
+      "step": 8843
+    },
+    {
+      "epoch": 0.6802553649719253,
+      "learning_rate": 0.0006952440197259266,
+      "loss": 1.0327,
+      "step": 8844
+    },
+    {
+      "epoch": 0.6803322821321437,
+      "learning_rate": 0.0006949381610132525,
+      "loss": 1.1075,
+      "step": 8845
+    },
+    {
+      "epoch": 0.6804091992923621,
+      "learning_rate": 0.0006946323493089785,
+      "loss": 1.1752,
+      "step": 8846
+    },
+    {
+      "epoch": 0.6804861164525806,
+      "learning_rate": 0.0006943265846309601,
+      "loss": 1.2647,
+      "step": 8847
+    },
+    {
+      "epoch": 0.680563033612799,
+      "learning_rate": 0.0006940208669970524,
+      "loss": 1.1476,
+      "step": 8848
+    },
+    {
+      "epoch": 0.6806399507730174,
+      "learning_rate": 0.0006937151964251065,
+      "loss": 1.2381,
+      "step": 8849
+    },
+    {
+      "epoch": 0.6807168679332359,
+      "learning_rate": 0.0006934095729329705,
+      "loss": 1.0425,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6807937850934543,
+      "learning_rate": 0.0006931039965384899,
+      "loss": 1.1445,
+      "step": 8851
+    },
+    {
+      "epoch": 0.6808707022536727,
+      "learning_rate": 0.0006927984672595081,
+      "loss": 1.3931,
+      "step": 8852
+    },
+    {
+      "epoch": 0.6809476194138913,
+      "learning_rate": 0.0006924929851138644,
+      "loss": 1.1622,
+      "step": 8853
+    },
+    {
+      "epoch": 0.6810245365741097,
+      "learning_rate": 0.0006921875501193974,
+      "loss": 0.9951,
+      "step": 8854
+    },
+    {
+      "epoch": 0.6811014537343282,
+      "learning_rate": 0.0006918821622939413,
+      "loss": 1.3899,
+      "step": 8855
+    },
+    {
+      "epoch": 0.6811783708945466,
+      "learning_rate": 0.0006915768216553278,
+      "loss": 1.1107,
+      "step": 8856
+    },
+    {
+      "epoch": 0.681255288054765,
+      "learning_rate": 0.0006912715282213866,
+      "loss": 1.1527,
+      "step": 8857
+    },
+    {
+      "epoch": 0.6813322052149835,
+      "learning_rate": 0.0006909662820099428,
+      "loss": 1.335,
+      "step": 8858
+    },
+    {
+      "epoch": 0.6814091223752019,
+      "learning_rate": 0.0006906610830388221,
+      "loss": 1.2401,
+      "step": 8859
+    },
+    {
+      "epoch": 0.6814860395354203,
+      "learning_rate": 0.0006903559313258437,
+      "loss": 0.9982,
+      "step": 8860
+    },
+    {
+      "epoch": 0.6815629566956388,
+      "learning_rate": 0.0006900508268888257,
+      "loss": 1.1287,
+      "step": 8861
+    },
+    {
+      "epoch": 0.6816398738558572,
+      "learning_rate": 0.0006897457697455845,
+      "loss": 1.123,
+      "step": 8862
+    },
+    {
+      "epoch": 0.6817167910160757,
+      "learning_rate": 0.0006894407599139324,
+      "loss": 1.1541,
+      "step": 8863
+    },
+    {
+      "epoch": 0.6817937081762941,
+      "learning_rate": 0.000689135797411679,
+      "loss": 0.884,
+      "step": 8864
+    },
+    {
+      "epoch": 0.6818706253365125,
+      "learning_rate": 0.0006888308822566316,
+      "loss": 1.2041,
+      "step": 8865
+    },
+    {
+      "epoch": 0.6819475424967311,
+      "learning_rate": 0.0006885260144665936,
+      "loss": 1.3779,
+      "step": 8866
+    },
+    {
+      "epoch": 0.6820244596569495,
+      "learning_rate": 0.0006882211940593678,
+      "loss": 1.1393,
+      "step": 8867
+    },
+    {
+      "epoch": 0.6821013768171679,
+      "learning_rate": 0.0006879164210527528,
+      "loss": 0.8734,
+      "step": 8868
+    },
+    {
+      "epoch": 0.6821782939773864,
+      "learning_rate": 0.000687611695464544,
+      "loss": 1.1687,
+      "step": 8869
+    },
+    {
+      "epoch": 0.6822552111376048,
+      "learning_rate": 0.0006873070173125352,
+      "loss": 1.2036,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6823321282978232,
+      "learning_rate": 0.0006870023866145165,
+      "loss": 1.1798,
+      "step": 8871
+    },
+    {
+      "epoch": 0.6824090454580417,
+      "learning_rate": 0.0006866978033882756,
+      "loss": 1.2519,
+      "step": 8872
+    },
+    {
+      "epoch": 0.6824859626182601,
+      "learning_rate": 0.0006863932676515978,
+      "loss": 1.0979,
+      "step": 8873
+    },
+    {
+      "epoch": 0.6825628797784786,
+      "learning_rate": 0.0006860887794222644,
+      "loss": 1.2568,
+      "step": 8874
+    },
+    {
+      "epoch": 0.682639796938697,
+      "learning_rate": 0.0006857843387180559,
+      "loss": 1.4729,
+      "step": 8875
+    },
+    {
+      "epoch": 0.6827167140989154,
+      "learning_rate": 0.0006854799455567484,
+      "loss": 1.1827,
+      "step": 8876
+    },
+    {
+      "epoch": 0.682793631259134,
+      "learning_rate": 0.0006851755999561159,
+      "loss": 1.2076,
+      "step": 8877
+    },
+    {
+      "epoch": 0.6828705484193524,
+      "learning_rate": 0.0006848713019339291,
+      "loss": 0.7952,
+      "step": 8878
+    },
+    {
+      "epoch": 0.6829474655795708,
+      "learning_rate": 0.0006845670515079567,
+      "loss": 1.3458,
+      "step": 8879
+    },
+    {
+      "epoch": 0.6830243827397893,
+      "learning_rate": 0.0006842628486959633,
+      "loss": 0.9705,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6831012999000077,
+      "learning_rate": 0.000683958693515713,
+      "loss": 0.8039,
+      "step": 8881
+    },
+    {
+      "epoch": 0.6831782170602262,
+      "learning_rate": 0.000683654585984965,
+      "loss": 1.0404,
+      "step": 8882
+    },
+    {
+      "epoch": 0.6832551342204446,
+      "learning_rate": 0.0006833505261214765,
+      "loss": 1.0742,
+      "step": 8883
+    },
+    {
+      "epoch": 0.683332051380663,
+      "learning_rate": 0.0006830465139430016,
+      "loss": 1.5145,
+      "step": 8884
+    },
+    {
+      "epoch": 0.6834089685408815,
+      "learning_rate": 0.0006827425494672925,
+      "loss": 1.3263,
+      "step": 8885
+    },
+    {
+      "epoch": 0.6834858857010999,
+      "learning_rate": 0.0006824386327120975,
+      "loss": 1.1496,
+      "step": 8886
+    },
+    {
+      "epoch": 0.6835628028613183,
+      "learning_rate": 0.0006821347636951627,
+      "loss": 0.8693,
+      "step": 8887
+    },
+    {
+      "epoch": 0.6836397200215368,
+      "learning_rate": 0.000681830942434231,
+      "loss": 1.0798,
+      "step": 8888
+    },
+    {
+      "epoch": 0.6837166371817552,
+      "learning_rate": 0.0006815271689470437,
+      "loss": 1.3729,
+      "step": 8889
+    },
+    {
+      "epoch": 0.6837935543419736,
+      "learning_rate": 0.0006812234432513381,
+      "loss": 0.9765,
+      "step": 8890
+    },
+    {
+      "epoch": 0.6838704715021922,
+      "learning_rate": 0.000680919765364849,
+      "loss": 1.0981,
+      "step": 8891
+    },
+    {
+      "epoch": 0.6839473886624106,
+      "learning_rate": 0.0006806161353053084,
+      "loss": 1.072,
+      "step": 8892
+    },
+    {
+      "epoch": 0.6840243058226291,
+      "learning_rate": 0.0006803125530904448,
+      "loss": 1.1629,
+      "step": 8893
+    },
+    {
+      "epoch": 0.6841012229828475,
+      "learning_rate": 0.0006800090187379863,
+      "loss": 1.1544,
+      "step": 8894
+    },
+    {
+      "epoch": 0.6841781401430659,
+      "learning_rate": 0.0006797055322656562,
+      "loss": 1.0568,
+      "step": 8895
+    },
+    {
+      "epoch": 0.6842550573032844,
+      "learning_rate": 0.0006794020936911739,
+      "loss": 1.0024,
+      "step": 8896
+    },
+    {
+      "epoch": 0.6843319744635028,
+      "learning_rate": 0.0006790987030322591,
+      "loss": 1.1567,
+      "step": 8897
+    },
+    {
+      "epoch": 0.6844088916237212,
+      "learning_rate": 0.0006787953603066267,
+      "loss": 1.0324,
+      "step": 8898
+    },
+    {
+      "epoch": 0.6844858087839397,
+      "learning_rate": 0.0006784920655319887,
+      "loss": 1.3817,
+      "step": 8899
+    },
+    {
+      "epoch": 0.6845627259441581,
+      "learning_rate": 0.0006781888187260555,
+      "loss": 1.2668,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6846396431043766,
+      "learning_rate": 0.0006778856199065329,
+      "loss": 0.9629,
+      "step": 8901
+    },
+    {
+      "epoch": 0.684716560264595,
+      "learning_rate": 0.0006775824690911264,
+      "loss": 1.3683,
+      "step": 8902
+    },
+    {
+      "epoch": 0.6847934774248134,
+      "learning_rate": 0.0006772793662975366,
+      "loss": 1.2192,
+      "step": 8903
+    },
+    {
+      "epoch": 0.684870394585032,
+      "learning_rate": 0.0006769763115434621,
+      "loss": 1.145,
+      "step": 8904
+    },
+    {
+      "epoch": 0.6849473117452504,
+      "learning_rate": 0.0006766733048465983,
+      "loss": 1.0979,
+      "step": 8905
+    },
+    {
+      "epoch": 0.6850242289054688,
+      "learning_rate": 0.0006763703462246384,
+      "loss": 1.3081,
+      "step": 8906
+    },
+    {
+      "epoch": 0.6851011460656873,
+      "learning_rate": 0.0006760674356952718,
+      "loss": 1.1022,
+      "step": 8907
+    },
+    {
+      "epoch": 0.6851780632259057,
+      "learning_rate": 0.0006757645732761873,
+      "loss": 0.8378,
+      "step": 8908
+    },
+    {
+      "epoch": 0.6852549803861241,
+      "learning_rate": 0.0006754617589850681,
+      "loss": 1.3486,
+      "step": 8909
+    },
+    {
+      "epoch": 0.6853318975463426,
+      "learning_rate": 0.0006751589928395953,
+      "loss": 1.3753,
+      "step": 8910
+    },
+    {
+      "epoch": 0.685408814706561,
+      "learning_rate": 0.0006748562748574491,
+      "loss": 1.0124,
+      "step": 8911
+    },
+    {
+      "epoch": 0.6854857318667795,
+      "learning_rate": 0.000674553605056305,
+      "loss": 1.3546,
+      "step": 8912
+    },
+    {
+      "epoch": 0.6855626490269979,
+      "learning_rate": 0.0006742509834538362,
+      "loss": 0.9445,
+      "step": 8913
+    },
+    {
+      "epoch": 0.6856395661872163,
+      "learning_rate": 0.000673948410067713,
+      "loss": 0.9991,
+      "step": 8914
+    },
+    {
+      "epoch": 0.6857164833474348,
+      "learning_rate": 0.0006736458849156026,
+      "loss": 1.0037,
+      "step": 8915
+    },
+    {
+      "epoch": 0.6857934005076533,
+      "learning_rate": 0.0006733434080151707,
+      "loss": 0.9724,
+      "step": 8916
+    },
+    {
+      "epoch": 0.6858703176678717,
+      "learning_rate": 0.0006730409793840791,
+      "loss": 1.0462,
+      "step": 8917
+    },
+    {
+      "epoch": 0.6859472348280902,
+      "learning_rate": 0.0006727385990399857,
+      "loss": 0.8091,
+      "step": 8918
+    },
+    {
+      "epoch": 0.6860241519883086,
+      "learning_rate": 0.000672436267000548,
+      "loss": 1.1152,
+      "step": 8919
+    },
+    {
+      "epoch": 0.6861010691485271,
+      "learning_rate": 0.0006721339832834186,
+      "loss": 1.4705,
+      "step": 8920
+    },
+    {
+      "epoch": 0.6861779863087455,
+      "learning_rate": 0.0006718317479062499,
+      "loss": 1.2821,
+      "step": 8921
+    },
+    {
+      "epoch": 0.6862549034689639,
+      "learning_rate": 0.0006715295608866878,
+      "loss": 1.1767,
+      "step": 8922
+    },
+    {
+      "epoch": 0.6863318206291824,
+      "learning_rate": 0.0006712274222423777,
+      "loss": 1.3717,
+      "step": 8923
+    },
+    {
+      "epoch": 0.6864087377894008,
+      "learning_rate": 0.0006709253319909627,
+      "loss": 1.3316,
+      "step": 8924
+    },
+    {
+      "epoch": 0.6864856549496192,
+      "learning_rate": 0.0006706232901500814,
+      "loss": 1.103,
+      "step": 8925
+    },
+    {
+      "epoch": 0.6865625721098377,
+      "learning_rate": 0.0006703212967373708,
+      "loss": 1.4027,
+      "step": 8926
+    },
+    {
+      "epoch": 0.6866394892700561,
+      "learning_rate": 0.0006700193517704641,
+      "loss": 1.0511,
+      "step": 8927
+    },
+    {
+      "epoch": 0.6867164064302745,
+      "learning_rate": 0.0006697174552669926,
+      "loss": 1.2876,
+      "step": 8928
+    },
+    {
+      "epoch": 0.686793323590493,
+      "learning_rate": 0.0006694156072445836,
+      "loss": 0.8889,
+      "step": 8929
+    },
+    {
+      "epoch": 0.6868702407507115,
+      "learning_rate": 0.0006691138077208641,
+      "loss": 0.9983,
+      "step": 8930
+    },
+    {
+      "epoch": 0.68694715791093,
+      "learning_rate": 0.0006688120567134541,
+      "loss": 1.0804,
+      "step": 8931
+    },
+    {
+      "epoch": 0.6870240750711484,
+      "learning_rate": 0.0006685103542399749,
+      "loss": 1.1354,
+      "step": 8932
+    },
+    {
+      "epoch": 0.6871009922313668,
+      "learning_rate": 0.0006682087003180427,
+      "loss": 1.2051,
+      "step": 8933
+    },
+    {
+      "epoch": 0.6871779093915853,
+      "learning_rate": 0.0006679070949652714,
+      "loss": 1.5191,
+      "step": 8934
+    },
+    {
+      "epoch": 0.6872548265518037,
+      "learning_rate": 0.0006676055381992718,
+      "loss": 1.4377,
+      "step": 8935
+    },
+    {
+      "epoch": 0.6873317437120221,
+      "learning_rate": 0.0006673040300376526,
+      "loss": 1.211,
+      "step": 8936
+    },
+    {
+      "epoch": 0.6874086608722406,
+      "learning_rate": 0.0006670025704980181,
+      "loss": 1.0946,
+      "step": 8937
+    },
+    {
+      "epoch": 0.687485578032459,
+      "learning_rate": 0.0006667011595979723,
+      "loss": 0.8467,
+      "step": 8938
+    },
+    {
+      "epoch": 0.6875624951926775,
+      "learning_rate": 0.0006663997973551144,
+      "loss": 1.3303,
+      "step": 8939
+    },
+    {
+      "epoch": 0.6876394123528959,
+      "learning_rate": 0.0006660984837870409,
+      "loss": 1.2592,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6877163295131143,
+      "learning_rate": 0.000665797218911346,
+      "loss": 1.1655,
+      "step": 8941
+    },
+    {
+      "epoch": 0.6877932466733329,
+      "learning_rate": 0.0006654960027456204,
+      "loss": 1.1318,
+      "step": 8942
+    },
+    {
+      "epoch": 0.6878701638335513,
+      "learning_rate": 0.0006651948353074539,
+      "loss": 1.2717,
+      "step": 8943
+    },
+    {
+      "epoch": 0.6879470809937697,
+      "learning_rate": 0.0006648937166144305,
+      "loss": 0.7865,
+      "step": 8944
+    },
+    {
+      "epoch": 0.6880239981539882,
+      "learning_rate": 0.0006645926466841328,
+      "loss": 1.4333,
+      "step": 8945
+    },
+    {
+      "epoch": 0.6881009153142066,
+      "learning_rate": 0.0006642916255341415,
+      "loss": 1.3917,
+      "step": 8946
+    },
+    {
+      "epoch": 0.6881778324744251,
+      "learning_rate": 0.0006639906531820334,
+      "loss": 1.0905,
+      "step": 8947
+    },
+    {
+      "epoch": 0.6882547496346435,
+      "learning_rate": 0.0006636897296453821,
+      "loss": 1.5172,
+      "step": 8948
+    },
+    {
+      "epoch": 0.6883316667948619,
+      "learning_rate": 0.0006633888549417591,
+      "loss": 1.0846,
+      "step": 8949
+    },
+    {
+      "epoch": 0.6884085839550804,
+      "learning_rate": 0.0006630880290887323,
+      "loss": 1.1195,
+      "step": 8950
+    },
+    {
+      "epoch": 0.6884855011152988,
+      "learning_rate": 0.0006627872521038682,
+      "loss": 1.0971,
+      "step": 8951
+    },
+    {
+      "epoch": 0.6885624182755172,
+      "learning_rate": 0.0006624865240047295,
+      "loss": 1.135,
+      "step": 8952
+    },
+    {
+      "epoch": 0.6886393354357357,
+      "learning_rate": 0.0006621858448088745,
+      "loss": 1.0745,
+      "step": 8953
+    },
+    {
+      "epoch": 0.6887162525959541,
+      "learning_rate": 0.0006618852145338617,
+      "loss": 1.1851,
+      "step": 8954
+    },
+    {
+      "epoch": 0.6887931697561726,
+      "learning_rate": 0.0006615846331972445,
+      "loss": 1.2612,
+      "step": 8955
+    },
+    {
+      "epoch": 0.6888700869163911,
+      "learning_rate": 0.0006612841008165744,
+      "loss": 1.1767,
+      "step": 8956
+    },
+    {
+      "epoch": 0.6889470040766095,
+      "learning_rate": 0.0006609836174093998,
+      "loss": 1.0124,
+      "step": 8957
+    },
+    {
+      "epoch": 0.689023921236828,
+      "learning_rate": 0.0006606831829932657,
+      "loss": 1.03,
+      "step": 8958
+    },
+    {
+      "epoch": 0.6891008383970464,
+      "learning_rate": 0.0006603827975857157,
+      "loss": 1.1947,
+      "step": 8959
+    },
+    {
+      "epoch": 0.6891777555572648,
+      "learning_rate": 0.0006600824612042894,
+      "loss": 1.1655,
+      "step": 8960
+    },
+    {
+      "epoch": 0.6892546727174833,
+      "learning_rate": 0.0006597821738665233,
+      "loss": 1.2997,
+      "step": 8961
+    },
+    {
+      "epoch": 0.6893315898777017,
+      "learning_rate": 0.0006594819355899517,
+      "loss": 1.5447,
+      "step": 8962
+    },
+    {
+      "epoch": 0.6894085070379201,
+      "learning_rate": 0.000659181746392106,
+      "loss": 1.4082,
+      "step": 8963
+    },
+    {
+      "epoch": 0.6894854241981386,
+      "learning_rate": 0.0006588816062905137,
+      "loss": 1.1373,
+      "step": 8964
+    },
+    {
+      "epoch": 0.689562341358357,
+      "learning_rate": 0.0006585815153027021,
+      "loss": 1.2028,
+      "step": 8965
+    },
+    {
+      "epoch": 0.6896392585185755,
+      "learning_rate": 0.0006582814734461917,
+      "loss": 1.1343,
+      "step": 8966
+    },
+    {
+      "epoch": 0.689716175678794,
+      "learning_rate": 0.000657981480738504,
+      "loss": 1.079,
+      "step": 8967
+    },
+    {
+      "epoch": 0.6897930928390124,
+      "learning_rate": 0.0006576815371971549,
+      "loss": 1.0593,
+      "step": 8968
+    },
+    {
+      "epoch": 0.6898700099992309,
+      "learning_rate": 0.0006573816428396589,
+      "loss": 1.5121,
+      "step": 8969
+    },
+    {
+      "epoch": 0.6899469271594493,
+      "learning_rate": 0.0006570817976835269,
+      "loss": 0.6794,
+      "step": 8970
+    },
+    {
+      "epoch": 0.6900238443196677,
+      "learning_rate": 0.000656782001746267,
+      "loss": 1.1135,
+      "step": 8971
+    },
+    {
+      "epoch": 0.6901007614798862,
+      "learning_rate": 0.0006564822550453845,
+      "loss": 0.9261,
+      "step": 8972
+    },
+    {
+      "epoch": 0.6901776786401046,
+      "learning_rate": 0.0006561825575983825,
+      "loss": 1.6339,
+      "step": 8973
+    },
+    {
+      "epoch": 0.690254595800323,
+      "learning_rate": 0.0006558829094227612,
+      "loss": 1.1517,
+      "step": 8974
+    },
+    {
+      "epoch": 0.6903315129605415,
+      "learning_rate": 0.0006555833105360153,
+      "loss": 1.0657,
+      "step": 8975
+    },
+    {
+      "epoch": 0.6904084301207599,
+      "learning_rate": 0.0006552837609556404,
+      "loss": 1.1853,
+      "step": 8976
+    },
+    {
+      "epoch": 0.6904853472809784,
+      "learning_rate": 0.0006549842606991265,
+      "loss": 0.9982,
+      "step": 8977
+    },
+    {
+      "epoch": 0.6905622644411968,
+      "learning_rate": 0.0006546848097839635,
+      "loss": 0.9886,
+      "step": 8978
+    },
+    {
+      "epoch": 0.6906391816014152,
+      "learning_rate": 0.0006543854082276348,
+      "loss": 1.267,
+      "step": 8979
+    },
+    {
+      "epoch": 0.6907160987616338,
+      "learning_rate": 0.0006540860560476228,
+      "loss": 1.11,
+      "step": 8980
+    },
+    {
+      "epoch": 0.6907930159218522,
+      "learning_rate": 0.0006537867532614083,
+      "loss": 1.1821,
+      "step": 8981
+    },
+    {
+      "epoch": 0.6908699330820706,
+      "learning_rate": 0.0006534874998864671,
+      "loss": 1.3343,
+      "step": 8982
+    },
+    {
+      "epoch": 0.6909468502422891,
+      "learning_rate": 0.0006531882959402729,
+      "loss": 0.9674,
+      "step": 8983
+    },
+    {
+      "epoch": 0.6910237674025075,
+      "learning_rate": 0.0006528891414402967,
+      "loss": 1.2381,
+      "step": 8984
+    },
+    {
+      "epoch": 0.691100684562726,
+      "learning_rate": 0.0006525900364040059,
+      "loss": 1.0584,
+      "step": 8985
+    },
+    {
+      "epoch": 0.6911776017229444,
+      "learning_rate": 0.0006522909808488664,
+      "loss": 1.3482,
+      "step": 8986
+    },
+    {
+      "epoch": 0.6912545188831628,
+      "learning_rate": 0.0006519919747923406,
+      "loss": 1.2494,
+      "step": 8987
+    },
+    {
+      "epoch": 0.6913314360433813,
+      "learning_rate": 0.0006516930182518861,
+      "loss": 1.1126,
+      "step": 8988
+    },
+    {
+      "epoch": 0.6914083532035997,
+      "learning_rate": 0.000651394111244961,
+      "loss": 1.2531,
+      "step": 8989
+    },
+    {
+      "epoch": 0.6914852703638181,
+      "learning_rate": 0.0006510952537890179,
+      "loss": 1.2582,
+      "step": 8990
+    },
+    {
+      "epoch": 0.6915621875240366,
+      "learning_rate": 0.0006507964459015078,
+      "loss": 1.3781,
+      "step": 8991
+    },
+    {
+      "epoch": 0.691639104684255,
+      "learning_rate": 0.0006504976875998783,
+      "loss": 1.3166,
+      "step": 8992
+    },
+    {
+      "epoch": 0.6917160218444734,
+      "learning_rate": 0.0006501989789015739,
+      "loss": 1.2769,
+      "step": 8993
+    },
+    {
+      "epoch": 0.691792939004692,
+      "learning_rate": 0.0006499003198240364,
+      "loss": 1.0458,
+      "step": 8994
+    },
+    {
+      "epoch": 0.6918698561649104,
+      "learning_rate": 0.0006496017103847058,
+      "loss": 1.1119,
+      "step": 8995
+    },
+    {
+      "epoch": 0.6919467733251289,
+      "learning_rate": 0.0006493031506010174,
+      "loss": 1.2924,
+      "step": 8996
+    },
+    {
+      "epoch": 0.6920236904853473,
+      "learning_rate": 0.0006490046404904048,
+      "loss": 1.3174,
+      "step": 8997
+    },
+    {
+      "epoch": 0.6921006076455657,
+      "learning_rate": 0.0006487061800702981,
+      "loss": 0.9987,
+      "step": 8998
+    },
+    {
+      "epoch": 0.6921775248057842,
+      "learning_rate": 0.0006484077693581239,
+      "loss": 0.9899,
+      "step": 8999
+    },
+    {
+      "epoch": 0.6922544419660026,
+      "learning_rate": 0.0006481094083713089,
+      "loss": 1.1364,
+      "step": 9000
+    },
+    {
+      "epoch": 0.692331359126221,
+      "learning_rate": 0.0006478110971272729,
+      "loss": 1.058,
+      "step": 9001
+    },
+    {
+      "epoch": 0.6924082762864395,
+      "learning_rate": 0.0006475128356434345,
+      "loss": 1.2429,
+      "step": 9002
+    },
+    {
+      "epoch": 0.6924851934466579,
+      "learning_rate": 0.0006472146239372104,
+      "loss": 1.052,
+      "step": 9003
+    },
+    {
+      "epoch": 0.6925621106068764,
+      "learning_rate": 0.0006469164620260132,
+      "loss": 0.8947,
+      "step": 9004
+    },
+    {
+      "epoch": 0.6926390277670949,
+      "learning_rate": 0.0006466183499272529,
+      "loss": 1.4427,
+      "step": 9005
+    },
+    {
+      "epoch": 0.6927159449273133,
+      "learning_rate": 0.0006463202876583367,
+      "loss": 1.3991,
+      "step": 9006
+    },
+    {
+      "epoch": 0.6927928620875318,
+      "learning_rate": 0.0006460222752366679,
+      "loss": 0.8358,
+      "step": 9007
+    },
+    {
+      "epoch": 0.6928697792477502,
+      "learning_rate": 0.0006457243126796489,
+      "loss": 1.2745,
+      "step": 9008
+    },
+    {
+      "epoch": 0.6929466964079686,
+      "learning_rate": 0.0006454264000046778,
+      "loss": 1.1079,
+      "step": 9009
+    },
+    {
+      "epoch": 0.6930236135681871,
+      "learning_rate": 0.0006451285372291496,
+      "loss": 0.9257,
+      "step": 9010
+    },
+    {
+      "epoch": 0.6931005307284055,
+      "learning_rate": 0.000644830724370457,
+      "loss": 1.387,
+      "step": 9011
+    },
+    {
+      "epoch": 0.6931774478886239,
+      "learning_rate": 0.00064453296144599,
+      "loss": 1.0353,
+      "step": 9012
+    },
+    {
+      "epoch": 0.6932543650488424,
+      "learning_rate": 0.0006442352484731346,
+      "loss": 1.0631,
+      "step": 9013
+    },
+    {
+      "epoch": 0.6933312822090608,
+      "learning_rate": 0.000643937585469275,
+      "loss": 1.3631,
+      "step": 9014
+    },
+    {
+      "epoch": 0.6934081993692793,
+      "learning_rate": 0.0006436399724517915,
+      "loss": 0.8408,
+      "step": 9015
+    },
+    {
+      "epoch": 0.6934851165294977,
+      "learning_rate": 0.0006433424094380628,
+      "loss": 1.1477,
+      "step": 9016
+    },
+    {
+      "epoch": 0.6935620336897161,
+      "learning_rate": 0.0006430448964454638,
+      "loss": 1.5772,
+      "step": 9017
+    },
+    {
+      "epoch": 0.6936389508499347,
+      "learning_rate": 0.0006427474334913664,
+      "loss": 1.3088,
+      "step": 9018
+    },
+    {
+      "epoch": 0.6937158680101531,
+      "learning_rate": 0.0006424500205931395,
+      "loss": 1.3001,
+      "step": 9019
+    },
+    {
+      "epoch": 0.6937927851703715,
+      "learning_rate": 0.00064215265776815,
+      "loss": 1.0358,
+      "step": 9020
+    },
+    {
+      "epoch": 0.69386970233059,
+      "learning_rate": 0.00064185534503376,
+      "loss": 1.4033,
+      "step": 9021
+    },
+    {
+      "epoch": 0.6939466194908084,
+      "learning_rate": 0.0006415580824073314,
+      "loss": 1.3163,
+      "step": 9022
+    },
+    {
+      "epoch": 0.6940235366510269,
+      "learning_rate": 0.000641260869906221,
+      "loss": 1.162,
+      "step": 9023
+    },
+    {
+      "epoch": 0.6941004538112453,
+      "learning_rate": 0.0006409637075477834,
+      "loss": 0.9415,
+      "step": 9024
+    },
+    {
+      "epoch": 0.6941773709714637,
+      "learning_rate": 0.0006406665953493701,
+      "loss": 1.0079,
+      "step": 9025
+    },
+    {
+      "epoch": 0.6942542881316822,
+      "learning_rate": 0.00064036953332833,
+      "loss": 1.2808,
+      "step": 9026
+    },
+    {
+      "epoch": 0.6943312052919006,
+      "learning_rate": 0.0006400725215020087,
+      "loss": 0.6978,
+      "step": 9027
+    },
+    {
+      "epoch": 0.694408122452119,
+      "learning_rate": 0.0006397755598877488,
+      "loss": 0.9175,
+      "step": 9028
+    },
+    {
+      "epoch": 0.6944850396123375,
+      "learning_rate": 0.0006394786485028901,
+      "loss": 0.9635,
+      "step": 9029
+    },
+    {
+      "epoch": 0.6945619567725559,
+      "learning_rate": 0.0006391817873647707,
+      "loss": 1.3319,
+      "step": 9030
+    },
+    {
+      "epoch": 0.6946388739327743,
+      "learning_rate": 0.0006388849764907236,
+      "loss": 1.0064,
+      "step": 9031
+    },
+    {
+      "epoch": 0.6947157910929929,
+      "learning_rate": 0.0006385882158980802,
+      "loss": 1.2103,
+      "step": 9032
+    },
+    {
+      "epoch": 0.6947927082532113,
+      "learning_rate": 0.0006382915056041686,
+      "loss": 1.1953,
+      "step": 9033
+    },
+    {
+      "epoch": 0.6948696254134298,
+      "learning_rate": 0.0006379948456263137,
+      "loss": 1.0738,
+      "step": 9034
+    },
+    {
+      "epoch": 0.6949465425736482,
+      "learning_rate": 0.0006376982359818384,
+      "loss": 0.757,
+      "step": 9035
+    },
+    {
+      "epoch": 0.6950234597338666,
+      "learning_rate": 0.0006374016766880625,
+      "loss": 1.3447,
+      "step": 9036
+    },
+    {
+      "epoch": 0.6951003768940851,
+      "learning_rate": 0.0006371051677623004,
+      "loss": 1.0917,
+      "step": 9037
+    },
+    {
+      "epoch": 0.6951772940543035,
+      "learning_rate": 0.0006368087092218675,
+      "loss": 1.1666,
+      "step": 9038
+    },
+    {
+      "epoch": 0.6952542112145219,
+      "learning_rate": 0.0006365123010840737,
+      "loss": 1.1845,
+      "step": 9039
+    },
+    {
+      "epoch": 0.6953311283747404,
+      "learning_rate": 0.0006362159433662265,
+      "loss": 1.0628,
+      "step": 9040
+    },
+    {
+      "epoch": 0.6954080455349588,
+      "learning_rate": 0.0006359196360856305,
+      "loss": 0.8804,
+      "step": 9041
+    },
+    {
+      "epoch": 0.6954849626951773,
+      "learning_rate": 0.000635623379259587,
+      "loss": 1.5282,
+      "step": 9042
+    },
+    {
+      "epoch": 0.6955618798553957,
+      "learning_rate": 0.0006353271729053957,
+      "loss": 1.197,
+      "step": 9043
+    },
+    {
+      "epoch": 0.6956387970156142,
+      "learning_rate": 0.0006350310170403522,
+      "loss": 1.1779,
+      "step": 9044
+    },
+    {
+      "epoch": 0.6957157141758327,
+      "learning_rate": 0.0006347349116817487,
+      "loss": 1.1069,
+      "step": 9045
+    },
+    {
+      "epoch": 0.6957926313360511,
+      "learning_rate": 0.0006344388568468757,
+      "loss": 0.9396,
+      "step": 9046
+    },
+    {
+      "epoch": 0.6958695484962695,
+      "learning_rate": 0.0006341428525530198,
+      "loss": 1.3949,
+      "step": 9047
+    },
+    {
+      "epoch": 0.695946465656488,
+      "learning_rate": 0.0006338468988174646,
+      "loss": 1.337,
+      "step": 9048
+    },
+    {
+      "epoch": 0.6960233828167064,
+      "learning_rate": 0.0006335509956574928,
+      "loss": 0.8997,
+      "step": 9049
+    },
+    {
+      "epoch": 0.6961002999769248,
+      "learning_rate": 0.0006332551430903808,
+      "loss": 0.7967,
+      "step": 9050
+    },
+    {
+      "epoch": 0.6961772171371433,
+      "learning_rate": 0.0006329593411334037,
+      "loss": 1.2711,
+      "step": 9051
+    },
+    {
+      "epoch": 0.6962541342973617,
+      "learning_rate": 0.0006326635898038347,
+      "loss": 1.3538,
+      "step": 9052
+    },
+    {
+      "epoch": 0.6963310514575802,
+      "learning_rate": 0.0006323678891189429,
+      "loss": 1.4564,
+      "step": 9053
+    },
+    {
+      "epoch": 0.6964079686177986,
+      "learning_rate": 0.0006320722390959941,
+      "loss": 1.0628,
+      "step": 9054
+    },
+    {
+      "epoch": 0.696484885778017,
+      "learning_rate": 0.0006317766397522517,
+      "loss": 0.4444,
+      "step": 9055
+    },
+    {
+      "epoch": 0.6965618029382356,
+      "learning_rate": 0.0006314810911049758,
+      "loss": 1.34,
+      "step": 9056
+    },
+    {
+      "epoch": 0.696638720098454,
+      "learning_rate": 0.0006311855931714244,
+      "loss": 1.4358,
+      "step": 9057
+    },
+    {
+      "epoch": 0.6967156372586724,
+      "learning_rate": 0.0006308901459688526,
+      "loss": 1.3789,
+      "step": 9058
+    },
+    {
+      "epoch": 0.6967925544188909,
+      "learning_rate": 0.0006305947495145096,
+      "loss": 1.0355,
+      "step": 9059
+    },
+    {
+      "epoch": 0.6968694715791093,
+      "learning_rate": 0.0006302994038256458,
+      "loss": 1.0603,
+      "step": 9060
+    },
+    {
+      "epoch": 0.6969463887393278,
+      "learning_rate": 0.0006300041089195057,
+      "loss": 1.0929,
+      "step": 9061
+    },
+    {
+      "epoch": 0.6970233058995462,
+      "learning_rate": 0.0006297088648133336,
+      "loss": 0.8972,
+      "step": 9062
+    },
+    {
+      "epoch": 0.6971002230597646,
+      "learning_rate": 0.000629413671524367,
+      "loss": 1.1272,
+      "step": 9063
+    },
+    {
+      "epoch": 0.6971771402199831,
+      "learning_rate": 0.0006291185290698433,
+      "loss": 1.2359,
+      "step": 9064
+    },
+    {
+      "epoch": 0.6972540573802015,
+      "learning_rate": 0.0006288234374669966,
+      "loss": 1.0291,
+      "step": 9065
+    },
+    {
+      "epoch": 0.6973309745404199,
+      "learning_rate": 0.0006285283967330573,
+      "loss": 1.2536,
+      "step": 9066
+    },
+    {
+      "epoch": 0.6974078917006384,
+      "learning_rate": 0.0006282334068852531,
+      "loss": 1.2681,
+      "step": 9067
+    },
+    {
+      "epoch": 0.6974848088608568,
+      "learning_rate": 0.0006279384679408089,
+      "loss": 1.262,
+      "step": 9068
+    },
+    {
+      "epoch": 0.6975617260210754,
+      "learning_rate": 0.0006276435799169458,
+      "loss": 1.1998,
+      "step": 9069
+    },
+    {
+      "epoch": 0.6976386431812938,
+      "learning_rate": 0.0006273487428308837,
+      "loss": 1.0804,
+      "step": 9070
+    },
+    {
+      "epoch": 0.6977155603415122,
+      "learning_rate": 0.0006270539566998386,
+      "loss": 1.2336,
+      "step": 9071
+    },
+    {
+      "epoch": 0.6977924775017307,
+      "learning_rate": 0.0006267592215410217,
+      "loss": 1.2207,
+      "step": 9072
+    },
+    {
+      "epoch": 0.6978693946619491,
+      "learning_rate": 0.0006264645373716444,
+      "loss": 1.2066,
+      "step": 9073
+    },
+    {
+      "epoch": 0.6979463118221675,
+      "learning_rate": 0.000626169904208913,
+      "loss": 1.2549,
+      "step": 9074
+    },
+    {
+      "epoch": 0.698023228982386,
+      "learning_rate": 0.0006258753220700318,
+      "loss": 1.4719,
+      "step": 9075
+    },
+    {
+      "epoch": 0.6981001461426044,
+      "learning_rate": 0.0006255807909722015,
+      "loss": 1.0409,
+      "step": 9076
+    },
+    {
+      "epoch": 0.6981770633028228,
+      "learning_rate": 0.0006252863109326201,
+      "loss": 1.3939,
+      "step": 9077
+    },
+    {
+      "epoch": 0.6982539804630413,
+      "learning_rate": 0.000624991881968482,
+      "loss": 1.1164,
+      "step": 9078
+    },
+    {
+      "epoch": 0.6983308976232597,
+      "learning_rate": 0.0006246975040969804,
+      "loss": 1.4007,
+      "step": 9079
+    },
+    {
+      "epoch": 0.6984078147834782,
+      "learning_rate": 0.0006244031773353037,
+      "loss": 1.0795,
+      "step": 9080
+    },
+    {
+      "epoch": 0.6984847319436966,
+      "learning_rate": 0.0006241089017006382,
+      "loss": 1.3028,
+      "step": 9081
+    },
+    {
+      "epoch": 0.698561649103915,
+      "learning_rate": 0.0006238146772101665,
+      "loss": 1.1655,
+      "step": 9082
+    },
+    {
+      "epoch": 0.6986385662641336,
+      "learning_rate": 0.0006235205038810685,
+      "loss": 1.1177,
+      "step": 9083
+    },
+    {
+      "epoch": 0.698715483424352,
+      "learning_rate": 0.0006232263817305228,
+      "loss": 1.3024,
+      "step": 9084
+    },
+    {
+      "epoch": 0.6987924005845704,
+      "learning_rate": 0.000622932310775702,
+      "loss": 1.2065,
+      "step": 9085
+    },
+    {
+      "epoch": 0.6988693177447889,
+      "learning_rate": 0.000622638291033777,
+      "loss": 1.229,
+      "step": 9086
+    },
+    {
+      "epoch": 0.6989462349050073,
+      "learning_rate": 0.0006223443225219169,
+      "loss": 1.0655,
+      "step": 9087
+    },
+    {
+      "epoch": 0.6990231520652258,
+      "learning_rate": 0.0006220504052572867,
+      "loss": 0.8928,
+      "step": 9088
+    },
+    {
+      "epoch": 0.6991000692254442,
+      "learning_rate": 0.0006217565392570483,
+      "loss": 0.8444,
+      "step": 9089
+    },
+    {
+      "epoch": 0.6991769863856626,
+      "learning_rate": 0.0006214627245383609,
+      "loss": 1.1482,
+      "step": 9090
+    },
+    {
+      "epoch": 0.6992539035458811,
+      "learning_rate": 0.0006211689611183799,
+      "loss": 1.3874,
+      "step": 9091
+    },
+    {
+      "epoch": 0.6993308207060995,
+      "learning_rate": 0.0006208752490142597,
+      "loss": 1.3886,
+      "step": 9092
+    },
+    {
+      "epoch": 0.6994077378663179,
+      "learning_rate": 0.0006205815882431506,
+      "loss": 1.3982,
+      "step": 9093
+    },
+    {
+      "epoch": 0.6994846550265365,
+      "learning_rate": 0.000620287978822198,
+      "loss": 0.9534,
+      "step": 9094
+    },
+    {
+      "epoch": 0.6995615721867549,
+      "learning_rate": 0.0006199944207685475,
+      "loss": 1.1033,
+      "step": 9095
+    },
+    {
+      "epoch": 0.6996384893469733,
+      "learning_rate": 0.0006197009140993401,
+      "loss": 1.2295,
+      "step": 9096
+    },
+    {
+      "epoch": 0.6997154065071918,
+      "learning_rate": 0.0006194074588317137,
+      "loss": 0.8809,
+      "step": 9097
+    },
+    {
+      "epoch": 0.6997923236674102,
+      "learning_rate": 0.0006191140549828034,
+      "loss": 0.9768,
+      "step": 9098
+    },
+    {
+      "epoch": 0.6998692408276287,
+      "learning_rate": 0.000618820702569741,
+      "loss": 1.5388,
+      "step": 9099
+    },
+    {
+      "epoch": 0.6999461579878471,
+      "learning_rate": 0.0006185274016096567,
+      "loss": 1.1944,
+      "step": 9100
+    },
+    {
+      "epoch": 0.7000230751480655,
+      "learning_rate": 0.0006182341521196759,
+      "loss": 1.3062,
+      "step": 9101
+    },
+    {
+      "epoch": 0.700099992308284,
+      "learning_rate": 0.000617940954116922,
+      "loss": 1.0186,
+      "step": 9102
+    },
+    {
+      "epoch": 0.7001769094685024,
+      "learning_rate": 0.0006176478076185151,
+      "loss": 1.1119,
+      "step": 9103
+    },
+    {
+      "epoch": 0.7002538266287208,
+      "learning_rate": 0.0006173547126415722,
+      "loss": 0.9446,
+      "step": 9104
+    },
+    {
+      "epoch": 0.7003307437889393,
+      "learning_rate": 0.0006170616692032071,
+      "loss": 1.0974,
+      "step": 9105
+    },
+    {
+      "epoch": 0.7004076609491577,
+      "learning_rate": 0.0006167686773205322,
+      "loss": 0.9449,
+      "step": 9106
+    },
+    {
+      "epoch": 0.7004845781093763,
+      "learning_rate": 0.0006164757370106537,
+      "loss": 1.1489,
+      "step": 9107
+    },
+    {
+      "epoch": 0.7005614952695947,
+      "learning_rate": 0.0006161828482906783,
+      "loss": 1.2107,
+      "step": 9108
+    },
+    {
+      "epoch": 0.7006384124298131,
+      "learning_rate": 0.0006158900111777075,
+      "loss": 1.2626,
+      "step": 9109
+    },
+    {
+      "epoch": 0.7007153295900316,
+      "learning_rate": 0.0006155972256888402,
+      "loss": 1.3257,
+      "step": 9110
+    },
+    {
+      "epoch": 0.70079224675025,
+      "learning_rate": 0.0006153044918411727,
+      "loss": 1.0702,
+      "step": 9111
+    },
+    {
+      "epoch": 0.7008691639104684,
+      "learning_rate": 0.000615011809651798,
+      "loss": 0.8849,
+      "step": 9112
+    },
+    {
+      "epoch": 0.7009460810706869,
+      "learning_rate": 0.0006147191791378053,
+      "loss": 0.9322,
+      "step": 9113
+    },
+    {
+      "epoch": 0.7010229982309053,
+      "learning_rate": 0.0006144266003162831,
+      "loss": 1.2252,
+      "step": 9114
+    },
+    {
+      "epoch": 0.7010999153911237,
+      "learning_rate": 0.0006141340732043152,
+      "loss": 1.1744,
+      "step": 9115
+    },
+    {
+      "epoch": 0.7011768325513422,
+      "learning_rate": 0.0006138415978189808,
+      "loss": 1.2032,
+      "step": 9116
+    },
+    {
+      "epoch": 0.7012537497115606,
+      "learning_rate": 0.0006135491741773594,
+      "loss": 0.8612,
+      "step": 9117
+    },
+    {
+      "epoch": 0.7013306668717791,
+      "learning_rate": 0.0006132568022965253,
+      "loss": 1.3158,
+      "step": 9118
+    },
+    {
+      "epoch": 0.7014075840319975,
+      "learning_rate": 0.0006129644821935517,
+      "loss": 1.1177,
+      "step": 9119
+    },
+    {
+      "epoch": 0.701484501192216,
+      "learning_rate": 0.0006126722138855057,
+      "loss": 1.1121,
+      "step": 9120
+    },
+    {
+      "epoch": 0.7015614183524345,
+      "learning_rate": 0.0006123799973894534,
+      "loss": 1.1532,
+      "step": 9121
+    },
+    {
+      "epoch": 0.7016383355126529,
+      "learning_rate": 0.0006120878327224587,
+      "loss": 1.048,
+      "step": 9122
+    },
+    {
+      "epoch": 0.7017152526728713,
+      "learning_rate": 0.0006117957199015809,
+      "loss": 1.0144,
+      "step": 9123
+    },
+    {
+      "epoch": 0.7017921698330898,
+      "learning_rate": 0.0006115036589438767,
+      "loss": 1.1654,
+      "step": 9124
+    },
+    {
+      "epoch": 0.7018690869933082,
+      "learning_rate": 0.0006112116498663996,
+      "loss": 1.3741,
+      "step": 9125
+    },
+    {
+      "epoch": 0.7019460041535267,
+      "learning_rate": 0.0006109196926862001,
+      "loss": 0.7862,
+      "step": 9126
+    },
+    {
+      "epoch": 0.7020229213137451,
+      "learning_rate": 0.000610627787420327,
+      "loss": 1.3426,
+      "step": 9127
+    },
+    {
+      "epoch": 0.7020998384739635,
+      "learning_rate": 0.0006103359340858247,
+      "loss": 1.1919,
+      "step": 9128
+    },
+    {
+      "epoch": 0.702176755634182,
+      "learning_rate": 0.0006100441326997332,
+      "loss": 1.0977,
+      "step": 9129
+    },
+    {
+      "epoch": 0.7022536727944004,
+      "learning_rate": 0.0006097523832790929,
+      "loss": 1.0682,
+      "step": 9130
+    },
+    {
+      "epoch": 0.7023305899546188,
+      "learning_rate": 0.0006094606858409388,
+      "loss": 1.3242,
+      "step": 9131
+    },
+    {
+      "epoch": 0.7024075071148373,
+      "learning_rate": 0.0006091690404023034,
+      "loss": 1.0938,
+      "step": 9132
+    },
+    {
+      "epoch": 0.7024844242750558,
+      "learning_rate": 0.000608877446980216,
+      "loss": 0.9688,
+      "step": 9133
+    },
+    {
+      "epoch": 0.7025613414352742,
+      "learning_rate": 0.0006085859055917031,
+      "loss": 1.1817,
+      "step": 9134
+    },
+    {
+      "epoch": 0.7026382585954927,
+      "learning_rate": 0.000608294416253788,
+      "loss": 0.8119,
+      "step": 9135
+    },
+    {
+      "epoch": 0.7027151757557111,
+      "learning_rate": 0.0006080029789834913,
+      "loss": 1.3178,
+      "step": 9136
+    },
+    {
+      "epoch": 0.7027920929159296,
+      "learning_rate": 0.0006077115937978306,
+      "loss": 0.8258,
+      "step": 9137
+    },
+    {
+      "epoch": 0.702869010076148,
+      "learning_rate": 0.0006074202607138197,
+      "loss": 0.9626,
+      "step": 9138
+    },
+    {
+      "epoch": 0.7029459272363664,
+      "learning_rate": 0.0006071289797484703,
+      "loss": 1.0446,
+      "step": 9139
+    },
+    {
+      "epoch": 0.7030228443965849,
+      "learning_rate": 0.0006068377509187894,
+      "loss": 1.1184,
+      "step": 9140
+    },
+    {
+      "epoch": 0.7030997615568033,
+      "learning_rate": 0.0006065465742417842,
+      "loss": 1.6268,
+      "step": 9141
+    },
+    {
+      "epoch": 0.7031766787170217,
+      "learning_rate": 0.0006062554497344553,
+      "loss": 0.9224,
+      "step": 9142
+    },
+    {
+      "epoch": 0.7032535958772402,
+      "learning_rate": 0.0006059643774138016,
+      "loss": 0.7972,
+      "step": 9143
+    },
+    {
+      "epoch": 0.7033305130374586,
+      "learning_rate": 0.00060567335729682,
+      "loss": 0.965,
+      "step": 9144
+    },
+    {
+      "epoch": 0.7034074301976772,
+      "learning_rate": 0.0006053823894005031,
+      "loss": 1.191,
+      "step": 9145
+    },
+    {
+      "epoch": 0.7034843473578956,
+      "learning_rate": 0.000605091473741841,
+      "loss": 1.2312,
+      "step": 9146
+    },
+    {
+      "epoch": 0.703561264518114,
+      "learning_rate": 0.0006048006103378203,
+      "loss": 0.6918,
+      "step": 9147
+    },
+    {
+      "epoch": 0.7036381816783325,
+      "learning_rate": 0.0006045097992054245,
+      "loss": 0.7676,
+      "step": 9148
+    },
+    {
+      "epoch": 0.7037150988385509,
+      "learning_rate": 0.0006042190403616351,
+      "loss": 1.4219,
+      "step": 9149
+    },
+    {
+      "epoch": 0.7037920159987693,
+      "learning_rate": 0.0006039283338234303,
+      "loss": 0.998,
+      "step": 9150
+    },
+    {
+      "epoch": 0.7038689331589878,
+      "learning_rate": 0.0006036376796077826,
+      "loss": 1.0559,
+      "step": 9151
+    },
+    {
+      "epoch": 0.7039458503192062,
+      "learning_rate": 0.0006033470777316657,
+      "loss": 1.4876,
+      "step": 9152
+    },
+    {
+      "epoch": 0.7040227674794246,
+      "learning_rate": 0.0006030565282120473,
+      "loss": 1.4646,
+      "step": 9153
+    },
+    {
+      "epoch": 0.7040996846396431,
+      "learning_rate": 0.000602766031065893,
+      "loss": 1.0207,
+      "step": 9154
+    },
+    {
+      "epoch": 0.7041766017998615,
+      "learning_rate": 0.0006024755863101652,
+      "loss": 1.0601,
+      "step": 9155
+    },
+    {
+      "epoch": 0.70425351896008,
+      "learning_rate": 0.0006021851939618228,
+      "loss": 1.3219,
+      "step": 9156
+    },
+    {
+      "epoch": 0.7043304361202984,
+      "learning_rate": 0.0006018948540378232,
+      "loss": 1.1121,
+      "step": 9157
+    },
+    {
+      "epoch": 0.7044073532805168,
+      "learning_rate": 0.000601604566555119,
+      "loss": 1.3604,
+      "step": 9158
+    },
+    {
+      "epoch": 0.7044842704407354,
+      "learning_rate": 0.0006013143315306603,
+      "loss": 1.2011,
+      "step": 9159
+    },
+    {
+      "epoch": 0.7045611876009538,
+      "learning_rate": 0.0006010241489813945,
+      "loss": 1.1483,
+      "step": 9160
+    },
+    {
+      "epoch": 0.7046381047611722,
+      "learning_rate": 0.0006007340189242656,
+      "loss": 1.1276,
+      "step": 9161
+    },
+    {
+      "epoch": 0.7047150219213907,
+      "learning_rate": 0.0006004439413762138,
+      "loss": 1.2896,
+      "step": 9162
+    },
+    {
+      "epoch": 0.7047919390816091,
+      "learning_rate": 0.000600153916354179,
+      "loss": 0.8891,
+      "step": 9163
+    },
+    {
+      "epoch": 0.7048688562418276,
+      "learning_rate": 0.0005998639438750936,
+      "loss": 0.9449,
+      "step": 9164
+    },
+    {
+      "epoch": 0.704945773402046,
+      "learning_rate": 0.0005995740239558914,
+      "loss": 1.0639,
+      "step": 9165
+    },
+    {
+      "epoch": 0.7050226905622644,
+      "learning_rate": 0.0005992841566135002,
+      "loss": 1.0862,
+      "step": 9166
+    },
+    {
+      "epoch": 0.7050996077224829,
+      "learning_rate": 0.0005989943418648457,
+      "loss": 1.1619,
+      "step": 9167
+    },
+    {
+      "epoch": 0.7051765248827013,
+      "learning_rate": 0.0005987045797268508,
+      "loss": 1.0624,
+      "step": 9168
+    },
+    {
+      "epoch": 0.7052534420429197,
+      "learning_rate": 0.0005984148702164346,
+      "loss": 1.1822,
+      "step": 9169
+    },
+    {
+      "epoch": 0.7053303592031382,
+      "learning_rate": 0.0005981252133505134,
+      "loss": 1.2507,
+      "step": 9170
+    },
+    {
+      "epoch": 0.7054072763633567,
+      "learning_rate": 0.0005978356091460011,
+      "loss": 1.0611,
+      "step": 9171
+    },
+    {
+      "epoch": 0.7054841935235752,
+      "learning_rate": 0.0005975460576198081,
+      "loss": 1.2883,
+      "step": 9172
+    },
+    {
+      "epoch": 0.7055611106837936,
+      "learning_rate": 0.0005972565587888413,
+      "loss": 1.1024,
+      "step": 9173
+    },
+    {
+      "epoch": 0.705638027844012,
+      "learning_rate": 0.0005969671126700047,
+      "loss": 0.968,
+      "step": 9174
+    },
+    {
+      "epoch": 0.7057149450042305,
+      "learning_rate": 0.0005966777192801991,
+      "loss": 1.2252,
+      "step": 9175
+    },
+    {
+      "epoch": 0.7057918621644489,
+      "learning_rate": 0.0005963883786363239,
+      "loss": 1.1141,
+      "step": 9176
+    },
+    {
+      "epoch": 0.7058687793246673,
+      "learning_rate": 0.0005960990907552724,
+      "loss": 1.105,
+      "step": 9177
+    },
+    {
+      "epoch": 0.7059456964848858,
+      "learning_rate": 0.0005958098556539365,
+      "loss": 1.1531,
+      "step": 9178
+    },
+    {
+      "epoch": 0.7060226136451042,
+      "learning_rate": 0.000595520673349206,
+      "loss": 1.1202,
+      "step": 9179
+    },
+    {
+      "epoch": 0.7060995308053226,
+      "learning_rate": 0.000595231543857966,
+      "loss": 1.0094,
+      "step": 9180
+    },
+    {
+      "epoch": 0.7061764479655411,
+      "learning_rate": 0.0005949424671970991,
+      "loss": 1.0983,
+      "step": 9181
+    },
+    {
+      "epoch": 0.7062533651257595,
+      "learning_rate": 0.000594653443383485,
+      "loss": 1.1388,
+      "step": 9182
+    },
+    {
+      "epoch": 0.706330282285978,
+      "learning_rate": 0.000594364472433999,
+      "loss": 1.2372,
+      "step": 9183
+    },
+    {
+      "epoch": 0.7064071994461965,
+      "learning_rate": 0.0005940755543655161,
+      "loss": 1.3763,
+      "step": 9184
+    },
+    {
+      "epoch": 0.7064841166064149,
+      "learning_rate": 0.0005937866891949054,
+      "loss": 1.2527,
+      "step": 9185
+    },
+    {
+      "epoch": 0.7065610337666334,
+      "learning_rate": 0.0005934978769390345,
+      "loss": 0.9266,
+      "step": 9186
+    },
+    {
+      "epoch": 0.7066379509268518,
+      "learning_rate": 0.0005932091176147674,
+      "loss": 1.3706,
+      "step": 9187
+    },
+    {
+      "epoch": 0.7067148680870702,
+      "learning_rate": 0.0005929204112389648,
+      "loss": 1.2825,
+      "step": 9188
+    },
+    {
+      "epoch": 0.7067917852472887,
+      "learning_rate": 0.0005926317578284844,
+      "loss": 1.0201,
+      "step": 9189
+    },
+    {
+      "epoch": 0.7068687024075071,
+      "learning_rate": 0.0005923431574001821,
+      "loss": 0.7452,
+      "step": 9190
+    },
+    {
+      "epoch": 0.7069456195677256,
+      "learning_rate": 0.0005920546099709077,
+      "loss": 1.5096,
+      "step": 9191
+    },
+    {
+      "epoch": 0.707022536727944,
+      "learning_rate": 0.0005917661155575116,
+      "loss": 1.0856,
+      "step": 9192
+    },
+    {
+      "epoch": 0.7070994538881624,
+      "learning_rate": 0.0005914776741768385,
+      "loss": 0.4657,
+      "step": 9193
+    },
+    {
+      "epoch": 0.7071763710483809,
+      "learning_rate": 0.0005911892858457309,
+      "loss": 1.2545,
+      "step": 9194
+    },
+    {
+      "epoch": 0.7072532882085993,
+      "learning_rate": 0.000590900950581028,
+      "loss": 1.0548,
+      "step": 9195
+    },
+    {
+      "epoch": 0.7073302053688177,
+      "learning_rate": 0.000590612668399566,
+      "loss": 1.1633,
+      "step": 9196
+    },
+    {
+      "epoch": 0.7074071225290363,
+      "learning_rate": 0.0005903244393181774,
+      "loss": 1.0331,
+      "step": 9197
+    },
+    {
+      "epoch": 0.7074840396892547,
+      "learning_rate": 0.0005900362633536934,
+      "loss": 1.3973,
+      "step": 9198
+    },
+    {
+      "epoch": 0.7075609568494731,
+      "learning_rate": 0.000589748140522941,
+      "loss": 1.3848,
+      "step": 9199
+    },
+    {
+      "epoch": 0.7076378740096916,
+      "learning_rate": 0.0005894600708427421,
+      "loss": 1.1831,
+      "step": 9200
+    },
+    {
+      "epoch": 0.70771479116991,
+      "learning_rate": 0.0005891720543299193,
+      "loss": 1.0076,
+      "step": 9201
+    },
+    {
+      "epoch": 0.7077917083301285,
+      "learning_rate": 0.0005888840910012887,
+      "loss": 1.3635,
+      "step": 9202
+    },
+    {
+      "epoch": 0.7078686254903469,
+      "learning_rate": 0.0005885961808736668,
+      "loss": 1.1163,
+      "step": 9203
+    },
+    {
+      "epoch": 0.7079455426505653,
+      "learning_rate": 0.0005883083239638632,
+      "loss": 1.1741,
+      "step": 9204
+    },
+    {
+      "epoch": 0.7080224598107838,
+      "learning_rate": 0.0005880205202886863,
+      "loss": 1.1505,
+      "step": 9205
+    },
+    {
+      "epoch": 0.7080993769710022,
+      "learning_rate": 0.000587732769864942,
+      "loss": 1.0987,
+      "step": 9206
+    },
+    {
+      "epoch": 0.7081762941312206,
+      "learning_rate": 0.000587445072709432,
+      "loss": 0.9515,
+      "step": 9207
+    },
+    {
+      "epoch": 0.7082532112914391,
+      "learning_rate": 0.0005871574288389555,
+      "loss": 1.1378,
+      "step": 9208
+    },
+    {
+      "epoch": 0.7083301284516575,
+      "learning_rate": 0.0005868698382703078,
+      "loss": 1.099,
+      "step": 9209
+    },
+    {
+      "epoch": 0.7084070456118761,
+      "learning_rate": 0.0005865823010202815,
+      "loss": 1.2283,
+      "step": 9210
+    },
+    {
+      "epoch": 0.7084839627720945,
+      "learning_rate": 0.000586294817105667,
+      "loss": 1.3589,
+      "step": 9211
+    },
+    {
+      "epoch": 0.7085608799323129,
+      "learning_rate": 0.000586007386543251,
+      "loss": 1.1269,
+      "step": 9212
+    },
+    {
+      "epoch": 0.7086377970925314,
+      "learning_rate": 0.0005857200093498152,
+      "loss": 1.2875,
+      "step": 9213
+    },
+    {
+      "epoch": 0.7087147142527498,
+      "learning_rate": 0.0005854326855421414,
+      "loss": 1.111,
+      "step": 9214
+    },
+    {
+      "epoch": 0.7087916314129682,
+      "learning_rate": 0.0005851454151370061,
+      "loss": 1.3645,
+      "step": 9215
+    },
+    {
+      "epoch": 0.7088685485731867,
+      "learning_rate": 0.0005848581981511837,
+      "loss": 1.1044,
+      "step": 9216
+    },
+    {
+      "epoch": 0.7089454657334051,
+      "learning_rate": 0.0005845710346014445,
+      "loss": 1.2185,
+      "step": 9217
+    },
+    {
+      "epoch": 0.7090223828936235,
+      "learning_rate": 0.000584283924504557,
+      "loss": 1.1358,
+      "step": 9218
+    },
+    {
+      "epoch": 0.709099300053842,
+      "learning_rate": 0.0005839968678772846,
+      "loss": 1.1562,
+      "step": 9219
+    },
+    {
+      "epoch": 0.7091762172140604,
+      "learning_rate": 0.0005837098647363904,
+      "loss": 0.9352,
+      "step": 9220
+    },
+    {
+      "epoch": 0.709253134374279,
+      "learning_rate": 0.0005834229150986323,
+      "loss": 0.9012,
+      "step": 9221
+    },
+    {
+      "epoch": 0.7093300515344974,
+      "learning_rate": 0.0005831360189807653,
+      "loss": 0.9845,
+      "step": 9222
+    },
+    {
+      "epoch": 0.7094069686947158,
+      "learning_rate": 0.0005828491763995417,
+      "loss": 0.8973,
+      "step": 9223
+    },
+    {
+      "epoch": 0.7094838858549343,
+      "learning_rate": 0.0005825623873717098,
+      "loss": 1.1538,
+      "step": 9224
+    },
+    {
+      "epoch": 0.7095608030151527,
+      "learning_rate": 0.0005822756519140176,
+      "loss": 1.1007,
+      "step": 9225
+    },
+    {
+      "epoch": 0.7096377201753711,
+      "learning_rate": 0.0005819889700432058,
+      "loss": 1.1432,
+      "step": 9226
+    },
+    {
+      "epoch": 0.7097146373355896,
+      "learning_rate": 0.0005817023417760142,
+      "loss": 1.287,
+      "step": 9227
+    },
+    {
+      "epoch": 0.709791554495808,
+      "learning_rate": 0.0005814157671291805,
+      "loss": 1.0971,
+      "step": 9228
+    },
+    {
+      "epoch": 0.7098684716560265,
+      "learning_rate": 0.0005811292461194375,
+      "loss": 1.3493,
+      "step": 9229
+    },
+    {
+      "epoch": 0.7099453888162449,
+      "learning_rate": 0.0005808427787635154,
+      "loss": 0.7992,
+      "step": 9230
+    },
+    {
+      "epoch": 0.7100223059764633,
+      "learning_rate": 0.0005805563650781414,
+      "loss": 1.1275,
+      "step": 9231
+    },
+    {
+      "epoch": 0.7100992231366818,
+      "learning_rate": 0.000580270005080039,
+      "loss": 1.2756,
+      "step": 9232
+    },
+    {
+      "epoch": 0.7101761402969002,
+      "learning_rate": 0.00057998369878593,
+      "loss": 0.9373,
+      "step": 9233
+    },
+    {
+      "epoch": 0.7102530574571186,
+      "learning_rate": 0.0005796974462125321,
+      "loss": 1.2198,
+      "step": 9234
+    },
+    {
+      "epoch": 0.7103299746173372,
+      "learning_rate": 0.0005794112473765584,
+      "loss": 0.8973,
+      "step": 9235
+    },
+    {
+      "epoch": 0.7104068917775556,
+      "learning_rate": 0.0005791251022947218,
+      "loss": 0.9384,
+      "step": 9236
+    },
+    {
+      "epoch": 0.710483808937774,
+      "learning_rate": 0.0005788390109837305,
+      "loss": 1.0546,
+      "step": 9237
+    },
+    {
+      "epoch": 0.7105607260979925,
+      "learning_rate": 0.000578552973460289,
+      "loss": 1.2324,
+      "step": 9238
+    },
+    {
+      "epoch": 0.7106376432582109,
+      "learning_rate": 0.0005782669897411001,
+      "loss": 1.47,
+      "step": 9239
+    },
+    {
+      "epoch": 0.7107145604184294,
+      "learning_rate": 0.0005779810598428614,
+      "loss": 1.1954,
+      "step": 9240
+    },
+    {
+      "epoch": 0.7107914775786478,
+      "learning_rate": 0.0005776951837822702,
+      "loss": 0.8435,
+      "step": 9241
+    },
+    {
+      "epoch": 0.7108683947388662,
+      "learning_rate": 0.0005774093615760184,
+      "loss": 0.9694,
+      "step": 9242
+    },
+    {
+      "epoch": 0.7109453118990847,
+      "learning_rate": 0.0005771235932407955,
+      "loss": 1.1737,
+      "step": 9243
+    },
+    {
+      "epoch": 0.7110222290593031,
+      "learning_rate": 0.0005768378787932878,
+      "loss": 1.1425,
+      "step": 9244
+    },
+    {
+      "epoch": 0.7110991462195215,
+      "learning_rate": 0.0005765522182501784,
+      "loss": 1.056,
+      "step": 9245
+    },
+    {
+      "epoch": 0.71117606337974,
+      "learning_rate": 0.0005762666116281469,
+      "loss": 0.7269,
+      "step": 9246
+    },
+    {
+      "epoch": 0.7112529805399584,
+      "learning_rate": 0.0005759810589438715,
+      "loss": 1.2509,
+      "step": 9247
+    },
+    {
+      "epoch": 0.711329897700177,
+      "learning_rate": 0.0005756955602140242,
+      "loss": 1.0475,
+      "step": 9248
+    },
+    {
+      "epoch": 0.7114068148603954,
+      "learning_rate": 0.0005754101154552768,
+      "loss": 1.4002,
+      "step": 9249
+    },
+    {
+      "epoch": 0.7114837320206138,
+      "learning_rate": 0.0005751247246842965,
+      "loss": 1.3878,
+      "step": 9250
+    },
+    {
+      "epoch": 0.7115606491808323,
+      "learning_rate": 0.0005748393879177474,
+      "loss": 0.8607,
+      "step": 9251
+    },
+    {
+      "epoch": 0.7116375663410507,
+      "learning_rate": 0.0005745541051722906,
+      "loss": 0.9819,
+      "step": 9252
+    },
+    {
+      "epoch": 0.7117144835012691,
+      "learning_rate": 0.000574268876464584,
+      "loss": 0.9967,
+      "step": 9253
+    },
+    {
+      "epoch": 0.7117914006614876,
+      "learning_rate": 0.0005739837018112821,
+      "loss": 1.0851,
+      "step": 9254
+    },
+    {
+      "epoch": 0.711868317821706,
+      "learning_rate": 0.0005736985812290372,
+      "loss": 1.2694,
+      "step": 9255
+    },
+    {
+      "epoch": 0.7119452349819244,
+      "learning_rate": 0.0005734135147344983,
+      "loss": 1.0488,
+      "step": 9256
+    },
+    {
+      "epoch": 0.7120221521421429,
+      "learning_rate": 0.0005731285023443087,
+      "loss": 1.1041,
+      "step": 9257
+    },
+    {
+      "epoch": 0.7120990693023613,
+      "learning_rate": 0.0005728435440751125,
+      "loss": 1.4289,
+      "step": 9258
+    },
+    {
+      "epoch": 0.7121759864625798,
+      "learning_rate": 0.0005725586399435473,
+      "loss": 1.0462,
+      "step": 9259
+    },
+    {
+      "epoch": 0.7122529036227983,
+      "learning_rate": 0.0005722737899662509,
+      "loss": 0.9973,
+      "step": 9260
+    },
+    {
+      "epoch": 0.7123298207830167,
+      "learning_rate": 0.0005719889941598542,
+      "loss": 1.0765,
+      "step": 9261
+    },
+    {
+      "epoch": 0.7124067379432352,
+      "learning_rate": 0.0005717042525409866,
+      "loss": 1.0844,
+      "step": 9262
+    },
+    {
+      "epoch": 0.7124836551034536,
+      "learning_rate": 0.0005714195651262758,
+      "loss": 1.1386,
+      "step": 9263
+    },
+    {
+      "epoch": 0.712560572263672,
+      "learning_rate": 0.0005711349319323445,
+      "loss": 0.929,
+      "step": 9264
+    },
+    {
+      "epoch": 0.7126374894238905,
+      "learning_rate": 0.0005708503529758123,
+      "loss": 1.2356,
+      "step": 9265
+    },
+    {
+      "epoch": 0.7127144065841089,
+      "learning_rate": 0.0005705658282732963,
+      "loss": 0.8556,
+      "step": 9266
+    },
+    {
+      "epoch": 0.7127913237443274,
+      "learning_rate": 0.0005702813578414098,
+      "loss": 0.928,
+      "step": 9267
+    },
+    {
+      "epoch": 0.7128682409045458,
+      "learning_rate": 0.0005699969416967643,
+      "loss": 0.98,
+      "step": 9268
+    },
+    {
+      "epoch": 0.7129451580647642,
+      "learning_rate": 0.0005697125798559672,
+      "loss": 1.6185,
+      "step": 9269
+    },
+    {
+      "epoch": 0.7130220752249827,
+      "learning_rate": 0.0005694282723356209,
+      "loss": 1.1985,
+      "step": 9270
+    },
+    {
+      "epoch": 0.7130989923852011,
+      "learning_rate": 0.0005691440191523282,
+      "loss": 1.3479,
+      "step": 9271
+    },
+    {
+      "epoch": 0.7131759095454195,
+      "learning_rate": 0.0005688598203226862,
+      "loss": 1.0522,
+      "step": 9272
+    },
+    {
+      "epoch": 0.713252826705638,
+      "learning_rate": 0.0005685756758632898,
+      "loss": 1.0467,
+      "step": 9273
+    },
+    {
+      "epoch": 0.7133297438658565,
+      "learning_rate": 0.0005682915857907303,
+      "loss": 1.2088,
+      "step": 9274
+    },
+    {
+      "epoch": 0.7134066610260749,
+      "learning_rate": 0.0005680075501215963,
+      "loss": 1.1078,
+      "step": 9275
+    },
+    {
+      "epoch": 0.7134835781862934,
+      "learning_rate": 0.0005677235688724721,
+      "loss": 1.1156,
+      "step": 9276
+    },
+    {
+      "epoch": 0.7135604953465118,
+      "learning_rate": 0.0005674396420599408,
+      "loss": 1.2745,
+      "step": 9277
+    },
+    {
+      "epoch": 0.7136374125067303,
+      "learning_rate": 0.0005671557697005808,
+      "loss": 1.3483,
+      "step": 9278
+    },
+    {
+      "epoch": 0.7137143296669487,
+      "learning_rate": 0.0005668719518109675,
+      "loss": 1.489,
+      "step": 9279
+    },
+    {
+      "epoch": 0.7137912468271671,
+      "learning_rate": 0.0005665881884076735,
+      "loss": 1.2943,
+      "step": 9280
+    },
+    {
+      "epoch": 0.7138681639873856,
+      "learning_rate": 0.0005663044795072673,
+      "loss": 1.2696,
+      "step": 9281
+    },
+    {
+      "epoch": 0.713945081147604,
+      "learning_rate": 0.0005660208251263169,
+      "loss": 1.3986,
+      "step": 9282
+    },
+    {
+      "epoch": 0.7140219983078224,
+      "learning_rate": 0.0005657372252813833,
+      "loss": 1.054,
+      "step": 9283
+    },
+    {
+      "epoch": 0.7140989154680409,
+      "learning_rate": 0.0005654536799890261,
+      "loss": 1.4222,
+      "step": 9284
+    },
+    {
+      "epoch": 0.7141758326282593,
+      "learning_rate": 0.000565170189265803,
+      "loss": 0.6641,
+      "step": 9285
+    },
+    {
+      "epoch": 0.7142527497884779,
+      "learning_rate": 0.0005648867531282669,
+      "loss": 0.8956,
+      "step": 9286
+    },
+    {
+      "epoch": 0.7143296669486963,
+      "learning_rate": 0.0005646033715929678,
+      "loss": 0.9189,
+      "step": 9287
+    },
+    {
+      "epoch": 0.7144065841089147,
+      "learning_rate": 0.0005643200446764526,
+      "loss": 1.2879,
+      "step": 9288
+    },
+    {
+      "epoch": 0.7144835012691332,
+      "learning_rate": 0.0005640367723952648,
+      "loss": 1.2763,
+      "step": 9289
+    },
+    {
+      "epoch": 0.7145604184293516,
+      "learning_rate": 0.0005637535547659458,
+      "loss": 1.0203,
+      "step": 9290
+    },
+    {
+      "epoch": 0.71463733558957,
+      "learning_rate": 0.0005634703918050329,
+      "loss": 1.4123,
+      "step": 9291
+    },
+    {
+      "epoch": 0.7147142527497885,
+      "learning_rate": 0.0005631872835290589,
+      "loss": 0.8268,
+      "step": 9292
+    },
+    {
+      "epoch": 0.7147911699100069,
+      "learning_rate": 0.0005629042299545561,
+      "loss": 1.198,
+      "step": 9293
+    },
+    {
+      "epoch": 0.7148680870702254,
+      "learning_rate": 0.0005626212310980514,
+      "loss": 1.0468,
+      "step": 9294
+    },
+    {
+      "epoch": 0.7149450042304438,
+      "learning_rate": 0.0005623382869760713,
+      "loss": 1.1087,
+      "step": 9295
+    },
+    {
+      "epoch": 0.7150219213906622,
+      "learning_rate": 0.0005620553976051351,
+      "loss": 1.0472,
+      "step": 9296
+    },
+    {
+      "epoch": 0.7150988385508807,
+      "learning_rate": 0.0005617725630017611,
+      "loss": 1.1679,
+      "step": 9297
+    },
+    {
+      "epoch": 0.7151757557110991,
+      "learning_rate": 0.0005614897831824656,
+      "loss": 1.5574,
+      "step": 9298
+    },
+    {
+      "epoch": 0.7152526728713176,
+      "learning_rate": 0.0005612070581637596,
+      "loss": 1.0795,
+      "step": 9299
+    },
+    {
+      "epoch": 0.7153295900315361,
+      "learning_rate": 0.0005609243879621521,
+      "loss": 1.1924,
+      "step": 9300
+    },
+    {
+      "epoch": 0.7154065071917545,
+      "learning_rate": 0.0005606417725941481,
+      "loss": 1.1555,
+      "step": 9301
+    },
+    {
+      "epoch": 0.7154834243519729,
+      "learning_rate": 0.0005603592120762501,
+      "loss": 1.1698,
+      "step": 9302
+    },
+    {
+      "epoch": 0.7155603415121914,
+      "learning_rate": 0.0005600767064249562,
+      "loss": 0.9323,
+      "step": 9303
+    },
+    {
+      "epoch": 0.7156372586724098,
+      "learning_rate": 0.0005597942556567643,
+      "loss": 1.3894,
+      "step": 9304
+    },
+    {
+      "epoch": 0.7157141758326283,
+      "learning_rate": 0.0005595118597881646,
+      "loss": 1.0572,
+      "step": 9305
+    },
+    {
+      "epoch": 0.7157910929928467,
+      "learning_rate": 0.0005592295188356479,
+      "loss": 1.1573,
+      "step": 9306
+    },
+    {
+      "epoch": 0.7158680101530651,
+      "learning_rate": 0.0005589472328157002,
+      "loss": 1.0623,
+      "step": 9307
+    },
+    {
+      "epoch": 0.7159449273132836,
+      "learning_rate": 0.0005586650017448041,
+      "loss": 1.0767,
+      "step": 9308
+    },
+    {
+      "epoch": 0.716021844473502,
+      "learning_rate": 0.0005583828256394399,
+      "loss": 0.9451,
+      "step": 9309
+    },
+    {
+      "epoch": 0.7160987616337204,
+      "learning_rate": 0.0005581007045160837,
+      "loss": 1.0433,
+      "step": 9310
+    },
+    {
+      "epoch": 0.716175678793939,
+      "learning_rate": 0.0005578186383912085,
+      "loss": 0.8603,
+      "step": 9311
+    },
+    {
+      "epoch": 0.7162525959541574,
+      "learning_rate": 0.0005575366272812854,
+      "loss": 1.2741,
+      "step": 9312
+    },
+    {
+      "epoch": 0.7163295131143759,
+      "learning_rate": 0.0005572546712027808,
+      "loss": 1.3708,
+      "step": 9313
+    },
+    {
+      "epoch": 0.7164064302745943,
+      "learning_rate": 0.0005569727701721584,
+      "loss": 1.356,
+      "step": 9314
+    },
+    {
+      "epoch": 0.7164833474348127,
+      "learning_rate": 0.0005566909242058789,
+      "loss": 1.1124,
+      "step": 9315
+    },
+    {
+      "epoch": 0.7165602645950312,
+      "learning_rate": 0.0005564091333203988,
+      "loss": 1.0062,
+      "step": 9316
+    },
+    {
+      "epoch": 0.7166371817552496,
+      "learning_rate": 0.000556127397532174,
+      "loss": 1.0108,
+      "step": 9317
+    },
+    {
+      "epoch": 0.716714098915468,
+      "learning_rate": 0.0005558457168576534,
+      "loss": 1.0154,
+      "step": 9318
+    },
+    {
+      "epoch": 0.7167910160756865,
+      "learning_rate": 0.000555564091313285,
+      "loss": 1.1735,
+      "step": 9319
+    },
+    {
+      "epoch": 0.7168679332359049,
+      "learning_rate": 0.000555282520915514,
+      "loss": 0.9607,
+      "step": 9320
+    },
+    {
+      "epoch": 0.7169448503961233,
+      "learning_rate": 0.0005550010056807813,
+      "loss": 1.3898,
+      "step": 9321
+    },
+    {
+      "epoch": 0.7170217675563418,
+      "learning_rate": 0.0005547195456255246,
+      "loss": 1.3058,
+      "step": 9322
+    },
+    {
+      "epoch": 0.7170986847165602,
+      "learning_rate": 0.0005544381407661787,
+      "loss": 1.3691,
+      "step": 9323
+    },
+    {
+      "epoch": 0.7171756018767788,
+      "learning_rate": 0.0005541567911191749,
+      "loss": 1.0796,
+      "step": 9324
+    },
+    {
+      "epoch": 0.7172525190369972,
+      "learning_rate": 0.0005538754967009423,
+      "loss": 1.2049,
+      "step": 9325
+    },
+    {
+      "epoch": 0.7173294361972156,
+      "learning_rate": 0.0005535942575279053,
+      "loss": 1.0439,
+      "step": 9326
+    },
+    {
+      "epoch": 0.7174063533574341,
+      "learning_rate": 0.0005533130736164861,
+      "loss": 1.515,
+      "step": 9327
+    },
+    {
+      "epoch": 0.7174832705176525,
+      "learning_rate": 0.000553031944983103,
+      "loss": 0.829,
+      "step": 9328
+    },
+    {
+      "epoch": 0.7175601876778709,
+      "learning_rate": 0.0005527508716441716,
+      "loss": 1.0455,
+      "step": 9329
+    },
+    {
+      "epoch": 0.7176371048380894,
+      "learning_rate": 0.0005524698536161041,
+      "loss": 1.173,
+      "step": 9330
+    },
+    {
+      "epoch": 0.7177140219983078,
+      "learning_rate": 0.0005521888909153092,
+      "loss": 0.8212,
+      "step": 9331
+    },
+    {
+      "epoch": 0.7177909391585263,
+      "learning_rate": 0.0005519079835581924,
+      "loss": 1.1177,
+      "step": 9332
+    },
+    {
+      "epoch": 0.7178678563187447,
+      "learning_rate": 0.0005516271315611568,
+      "loss": 1.0467,
+      "step": 9333
+    },
+    {
+      "epoch": 0.7179447734789631,
+      "learning_rate": 0.0005513463349406015,
+      "loss": 1.4146,
+      "step": 9334
+    },
+    {
+      "epoch": 0.7180216906391816,
+      "learning_rate": 0.0005510655937129224,
+      "loss": 0.9482,
+      "step": 9335
+    },
+    {
+      "epoch": 0.7180986077994,
+      "learning_rate": 0.0005507849078945121,
+      "loss": 1.023,
+      "step": 9336
+    },
+    {
+      "epoch": 0.7181755249596184,
+      "learning_rate": 0.0005505042775017603,
+      "loss": 0.8764,
+      "step": 9337
+    },
+    {
+      "epoch": 0.718252442119837,
+      "learning_rate": 0.0005502237025510529,
+      "loss": 1.1324,
+      "step": 9338
+    },
+    {
+      "epoch": 0.7183293592800554,
+      "learning_rate": 0.0005499431830587738,
+      "loss": 0.9934,
+      "step": 9339
+    },
+    {
+      "epoch": 0.7184062764402738,
+      "learning_rate": 0.0005496627190413028,
+      "loss": 1.209,
+      "step": 9340
+    },
+    {
+      "epoch": 0.7184831936004923,
+      "learning_rate": 0.0005493823105150151,
+      "loss": 1.2592,
+      "step": 9341
+    },
+    {
+      "epoch": 0.7185601107607107,
+      "learning_rate": 0.0005491019574962856,
+      "loss": 1.0501,
+      "step": 9342
+    },
+    {
+      "epoch": 0.7186370279209292,
+      "learning_rate": 0.0005488216600014836,
+      "loss": 1.2749,
+      "step": 9343
+    },
+    {
+      "epoch": 0.7187139450811476,
+      "learning_rate": 0.0005485414180469762,
+      "loss": 1.1647,
+      "step": 9344
+    },
+    {
+      "epoch": 0.718790862241366,
+      "learning_rate": 0.000548261231649127,
+      "loss": 0.8373,
+      "step": 9345
+    },
+    {
+      "epoch": 0.7188677794015845,
+      "learning_rate": 0.0005479811008242961,
+      "loss": 0.9246,
+      "step": 9346
+    },
+    {
+      "epoch": 0.7189446965618029,
+      "learning_rate": 0.0005477010255888412,
+      "loss": 1.1222,
+      "step": 9347
+    },
+    {
+      "epoch": 0.7190216137220213,
+      "learning_rate": 0.0005474210059591158,
+      "loss": 1.0607,
+      "step": 9348
+    },
+    {
+      "epoch": 0.7190985308822399,
+      "learning_rate": 0.0005471410419514707,
+      "loss": 1.3698,
+      "step": 9349
+    },
+    {
+      "epoch": 0.7191754480424583,
+      "learning_rate": 0.0005468611335822532,
+      "loss": 1.5374,
+      "step": 9350
+    },
+    {
+      "epoch": 0.7192523652026768,
+      "learning_rate": 0.000546581280867807,
+      "loss": 1.2769,
+      "step": 9351
+    },
+    {
+      "epoch": 0.7193292823628952,
+      "learning_rate": 0.0005463014838244738,
+      "loss": 1.0591,
+      "step": 9352
+    },
+    {
+      "epoch": 0.7194061995231136,
+      "learning_rate": 0.0005460217424685915,
+      "loss": 1.668,
+      "step": 9353
+    },
+    {
+      "epoch": 0.7194831166833321,
+      "learning_rate": 0.0005457420568164928,
+      "loss": 1.0583,
+      "step": 9354
+    },
+    {
+      "epoch": 0.7195600338435505,
+      "learning_rate": 0.0005454624268845105,
+      "loss": 0.8462,
+      "step": 9355
+    },
+    {
+      "epoch": 0.7196369510037689,
+      "learning_rate": 0.0005451828526889718,
+      "loss": 1.3111,
+      "step": 9356
+    },
+    {
+      "epoch": 0.7197138681639874,
+      "learning_rate": 0.0005449033342462015,
+      "loss": 1.2163,
+      "step": 9357
+    },
+    {
+      "epoch": 0.7197907853242058,
+      "learning_rate": 0.0005446238715725209,
+      "loss": 1.119,
+      "step": 9358
+    },
+    {
+      "epoch": 0.7198677024844242,
+      "learning_rate": 0.0005443444646842479,
+      "loss": 1.292,
+      "step": 9359
+    },
+    {
+      "epoch": 0.7199446196446427,
+      "learning_rate": 0.0005440651135976973,
+      "loss": 1.3125,
+      "step": 9360
+    },
+    {
+      "epoch": 0.7200215368048611,
+      "learning_rate": 0.0005437858183291814,
+      "loss": 1.2528,
+      "step": 9361
+    },
+    {
+      "epoch": 0.7200984539650797,
+      "learning_rate": 0.000543506578895008,
+      "loss": 0.8609,
+      "step": 9362
+    },
+    {
+      "epoch": 0.7201753711252981,
+      "learning_rate": 0.0005432273953114825,
+      "loss": 1.1988,
+      "step": 9363
+    },
+    {
+      "epoch": 0.7202522882855165,
+      "learning_rate": 0.0005429482675949063,
+      "loss": 1.0856,
+      "step": 9364
+    },
+    {
+      "epoch": 0.720329205445735,
+      "learning_rate": 0.0005426691957615778,
+      "loss": 1.2129,
+      "step": 9365
+    },
+    {
+      "epoch": 0.7204061226059534,
+      "learning_rate": 0.0005423901798277936,
+      "loss": 1.3119,
+      "step": 9366
+    },
+    {
+      "epoch": 0.7204830397661718,
+      "learning_rate": 0.0005421112198098444,
+      "loss": 1.2377,
+      "step": 9367
+    },
+    {
+      "epoch": 0.7205599569263903,
+      "learning_rate": 0.0005418323157240187,
+      "loss": 1.0323,
+      "step": 9368
+    },
+    {
+      "epoch": 0.7206368740866087,
+      "learning_rate": 0.0005415534675866033,
+      "loss": 1.3434,
+      "step": 9369
+    },
+    {
+      "epoch": 0.7207137912468272,
+      "learning_rate": 0.0005412746754138799,
+      "loss": 1.2895,
+      "step": 9370
+    },
+    {
+      "epoch": 0.7207907084070456,
+      "learning_rate": 0.0005409959392221271,
+      "loss": 0.9031,
+      "step": 9371
+    },
+    {
+      "epoch": 0.720867625567264,
+      "learning_rate": 0.000540717259027621,
+      "loss": 1.114,
+      "step": 9372
+    },
+    {
+      "epoch": 0.7209445427274825,
+      "learning_rate": 0.0005404386348466333,
+      "loss": 1.2747,
+      "step": 9373
+    },
+    {
+      "epoch": 0.7210214598877009,
+      "learning_rate": 0.0005401600666954344,
+      "loss": 1.0093,
+      "step": 9374
+    },
+    {
+      "epoch": 0.7210983770479193,
+      "learning_rate": 0.0005398815545902899,
+      "loss": 1.6238,
+      "step": 9375
+    },
+    {
+      "epoch": 0.7211752942081379,
+      "learning_rate": 0.0005396030985474611,
+      "loss": 1.0713,
+      "step": 9376
+    },
+    {
+      "epoch": 0.7212522113683563,
+      "learning_rate": 0.0005393246985832087,
+      "loss": 0.9556,
+      "step": 9377
+    },
+    {
+      "epoch": 0.7213291285285747,
+      "learning_rate": 0.0005390463547137883,
+      "loss": 0.9452,
+      "step": 9378
+    },
+    {
+      "epoch": 0.7214060456887932,
+      "learning_rate": 0.000538768066955453,
+      "loss": 1.1993,
+      "step": 9379
+    },
+    {
+      "epoch": 0.7214829628490116,
+      "learning_rate": 0.0005384898353244517,
+      "loss": 1.2313,
+      "step": 9380
+    },
+    {
+      "epoch": 0.7215598800092301,
+      "learning_rate": 0.0005382116598370307,
+      "loss": 1.1112,
+      "step": 9381
+    },
+    {
+      "epoch": 0.7216367971694485,
+      "learning_rate": 0.0005379335405094337,
+      "loss": 1.2084,
+      "step": 9382
+    },
+    {
+      "epoch": 0.7217137143296669,
+      "learning_rate": 0.0005376554773578998,
+      "loss": 0.8355,
+      "step": 9383
+    },
+    {
+      "epoch": 0.7217906314898854,
+      "learning_rate": 0.0005373774703986658,
+      "loss": 1.0795,
+      "step": 9384
+    },
+    {
+      "epoch": 0.7218675486501038,
+      "learning_rate": 0.0005370995196479642,
+      "loss": 1.1485,
+      "step": 9385
+    },
+    {
+      "epoch": 0.7219444658103222,
+      "learning_rate": 0.0005368216251220253,
+      "loss": 1.2273,
+      "step": 9386
+    },
+    {
+      "epoch": 0.7220213829705407,
+      "learning_rate": 0.0005365437868370751,
+      "loss": 1.2547,
+      "step": 9387
+    },
+    {
+      "epoch": 0.7220983001307592,
+      "learning_rate": 0.000536266004809338,
+      "loss": 1.2919,
+      "step": 9388
+    },
+    {
+      "epoch": 0.7221752172909777,
+      "learning_rate": 0.0005359882790550325,
+      "loss": 1.0553,
+      "step": 9389
+    },
+    {
+      "epoch": 0.7222521344511961,
+      "learning_rate": 0.0005357106095903763,
+      "loss": 1.1381,
+      "step": 9390
+    },
+    {
+      "epoch": 0.7223290516114145,
+      "learning_rate": 0.0005354329964315829,
+      "loss": 1.1194,
+      "step": 9391
+    },
+    {
+      "epoch": 0.722405968771633,
+      "learning_rate": 0.0005351554395948619,
+      "loss": 1.0312,
+      "step": 9392
+    },
+    {
+      "epoch": 0.7224828859318514,
+      "learning_rate": 0.0005348779390964203,
+      "loss": 0.7245,
+      "step": 9393
+    },
+    {
+      "epoch": 0.7225598030920698,
+      "learning_rate": 0.0005346004949524618,
+      "loss": 1.0681,
+      "step": 9394
+    },
+    {
+      "epoch": 0.7226367202522883,
+      "learning_rate": 0.0005343231071791858,
+      "loss": 1.0064,
+      "step": 9395
+    },
+    {
+      "epoch": 0.7227136374125067,
+      "learning_rate": 0.0005340457757927906,
+      "loss": 1.0391,
+      "step": 9396
+    },
+    {
+      "epoch": 0.7227905545727252,
+      "learning_rate": 0.0005337685008094694,
+      "loss": 0.9042,
+      "step": 9397
+    },
+    {
+      "epoch": 0.7228674717329436,
+      "learning_rate": 0.0005334912822454121,
+      "loss": 1.4803,
+      "step": 9398
+    },
+    {
+      "epoch": 0.722944388893162,
+      "learning_rate": 0.0005332141201168063,
+      "loss": 1.2519,
+      "step": 9399
+    },
+    {
+      "epoch": 0.7230213060533806,
+      "learning_rate": 0.0005329370144398351,
+      "loss": 1.1931,
+      "step": 9400
+    },
+    {
+      "epoch": 0.723098223213599,
+      "learning_rate": 0.0005326599652306804,
+      "loss": 1.1531,
+      "step": 9401
+    },
+    {
+      "epoch": 0.7231751403738174,
+      "learning_rate": 0.0005323829725055181,
+      "loss": 1.4472,
+      "step": 9402
+    },
+    {
+      "epoch": 0.7232520575340359,
+      "learning_rate": 0.0005321060362805217,
+      "loss": 0.9079,
+      "step": 9403
+    },
+    {
+      "epoch": 0.7233289746942543,
+      "learning_rate": 0.0005318291565718633,
+      "loss": 1.2469,
+      "step": 9404
+    },
+    {
+      "epoch": 0.7234058918544727,
+      "learning_rate": 0.0005315523333957094,
+      "loss": 1.1492,
+      "step": 9405
+    },
+    {
+      "epoch": 0.7234828090146912,
+      "learning_rate": 0.0005312755667682239,
+      "loss": 0.8142,
+      "step": 9406
+    },
+    {
+      "epoch": 0.7235597261749096,
+      "learning_rate": 0.0005309988567055679,
+      "loss": 1.0987,
+      "step": 9407
+    },
+    {
+      "epoch": 0.7236366433351281,
+      "learning_rate": 0.0005307222032238979,
+      "loss": 1.2596,
+      "step": 9408
+    },
+    {
+      "epoch": 0.7237135604953465,
+      "learning_rate": 0.0005304456063393691,
+      "loss": 1.2135,
+      "step": 9409
+    },
+    {
+      "epoch": 0.7237904776555649,
+      "learning_rate": 0.0005301690660681322,
+      "loss": 1.2055,
+      "step": 9410
+    },
+    {
+      "epoch": 0.7238673948157834,
+      "learning_rate": 0.0005298925824263333,
+      "loss": 1.2809,
+      "step": 9411
+    },
+    {
+      "epoch": 0.7239443119760018,
+      "learning_rate": 0.0005296161554301181,
+      "loss": 1.1521,
+      "step": 9412
+    },
+    {
+      "epoch": 0.7240212291362202,
+      "learning_rate": 0.0005293397850956268,
+      "loss": 1.1096,
+      "step": 9413
+    },
+    {
+      "epoch": 0.7240981462964388,
+      "learning_rate": 0.0005290634714389972,
+      "loss": 1.1432,
+      "step": 9414
+    },
+    {
+      "epoch": 0.7241750634566572,
+      "learning_rate": 0.0005287872144763633,
+      "loss": 1.1967,
+      "step": 9415
+    },
+    {
+      "epoch": 0.7242519806168757,
+      "learning_rate": 0.0005285110142238556,
+      "loss": 1.3089,
+      "step": 9416
+    },
+    {
+      "epoch": 0.7243288977770941,
+      "learning_rate": 0.0005282348706976027,
+      "loss": 0.9607,
+      "step": 9417
+    },
+    {
+      "epoch": 0.7244058149373125,
+      "learning_rate": 0.0005279587839137286,
+      "loss": 1.2065,
+      "step": 9418
+    },
+    {
+      "epoch": 0.724482732097531,
+      "learning_rate": 0.0005276827538883541,
+      "loss": 0.6844,
+      "step": 9419
+    },
+    {
+      "epoch": 0.7245596492577494,
+      "learning_rate": 0.0005274067806375968,
+      "loss": 1.1616,
+      "step": 9420
+    },
+    {
+      "epoch": 0.7246365664179678,
+      "learning_rate": 0.0005271308641775713,
+      "loss": 1.3268,
+      "step": 9421
+    },
+    {
+      "epoch": 0.7247134835781863,
+      "learning_rate": 0.000526855004524388,
+      "loss": 1.3412,
+      "step": 9422
+    },
+    {
+      "epoch": 0.7247904007384047,
+      "learning_rate": 0.0005265792016941563,
+      "loss": 1.2578,
+      "step": 9423
+    },
+    {
+      "epoch": 0.7248673178986231,
+      "learning_rate": 0.0005263034557029788,
+      "loss": 1.1353,
+      "step": 9424
+    },
+    {
+      "epoch": 0.7249442350588416,
+      "learning_rate": 0.0005260277665669569,
+      "loss": 1.0317,
+      "step": 9425
+    },
+    {
+      "epoch": 0.72502115221906,
+      "learning_rate": 0.0005257521343021893,
+      "loss": 1.2119,
+      "step": 9426
+    },
+    {
+      "epoch": 0.7250980693792786,
+      "learning_rate": 0.0005254765589247697,
+      "loss": 1.2605,
+      "step": 9427
+    },
+    {
+      "epoch": 0.725174986539497,
+      "learning_rate": 0.0005252010404507896,
+      "loss": 1.4108,
+      "step": 9428
+    },
+    {
+      "epoch": 0.7252519036997154,
+      "learning_rate": 0.0005249255788963367,
+      "loss": 1.1841,
+      "step": 9429
+    },
+    {
+      "epoch": 0.7253288208599339,
+      "learning_rate": 0.0005246501742774948,
+      "loss": 1.1969,
+      "step": 9430
+    },
+    {
+      "epoch": 0.7254057380201523,
+      "learning_rate": 0.0005243748266103462,
+      "loss": 0.8568,
+      "step": 9431
+    },
+    {
+      "epoch": 0.7254826551803707,
+      "learning_rate": 0.000524099535910969,
+      "loss": 0.8923,
+      "step": 9432
+    },
+    {
+      "epoch": 0.7255595723405892,
+      "learning_rate": 0.0005238243021954358,
+      "loss": 1.7811,
+      "step": 9433
+    },
+    {
+      "epoch": 0.7256364895008076,
+      "learning_rate": 0.0005235491254798193,
+      "loss": 1.0812,
+      "step": 9434
+    },
+    {
+      "epoch": 0.7257134066610261,
+      "learning_rate": 0.0005232740057801867,
+      "loss": 0.8363,
+      "step": 9435
+    },
+    {
+      "epoch": 0.7257903238212445,
+      "learning_rate": 0.0005229989431126037,
+      "loss": 1.0442,
+      "step": 9436
+    },
+    {
+      "epoch": 0.7258672409814629,
+      "learning_rate": 0.0005227239374931302,
+      "loss": 1.5356,
+      "step": 9437
+    },
+    {
+      "epoch": 0.7259441581416815,
+      "learning_rate": 0.0005224489889378241,
+      "loss": 1.0046,
+      "step": 9438
+    },
+    {
+      "epoch": 0.7260210753018999,
+      "learning_rate": 0.0005221740974627407,
+      "loss": 1.0041,
+      "step": 9439
+    },
+    {
+      "epoch": 0.7260979924621183,
+      "learning_rate": 0.0005218992630839307,
+      "loss": 0.9842,
+      "step": 9440
+    },
+    {
+      "epoch": 0.7261749096223368,
+      "learning_rate": 0.0005216244858174422,
+      "loss": 1.2465,
+      "step": 9441
+    },
+    {
+      "epoch": 0.7262518267825552,
+      "learning_rate": 0.0005213497656793195,
+      "loss": 1.1504,
+      "step": 9442
+    },
+    {
+      "epoch": 0.7263287439427736,
+      "learning_rate": 0.0005210751026856041,
+      "loss": 0.7373,
+      "step": 9443
+    },
+    {
+      "epoch": 0.7264056611029921,
+      "learning_rate": 0.000520800496852333,
+      "loss": 1.1089,
+      "step": 9444
+    },
+    {
+      "epoch": 0.7264825782632105,
+      "learning_rate": 0.0005205259481955424,
+      "loss": 1.1643,
+      "step": 9445
+    },
+    {
+      "epoch": 0.726559495423429,
+      "learning_rate": 0.0005202514567312616,
+      "loss": 1.0695,
+      "step": 9446
+    },
+    {
+      "epoch": 0.7266364125836474,
+      "learning_rate": 0.0005199770224755197,
+      "loss": 1.7981,
+      "step": 9447
+    },
+    {
+      "epoch": 0.7267133297438658,
+      "learning_rate": 0.0005197026454443408,
+      "loss": 0.8053,
+      "step": 9448
+    },
+    {
+      "epoch": 0.7267902469040843,
+      "learning_rate": 0.0005194283256537461,
+      "loss": 0.9412,
+      "step": 9449
+    },
+    {
+      "epoch": 0.7268671640643027,
+      "learning_rate": 0.0005191540631197534,
+      "loss": 0.9117,
+      "step": 9450
+    },
+    {
+      "epoch": 0.7269440812245211,
+      "learning_rate": 0.0005188798578583772,
+      "loss": 1.1715,
+      "step": 9451
+    },
+    {
+      "epoch": 0.7270209983847397,
+      "learning_rate": 0.000518605709885628,
+      "loss": 0.8628,
+      "step": 9452
+    },
+    {
+      "epoch": 0.7270979155449581,
+      "learning_rate": 0.0005183316192175149,
+      "loss": 1.2291,
+      "step": 9453
+    },
+    {
+      "epoch": 0.7271748327051766,
+      "learning_rate": 0.0005180575858700413,
+      "loss": 1.2169,
+      "step": 9454
+    },
+    {
+      "epoch": 0.727251749865395,
+      "learning_rate": 0.0005177836098592089,
+      "loss": 1.1948,
+      "step": 9455
+    },
+    {
+      "epoch": 0.7273286670256134,
+      "learning_rate": 0.000517509691201015,
+      "loss": 1.0877,
+      "step": 9456
+    },
+    {
+      "epoch": 0.7274055841858319,
+      "learning_rate": 0.0005172358299114538,
+      "loss": 1.093,
+      "step": 9457
+    },
+    {
+      "epoch": 0.7274825013460503,
+      "learning_rate": 0.0005169620260065175,
+      "loss": 0.9695,
+      "step": 9458
+    },
+    {
+      "epoch": 0.7275594185062687,
+      "learning_rate": 0.0005166882795021926,
+      "loss": 0.9559,
+      "step": 9459
+    },
+    {
+      "epoch": 0.7276363356664872,
+      "learning_rate": 0.0005164145904144634,
+      "loss": 1.4075,
+      "step": 9460
+    },
+    {
+      "epoch": 0.7277132528267056,
+      "learning_rate": 0.0005161409587593119,
+      "loss": 1.3804,
+      "step": 9461
+    },
+    {
+      "epoch": 0.727790169986924,
+      "learning_rate": 0.000515867384552715,
+      "loss": 0.9768,
+      "step": 9462
+    },
+    {
+      "epoch": 0.7278670871471425,
+      "learning_rate": 0.0005155938678106472,
+      "loss": 1.2198,
+      "step": 9463
+    },
+    {
+      "epoch": 0.727944004307361,
+      "learning_rate": 0.0005153204085490794,
+      "loss": 1.1318,
+      "step": 9464
+    },
+    {
+      "epoch": 0.7280209214675795,
+      "learning_rate": 0.0005150470067839787,
+      "loss": 1.1305,
+      "step": 9465
+    },
+    {
+      "epoch": 0.7280978386277979,
+      "learning_rate": 0.0005147736625313101,
+      "loss": 1.1635,
+      "step": 9466
+    },
+    {
+      "epoch": 0.7281747557880163,
+      "learning_rate": 0.0005145003758070348,
+      "loss": 1.1236,
+      "step": 9467
+    },
+    {
+      "epoch": 0.7282516729482348,
+      "learning_rate": 0.0005142271466271086,
+      "loss": 1.026,
+      "step": 9468
+    },
+    {
+      "epoch": 0.7283285901084532,
+      "learning_rate": 0.000513953975007487,
+      "loss": 0.9184,
+      "step": 9469
+    },
+    {
+      "epoch": 0.7284055072686716,
+      "learning_rate": 0.0005136808609641206,
+      "loss": 1.19,
+      "step": 9470
+    },
+    {
+      "epoch": 0.7284824244288901,
+      "learning_rate": 0.0005134078045129567,
+      "loss": 1.1412,
+      "step": 9471
+    },
+    {
+      "epoch": 0.7285593415891085,
+      "learning_rate": 0.0005131348056699391,
+      "loss": 1.1946,
+      "step": 9472
+    },
+    {
+      "epoch": 0.728636258749327,
+      "learning_rate": 0.0005128618644510083,
+      "loss": 1.1496,
+      "step": 9473
+    },
+    {
+      "epoch": 0.7287131759095454,
+      "learning_rate": 0.0005125889808721025,
+      "loss": 1.2636,
+      "step": 9474
+    },
+    {
+      "epoch": 0.7287900930697638,
+      "learning_rate": 0.0005123161549491551,
+      "loss": 1.34,
+      "step": 9475
+    },
+    {
+      "epoch": 0.7288670102299823,
+      "learning_rate": 0.0005120433866980968,
+      "loss": 1.1529,
+      "step": 9476
+    },
+    {
+      "epoch": 0.7289439273902008,
+      "learning_rate": 0.0005117706761348549,
+      "loss": 0.9917,
+      "step": 9477
+    },
+    {
+      "epoch": 0.7290208445504192,
+      "learning_rate": 0.0005114980232753529,
+      "loss": 1.0118,
+      "step": 9478
+    },
+    {
+      "epoch": 0.7290977617106377,
+      "learning_rate": 0.000511225428135511,
+      "loss": 1.16,
+      "step": 9479
+    },
+    {
+      "epoch": 0.7291746788708561,
+      "learning_rate": 0.0005109528907312476,
+      "loss": 1.4642,
+      "step": 9480
+    },
+    {
+      "epoch": 0.7292515960310745,
+      "learning_rate": 0.0005106804110784758,
+      "loss": 1.1429,
+      "step": 9481
+    },
+    {
+      "epoch": 0.729328513191293,
+      "learning_rate": 0.0005104079891931052,
+      "loss": 1.1273,
+      "step": 9482
+    },
+    {
+      "epoch": 0.7294054303515114,
+      "learning_rate": 0.0005101356250910438,
+      "loss": 1.2409,
+      "step": 9483
+    },
+    {
+      "epoch": 0.7294823475117299,
+      "learning_rate": 0.0005098633187881948,
+      "loss": 0.9331,
+      "step": 9484
+    },
+    {
+      "epoch": 0.7295592646719483,
+      "learning_rate": 0.0005095910703004586,
+      "loss": 1.1855,
+      "step": 9485
+    },
+    {
+      "epoch": 0.7296361818321667,
+      "learning_rate": 0.0005093188796437319,
+      "loss": 1.0239,
+      "step": 9486
+    },
+    {
+      "epoch": 0.7297130989923852,
+      "learning_rate": 0.0005090467468339078,
+      "loss": 1.1713,
+      "step": 9487
+    },
+    {
+      "epoch": 0.7297900161526036,
+      "learning_rate": 0.0005087746718868774,
+      "loss": 1.3517,
+      "step": 9488
+    },
+    {
+      "epoch": 0.729866933312822,
+      "learning_rate": 0.000508502654818527,
+      "loss": 1.2247,
+      "step": 9489
+    },
+    {
+      "epoch": 0.7299438504730406,
+      "learning_rate": 0.0005082306956447399,
+      "loss": 1.1024,
+      "step": 9490
+    },
+    {
+      "epoch": 0.730020767633259,
+      "learning_rate": 0.0005079587943813959,
+      "loss": 1.0691,
+      "step": 9491
+    },
+    {
+      "epoch": 0.7300976847934775,
+      "learning_rate": 0.0005076869510443715,
+      "loss": 1.4135,
+      "step": 9492
+    },
+    {
+      "epoch": 0.7301746019536959,
+      "learning_rate": 0.0005074151656495406,
+      "loss": 1.1331,
+      "step": 9493
+    },
+    {
+      "epoch": 0.7302515191139143,
+      "learning_rate": 0.0005071434382127732,
+      "loss": 0.9224,
+      "step": 9494
+    },
+    {
+      "epoch": 0.7303284362741328,
+      "learning_rate": 0.0005068717687499341,
+      "loss": 0.9198,
+      "step": 9495
+    },
+    {
+      "epoch": 0.7304053534343512,
+      "learning_rate": 0.000506600157276888,
+      "loss": 0.995,
+      "step": 9496
+    },
+    {
+      "epoch": 0.7304822705945696,
+      "learning_rate": 0.0005063286038094938,
+      "loss": 1.0685,
+      "step": 9497
+    },
+    {
+      "epoch": 0.7305591877547881,
+      "learning_rate": 0.0005060571083636083,
+      "loss": 1.2064,
+      "step": 9498
+    },
+    {
+      "epoch": 0.7306361049150065,
+      "learning_rate": 0.0005057856709550839,
+      "loss": 0.8912,
+      "step": 9499
+    },
+    {
+      "epoch": 0.7307130220752249,
+      "learning_rate": 0.0005055142915997705,
+      "loss": 1.7754,
+      "step": 9500
+    },
+    {
+      "epoch": 0.7307899392354434,
+      "learning_rate": 0.0005052429703135134,
+      "loss": 1.4261,
+      "step": 9501
+    },
+    {
+      "epoch": 0.7308668563956618,
+      "learning_rate": 0.0005049717071121565,
+      "loss": 1.5207,
+      "step": 9502
+    },
+    {
+      "epoch": 0.7309437735558804,
+      "learning_rate": 0.0005047005020115385,
+      "loss": 1.2914,
+      "step": 9503
+    },
+    {
+      "epoch": 0.7310206907160988,
+      "learning_rate": 0.0005044293550274957,
+      "loss": 1.3741,
+      "step": 9504
+    },
+    {
+      "epoch": 0.7310976078763172,
+      "learning_rate": 0.0005041582661758602,
+      "loss": 1.3324,
+      "step": 9505
+    },
+    {
+      "epoch": 0.7311745250365357,
+      "learning_rate": 0.000503887235472461,
+      "loss": 1.411,
+      "step": 9506
+    },
+    {
+      "epoch": 0.7312514421967541,
+      "learning_rate": 0.0005036162629331252,
+      "loss": 1.2625,
+      "step": 9507
+    },
+    {
+      "epoch": 0.7313283593569725,
+      "learning_rate": 0.0005033453485736737,
+      "loss": 0.9098,
+      "step": 9508
+    },
+    {
+      "epoch": 0.731405276517191,
+      "learning_rate": 0.0005030744924099256,
+      "loss": 0.9647,
+      "step": 9509
+    },
+    {
+      "epoch": 0.7314821936774094,
+      "learning_rate": 0.0005028036944576973,
+      "loss": 0.6814,
+      "step": 9510
+    },
+    {
+      "epoch": 0.7315591108376279,
+      "learning_rate": 0.0005025329547328006,
+      "loss": 0.8911,
+      "step": 9511
+    },
+    {
+      "epoch": 0.7316360279978463,
+      "learning_rate": 0.0005022622732510439,
+      "loss": 0.8692,
+      "step": 9512
+    },
+    {
+      "epoch": 0.7317129451580647,
+      "learning_rate": 0.0005019916500282332,
+      "loss": 1.288,
+      "step": 9513
+    },
+    {
+      "epoch": 0.7317898623182832,
+      "learning_rate": 0.0005017210850801696,
+      "loss": 1.1947,
+      "step": 9514
+    },
+    {
+      "epoch": 0.7318667794785017,
+      "learning_rate": 0.0005014505784226527,
+      "loss": 1.1111,
+      "step": 9515
+    },
+    {
+      "epoch": 0.73194369663872,
+      "learning_rate": 0.0005011801300714777,
+      "loss": 1.0163,
+      "step": 9516
+    },
+    {
+      "epoch": 0.7320206137989386,
+      "learning_rate": 0.0005009097400424347,
+      "loss": 1.1756,
+      "step": 9517
+    },
+    {
+      "epoch": 0.732097530959157,
+      "learning_rate": 0.0005006394083513139,
+      "loss": 1.3218,
+      "step": 9518
+    },
+    {
+      "epoch": 0.7321744481193755,
+      "learning_rate": 0.0005003691350138989,
+      "loss": 1.1557,
+      "step": 9519
+    },
+    {
+      "epoch": 0.7322513652795939,
+      "learning_rate": 0.000500098920045973,
+      "loss": 1.1851,
+      "step": 9520
+    },
+    {
+      "epoch": 0.7323282824398123,
+      "learning_rate": 0.0004998287634633128,
+      "loss": 1.1823,
+      "step": 9521
+    },
+    {
+      "epoch": 0.7324051996000308,
+      "learning_rate": 0.0004995586652816928,
+      "loss": 1.0033,
+      "step": 9522
+    },
+    {
+      "epoch": 0.7324821167602492,
+      "learning_rate": 0.0004992886255168855,
+      "loss": 1.279,
+      "step": 9523
+    },
+    {
+      "epoch": 0.7325590339204676,
+      "learning_rate": 0.0004990186441846584,
+      "loss": 0.9729,
+      "step": 9524
+    },
+    {
+      "epoch": 0.7326359510806861,
+      "learning_rate": 0.0004987487213007756,
+      "loss": 0.7133,
+      "step": 9525
+    },
+    {
+      "epoch": 0.7327128682409045,
+      "learning_rate": 0.0004984788568809988,
+      "loss": 1.163,
+      "step": 9526
+    },
+    {
+      "epoch": 0.7327897854011229,
+      "learning_rate": 0.000498209050941085,
+      "loss": 1.2016,
+      "step": 9527
+    },
+    {
+      "epoch": 0.7328667025613415,
+      "learning_rate": 0.0004979393034967884,
+      "loss": 1.334,
+      "step": 9528
+    },
+    {
+      "epoch": 0.7329436197215599,
+      "learning_rate": 0.0004976696145638612,
+      "loss": 0.9937,
+      "step": 9529
+    },
+    {
+      "epoch": 0.7330205368817784,
+      "learning_rate": 0.0004973999841580487,
+      "loss": 1.1001,
+      "step": 9530
+    },
+    {
+      "epoch": 0.7330974540419968,
+      "learning_rate": 0.0004971304122950966,
+      "loss": 1.2798,
+      "step": 9531
+    },
+    {
+      "epoch": 0.7331743712022152,
+      "learning_rate": 0.000496860898990745,
+      "loss": 0.9932,
+      "step": 9532
+    },
+    {
+      "epoch": 0.7332512883624337,
+      "learning_rate": 0.0004965914442607307,
+      "loss": 0.8744,
+      "step": 9533
+    },
+    {
+      "epoch": 0.7333282055226521,
+      "learning_rate": 0.000496322048120788,
+      "loss": 1.2469,
+      "step": 9534
+    },
+    {
+      "epoch": 0.7334051226828705,
+      "learning_rate": 0.0004960527105866467,
+      "loss": 0.3908,
+      "step": 9535
+    },
+    {
+      "epoch": 0.733482039843089,
+      "learning_rate": 0.0004957834316740335,
+      "loss": 1.1258,
+      "step": 9536
+    },
+    {
+      "epoch": 0.7335589570033074,
+      "learning_rate": 0.0004955142113986727,
+      "loss": 0.8403,
+      "step": 9537
+    },
+    {
+      "epoch": 0.7336358741635259,
+      "learning_rate": 0.000495245049776284,
+      "loss": 1.2507,
+      "step": 9538
+    },
+    {
+      "epoch": 0.7337127913237443,
+      "learning_rate": 0.0004949759468225839,
+      "loss": 1.1635,
+      "step": 9539
+    },
+    {
+      "epoch": 0.7337897084839627,
+      "learning_rate": 0.0004947069025532859,
+      "loss": 1.0787,
+      "step": 9540
+    },
+    {
+      "epoch": 0.7338666256441813,
+      "learning_rate": 0.0004944379169840988,
+      "loss": 1.2461,
+      "step": 9541
+    },
+    {
+      "epoch": 0.7339435428043997,
+      "learning_rate": 0.0004941689901307308,
+      "loss": 1.0702,
+      "step": 9542
+    },
+    {
+      "epoch": 0.7340204599646181,
+      "learning_rate": 0.0004939001220088833,
+      "loss": 1.0311,
+      "step": 9543
+    },
+    {
+      "epoch": 0.7340973771248366,
+      "learning_rate": 0.0004936313126342556,
+      "loss": 1.0072,
+      "step": 9544
+    },
+    {
+      "epoch": 0.734174294285055,
+      "learning_rate": 0.0004933625620225449,
+      "loss": 1.4752,
+      "step": 9545
+    },
+    {
+      "epoch": 0.7342512114452734,
+      "learning_rate": 0.0004930938701894434,
+      "loss": 1.0207,
+      "step": 9546
+    },
+    {
+      "epoch": 0.7343281286054919,
+      "learning_rate": 0.00049282523715064,
+      "loss": 1.3235,
+      "step": 9547
+    },
+    {
+      "epoch": 0.7344050457657103,
+      "learning_rate": 0.0004925566629218208,
+      "loss": 1.133,
+      "step": 9548
+    },
+    {
+      "epoch": 0.7344819629259288,
+      "learning_rate": 0.0004922881475186673,
+      "loss": 1.1452,
+      "step": 9549
+    },
+    {
+      "epoch": 0.7345588800861472,
+      "learning_rate": 0.0004920196909568599,
+      "loss": 1.055,
+      "step": 9550
+    },
+    {
+      "epoch": 0.7346357972463656,
+      "learning_rate": 0.0004917512932520734,
+      "loss": 0.8454,
+      "step": 9551
+    },
+    {
+      "epoch": 0.7347127144065841,
+      "learning_rate": 0.0004914829544199788,
+      "loss": 1.3494,
+      "step": 9552
+    },
+    {
+      "epoch": 0.7347896315668025,
+      "learning_rate": 0.000491214674476246,
+      "loss": 1.3645,
+      "step": 9553
+    },
+    {
+      "epoch": 0.734866548727021,
+      "learning_rate": 0.0004909464534365399,
+      "loss": 1.1012,
+      "step": 9554
+    },
+    {
+      "epoch": 0.7349434658872395,
+      "learning_rate": 0.0004906782913165217,
+      "loss": 1.4507,
+      "step": 9555
+    },
+    {
+      "epoch": 0.7350203830474579,
+      "learning_rate": 0.0004904101881318501,
+      "loss": 1.1934,
+      "step": 9556
+    },
+    {
+      "epoch": 0.7350973002076764,
+      "learning_rate": 0.0004901421438981792,
+      "loss": 1.6996,
+      "step": 9557
+    },
+    {
+      "epoch": 0.7351742173678948,
+      "learning_rate": 0.0004898741586311615,
+      "loss": 1.2667,
+      "step": 9558
+    },
+    {
+      "epoch": 0.7352511345281132,
+      "learning_rate": 0.0004896062323464447,
+      "loss": 1.1764,
+      "step": 9559
+    },
+    {
+      "epoch": 0.7353280516883317,
+      "learning_rate": 0.0004893383650596726,
+      "loss": 0.8002,
+      "step": 9560
+    },
+    {
+      "epoch": 0.7354049688485501,
+      "learning_rate": 0.0004890705567864868,
+      "loss": 0.9024,
+      "step": 9561
+    },
+    {
+      "epoch": 0.7354818860087685,
+      "learning_rate": 0.0004888028075425248,
+      "loss": 1.0302,
+      "step": 9562
+    },
+    {
+      "epoch": 0.735558803168987,
+      "learning_rate": 0.0004885351173434203,
+      "loss": 1.1395,
+      "step": 9563
+    },
+    {
+      "epoch": 0.7356357203292054,
+      "learning_rate": 0.00048826748620480517,
+      "loss": 1.3611,
+      "step": 9564
+    },
+    {
+      "epoch": 0.7357126374894238,
+      "learning_rate": 0.00048799991414230545,
+      "loss": 1.1677,
+      "step": 9565
+    },
+    {
+      "epoch": 0.7357895546496424,
+      "learning_rate": 0.00048773240117154504,
+      "loss": 1.1451,
+      "step": 9566
+    },
+    {
+      "epoch": 0.7358664718098608,
+      "learning_rate": 0.0004874649473081452,
+      "loss": 1.0619,
+      "step": 9567
+    },
+    {
+      "epoch": 0.7359433889700793,
+      "learning_rate": 0.0004871975525677223,
+      "loss": 1.2173,
+      "step": 9568
+    },
+    {
+      "epoch": 0.7360203061302977,
+      "learning_rate": 0.0004869302169658898,
+      "loss": 0.9331,
+      "step": 9569
+    },
+    {
+      "epoch": 0.7360972232905161,
+      "learning_rate": 0.00048666294051825763,
+      "loss": 1.1044,
+      "step": 9570
+    },
+    {
+      "epoch": 0.7361741404507346,
+      "learning_rate": 0.0004863957232404321,
+      "loss": 1.0849,
+      "step": 9571
+    },
+    {
+      "epoch": 0.736251057610953,
+      "learning_rate": 0.00048612856514801704,
+      "loss": 1.3422,
+      "step": 9572
+    },
+    {
+      "epoch": 0.7363279747711714,
+      "learning_rate": 0.00048586146625661234,
+      "loss": 1.2571,
+      "step": 9573
+    },
+    {
+      "epoch": 0.7364048919313899,
+      "learning_rate": 0.00048559442658181235,
+      "loss": 0.9953,
+      "step": 9574
+    },
+    {
+      "epoch": 0.7364818090916083,
+      "learning_rate": 0.00048532744613921154,
+      "loss": 1.3342,
+      "step": 9575
+    },
+    {
+      "epoch": 0.7365587262518268,
+      "learning_rate": 0.0004850605249443982,
+      "loss": 0.8317,
+      "step": 9576
+    },
+    {
+      "epoch": 0.7366356434120452,
+      "learning_rate": 0.00048479366301295944,
+      "loss": 1.3491,
+      "step": 9577
+    },
+    {
+      "epoch": 0.7367125605722636,
+      "learning_rate": 0.00048452686036047626,
+      "loss": 0.9034,
+      "step": 9578
+    },
+    {
+      "epoch": 0.7367894777324822,
+      "learning_rate": 0.00048426011700252754,
+      "loss": 1.0369,
+      "step": 9579
+    },
+    {
+      "epoch": 0.7368663948927006,
+      "learning_rate": 0.0004839934329546896,
+      "loss": 1.2174,
+      "step": 9580
+    },
+    {
+      "epoch": 0.736943312052919,
+      "learning_rate": 0.0004837268082325341,
+      "loss": 1.0057,
+      "step": 9581
+    },
+    {
+      "epoch": 0.7370202292131375,
+      "learning_rate": 0.0004834602428516293,
+      "loss": 1.2124,
+      "step": 9582
+    },
+    {
+      "epoch": 0.7370971463733559,
+      "learning_rate": 0.00048319373682754026,
+      "loss": 1.3084,
+      "step": 9583
+    },
+    {
+      "epoch": 0.7371740635335743,
+      "learning_rate": 0.00048292729017582875,
+      "loss": 1.3049,
+      "step": 9584
+    },
+    {
+      "epoch": 0.7372509806937928,
+      "learning_rate": 0.0004826609029120522,
+      "loss": 1.1122,
+      "step": 9585
+    },
+    {
+      "epoch": 0.7373278978540112,
+      "learning_rate": 0.0004823945750517667,
+      "loss": 1.2862,
+      "step": 9586
+    },
+    {
+      "epoch": 0.7374048150142297,
+      "learning_rate": 0.0004821283066105216,
+      "loss": 1.1289,
+      "step": 9587
+    },
+    {
+      "epoch": 0.7374817321744481,
+      "learning_rate": 0.0004818620976038659,
+      "loss": 1.1788,
+      "step": 9588
+    },
+    {
+      "epoch": 0.7375586493346665,
+      "learning_rate": 0.00048159594804734335,
+      "loss": 0.842,
+      "step": 9589
+    },
+    {
+      "epoch": 0.737635566494885,
+      "learning_rate": 0.00048132985795649487,
+      "loss": 1.2714,
+      "step": 9590
+    },
+    {
+      "epoch": 0.7377124836551034,
+      "learning_rate": 0.00048106382734685747,
+      "loss": 1.1957,
+      "step": 9591
+    },
+    {
+      "epoch": 0.7377894008153218,
+      "learning_rate": 0.00048079785623396533,
+      "loss": 1.3048,
+      "step": 9592
+    },
+    {
+      "epoch": 0.7378663179755404,
+      "learning_rate": 0.00048053194463334775,
+      "loss": 1.1409,
+      "step": 9593
+    },
+    {
+      "epoch": 0.7379432351357588,
+      "learning_rate": 0.0004802660925605332,
+      "loss": 1.3611,
+      "step": 9594
+    },
+    {
+      "epoch": 0.7380201522959773,
+      "learning_rate": 0.00048000030003104413,
+      "loss": 0.7823,
+      "step": 9595
+    },
+    {
+      "epoch": 0.7380970694561957,
+      "learning_rate": 0.00047973456706040037,
+      "loss": 0.967,
+      "step": 9596
+    },
+    {
+      "epoch": 0.7381739866164141,
+      "learning_rate": 0.0004794688936641187,
+      "loss": 1.2858,
+      "step": 9597
+    },
+    {
+      "epoch": 0.7382509037766326,
+      "learning_rate": 0.0004792032798577114,
+      "loss": 1.0721,
+      "step": 9598
+    },
+    {
+      "epoch": 0.738327820936851,
+      "learning_rate": 0.0004789377256566895,
+      "loss": 1.0331,
+      "step": 9599
+    },
+    {
+      "epoch": 0.7384047380970694,
+      "learning_rate": 0.0004786722310765574,
+      "loss": 1.0129,
+      "step": 9600
+    },
+    {
+      "epoch": 0.7384816552572879,
+      "learning_rate": 0.00047840679613281745,
+      "loss": 1.1207,
+      "step": 9601
+    },
+    {
+      "epoch": 0.7385585724175063,
+      "learning_rate": 0.0004781414208409701,
+      "loss": 1.2421,
+      "step": 9602
+    },
+    {
+      "epoch": 0.7386354895777247,
+      "learning_rate": 0.00047787610521651,
+      "loss": 1.2682,
+      "step": 9603
+    },
+    {
+      "epoch": 0.7387124067379433,
+      "learning_rate": 0.0004776108492749293,
+      "loss": 1.2052,
+      "step": 9604
+    },
+    {
+      "epoch": 0.7387893238981617,
+      "learning_rate": 0.0004773456530317167,
+      "loss": 1.1606,
+      "step": 9605
+    },
+    {
+      "epoch": 0.7388662410583802,
+      "learning_rate": 0.00047708051650235684,
+      "loss": 1.1792,
+      "step": 9606
+    },
+    {
+      "epoch": 0.7389431582185986,
+      "learning_rate": 0.00047681543970233195,
+      "loss": 1.2722,
+      "step": 9607
+    },
+    {
+      "epoch": 0.739020075378817,
+      "learning_rate": 0.00047655042264712044,
+      "loss": 1.2043,
+      "step": 9608
+    },
+    {
+      "epoch": 0.7390969925390355,
+      "learning_rate": 0.00047628546535219554,
+      "loss": 1.0525,
+      "step": 9609
+    },
+    {
+      "epoch": 0.7391739096992539,
+      "learning_rate": 0.0004760205678330294,
+      "loss": 0.9423,
+      "step": 9610
+    },
+    {
+      "epoch": 0.7392508268594723,
+      "learning_rate": 0.0004757557301050897,
+      "loss": 1.1264,
+      "step": 9611
+    },
+    {
+      "epoch": 0.7393277440196908,
+      "learning_rate": 0.0004754909521838403,
+      "loss": 1.4684,
+      "step": 9612
+    },
+    {
+      "epoch": 0.7394046611799092,
+      "learning_rate": 0.00047522623408474186,
+      "loss": 1.1883,
+      "step": 9613
+    },
+    {
+      "epoch": 0.7394815783401277,
+      "learning_rate": 0.00047496157582325126,
+      "loss": 0.9851,
+      "step": 9614
+    },
+    {
+      "epoch": 0.7395584955003461,
+      "learning_rate": 0.0004746969774148229,
+      "loss": 0.6258,
+      "step": 9615
+    },
+    {
+      "epoch": 0.7396354126605645,
+      "learning_rate": 0.00047443243887490663,
+      "loss": 1.7361,
+      "step": 9616
+    },
+    {
+      "epoch": 0.739712329820783,
+      "learning_rate": 0.000474167960218949,
+      "loss": 1.109,
+      "step": 9617
+    },
+    {
+      "epoch": 0.7397892469810015,
+      "learning_rate": 0.0004739035414623933,
+      "loss": 1.0861,
+      "step": 9618
+    },
+    {
+      "epoch": 0.7398661641412199,
+      "learning_rate": 0.0004736391826206792,
+      "loss": 1.15,
+      "step": 9619
+    },
+    {
+      "epoch": 0.7399430813014384,
+      "learning_rate": 0.0004733748837092423,
+      "loss": 0.8956,
+      "step": 9620
+    },
+    {
+      "epoch": 0.7400199984616568,
+      "learning_rate": 0.000473110644743517,
+      "loss": 1.4887,
+      "step": 9621
+    },
+    {
+      "epoch": 0.7400969156218753,
+      "learning_rate": 0.00047284646573893024,
+      "loss": 1.4459,
+      "step": 9622
+    },
+    {
+      "epoch": 0.7401738327820937,
+      "learning_rate": 0.00047258234671090937,
+      "loss": 0.9748,
+      "step": 9623
+    },
+    {
+      "epoch": 0.7402507499423121,
+      "learning_rate": 0.00047231828767487605,
+      "loss": 0.9643,
+      "step": 9624
+    },
+    {
+      "epoch": 0.7403276671025306,
+      "learning_rate": 0.0004720542886462488,
+      "loss": 1.1797,
+      "step": 9625
+    },
+    {
+      "epoch": 0.740404584262749,
+      "learning_rate": 0.00047179034964044304,
+      "loss": 1.0901,
+      "step": 9626
+    },
+    {
+      "epoch": 0.7404815014229674,
+      "learning_rate": 0.00047152647067287036,
+      "loss": 1.2084,
+      "step": 9627
+    },
+    {
+      "epoch": 0.7405584185831859,
+      "learning_rate": 0.0004712626517589384,
+      "loss": 1.4829,
+      "step": 9628
+    },
+    {
+      "epoch": 0.7406353357434043,
+      "learning_rate": 0.00047099889291405284,
+      "loss": 0.8356,
+      "step": 9629
+    },
+    {
+      "epoch": 0.7407122529036227,
+      "learning_rate": 0.0004707351941536142,
+      "loss": 1.2059,
+      "step": 9630
+    },
+    {
+      "epoch": 0.7407891700638413,
+      "learning_rate": 0.0004704715554930203,
+      "loss": 1.0607,
+      "step": 9631
+    },
+    {
+      "epoch": 0.7408660872240597,
+      "learning_rate": 0.0004702079769476653,
+      "loss": 1.2945,
+      "step": 9632
+    },
+    {
+      "epoch": 0.7409430043842782,
+      "learning_rate": 0.00046994445853293935,
+      "loss": 1.1079,
+      "step": 9633
+    },
+    {
+      "epoch": 0.7410199215444966,
+      "learning_rate": 0.0004696810002642307,
+      "loss": 1.1573,
+      "step": 9634
+    },
+    {
+      "epoch": 0.741096838704715,
+      "learning_rate": 0.0004694176021569218,
+      "loss": 1.144,
+      "step": 9635
+    },
+    {
+      "epoch": 0.7411737558649335,
+      "learning_rate": 0.0004691542642263928,
+      "loss": 0.8916,
+      "step": 9636
+    },
+    {
+      "epoch": 0.7412506730251519,
+      "learning_rate": 0.000468890986488021,
+      "loss": 1.2286,
+      "step": 9637
+    },
+    {
+      "epoch": 0.7413275901853703,
+      "learning_rate": 0.00046862776895717907,
+      "loss": 1.212,
+      "step": 9638
+    },
+    {
+      "epoch": 0.7414045073455888,
+      "learning_rate": 0.00046836461164923677,
+      "loss": 1.0431,
+      "step": 9639
+    },
+    {
+      "epoch": 0.7414814245058072,
+      "learning_rate": 0.0004681015145795597,
+      "loss": 1.1437,
+      "step": 9640
+    },
+    {
+      "epoch": 0.7415583416660257,
+      "learning_rate": 0.00046783847776351033,
+      "loss": 1.2865,
+      "step": 9641
+    },
+    {
+      "epoch": 0.7416352588262441,
+      "learning_rate": 0.0004675755012164483,
+      "loss": 1.388,
+      "step": 9642
+    },
+    {
+      "epoch": 0.7417121759864626,
+      "learning_rate": 0.0004673125849537288,
+      "loss": 1.1288,
+      "step": 9643
+    },
+    {
+      "epoch": 0.7417890931466811,
+      "learning_rate": 0.0004670497289907035,
+      "loss": 0.9553,
+      "step": 9644
+    },
+    {
+      "epoch": 0.7418660103068995,
+      "learning_rate": 0.0004667869333427212,
+      "loss": 1.2675,
+      "step": 9645
+    },
+    {
+      "epoch": 0.7419429274671179,
+      "learning_rate": 0.00046652419802512674,
+      "loss": 1.1955,
+      "step": 9646
+    },
+    {
+      "epoch": 0.7420198446273364,
+      "learning_rate": 0.00046626152305326133,
+      "loss": 1.226,
+      "step": 9647
+    },
+    {
+      "epoch": 0.7420967617875548,
+      "learning_rate": 0.00046599890844246293,
+      "loss": 0.531,
+      "step": 9648
+    },
+    {
+      "epoch": 0.7421736789477732,
+      "learning_rate": 0.0004657363542080658,
+      "loss": 0.9925,
+      "step": 9649
+    },
+    {
+      "epoch": 0.7422505961079917,
+      "learning_rate": 0.00046547386036540034,
+      "loss": 1.5166,
+      "step": 9650
+    },
+    {
+      "epoch": 0.7423275132682101,
+      "learning_rate": 0.00046521142692979494,
+      "loss": 1.3091,
+      "step": 9651
+    },
+    {
+      "epoch": 0.7424044304284286,
+      "learning_rate": 0.0004649490539165727,
+      "loss": 1.1894,
+      "step": 9652
+    },
+    {
+      "epoch": 0.742481347588647,
+      "learning_rate": 0.0004646867413410539,
+      "loss": 0.5401,
+      "step": 9653
+    },
+    {
+      "epoch": 0.7425582647488654,
+      "learning_rate": 0.0004644244892185553,
+      "loss": 1.1338,
+      "step": 9654
+    },
+    {
+      "epoch": 0.742635181909084,
+      "learning_rate": 0.0004641622975643895,
+      "loss": 1.0326,
+      "step": 9655
+    },
+    {
+      "epoch": 0.7427120990693024,
+      "learning_rate": 0.00046390016639386706,
+      "loss": 1.4126,
+      "step": 9656
+    },
+    {
+      "epoch": 0.7427890162295208,
+      "learning_rate": 0.00046363809572229434,
+      "loss": 1.1692,
+      "step": 9657
+    },
+    {
+      "epoch": 0.7428659333897393,
+      "learning_rate": 0.00046337608556497236,
+      "loss": 1.0901,
+      "step": 9658
+    },
+    {
+      "epoch": 0.7429428505499577,
+      "learning_rate": 0.0004631141359372014,
+      "loss": 1.406,
+      "step": 9659
+    },
+    {
+      "epoch": 0.7430197677101762,
+      "learning_rate": 0.0004628522468542763,
+      "loss": 0.714,
+      "step": 9660
+    },
+    {
+      "epoch": 0.7430966848703946,
+      "learning_rate": 0.00046259041833149024,
+      "loss": 1.1301,
+      "step": 9661
+    },
+    {
+      "epoch": 0.743173602030613,
+      "learning_rate": 0.00046232865038413036,
+      "loss": 0.8453,
+      "step": 9662
+    },
+    {
+      "epoch": 0.7432505191908315,
+      "learning_rate": 0.0004620669430274816,
+      "loss": 1.146,
+      "step": 9663
+    },
+    {
+      "epoch": 0.7433274363510499,
+      "learning_rate": 0.00046180529627682623,
+      "loss": 1.0231,
+      "step": 9664
+    },
+    {
+      "epoch": 0.7434043535112683,
+      "learning_rate": 0.0004615437101474415,
+      "loss": 0.9878,
+      "step": 9665
+    },
+    {
+      "epoch": 0.7434812706714868,
+      "learning_rate": 0.00046128218465460176,
+      "loss": 1.4212,
+      "step": 9666
+    },
+    {
+      "epoch": 0.7435581878317052,
+      "learning_rate": 0.0004610207198135779,
+      "loss": 0.8955,
+      "step": 9667
+    },
+    {
+      "epoch": 0.7436351049919236,
+      "learning_rate": 0.0004607593156396371,
+      "loss": 1.2782,
+      "step": 9668
+    },
+    {
+      "epoch": 0.7437120221521422,
+      "learning_rate": 0.0004604979721480422,
+      "loss": 1.2521,
+      "step": 9669
+    },
+    {
+      "epoch": 0.7437889393123606,
+      "learning_rate": 0.00046023668935405516,
+      "loss": 1.1578,
+      "step": 9670
+    },
+    {
+      "epoch": 0.7438658564725791,
+      "learning_rate": 0.00045997546727293047,
+      "loss": 0.9417,
+      "step": 9671
+    },
+    {
+      "epoch": 0.7439427736327975,
+      "learning_rate": 0.0004597143059199227,
+      "loss": 0.9857,
+      "step": 9672
+    },
+    {
+      "epoch": 0.7440196907930159,
+      "learning_rate": 0.00045945320531028057,
+      "loss": 1.1483,
+      "step": 9673
+    },
+    {
+      "epoch": 0.7440966079532344,
+      "learning_rate": 0.00045919216545925023,
+      "loss": 1.0915,
+      "step": 9674
+    },
+    {
+      "epoch": 0.7441735251134528,
+      "learning_rate": 0.0004589311863820742,
+      "loss": 1.2431,
+      "step": 9675
+    },
+    {
+      "epoch": 0.7442504422736712,
+      "learning_rate": 0.00045867026809399107,
+      "loss": 1.2593,
+      "step": 9676
+    },
+    {
+      "epoch": 0.7443273594338897,
+      "learning_rate": 0.00045840941061023587,
+      "loss": 1.0995,
+      "step": 9677
+    },
+    {
+      "epoch": 0.7444042765941081,
+      "learning_rate": 0.00045814861394604105,
+      "loss": 0.8031,
+      "step": 9678
+    },
+    {
+      "epoch": 0.7444811937543266,
+      "learning_rate": 0.0004578878781166346,
+      "loss": 1.0974,
+      "step": 9679
+    },
+    {
+      "epoch": 0.744558110914545,
+      "learning_rate": 0.00045762720313724086,
+      "loss": 0.8156,
+      "step": 9680
+    },
+    {
+      "epoch": 0.7446350280747634,
+      "learning_rate": 0.00045736658902308114,
+      "loss": 1.3662,
+      "step": 9681
+    },
+    {
+      "epoch": 0.744711945234982,
+      "learning_rate": 0.0004571060357893724,
+      "loss": 1.1884,
+      "step": 9682
+    },
+    {
+      "epoch": 0.7447888623952004,
+      "learning_rate": 0.00045684554345133005,
+      "loss": 1.1542,
+      "step": 9683
+    },
+    {
+      "epoch": 0.7448657795554188,
+      "learning_rate": 0.0004565851120241632,
+      "loss": 1.0133,
+      "step": 9684
+    },
+    {
+      "epoch": 0.7449426967156373,
+      "learning_rate": 0.00045632474152307854,
+      "loss": 0.6093,
+      "step": 9685
+    },
+    {
+      "epoch": 0.7450196138758557,
+      "learning_rate": 0.0004560644319632802,
+      "loss": 1.0068,
+      "step": 9686
+    },
+    {
+      "epoch": 0.7450965310360741,
+      "learning_rate": 0.00045580418335996786,
+      "loss": 1.1714,
+      "step": 9687
+    },
+    {
+      "epoch": 0.7451734481962926,
+      "learning_rate": 0.00045554399572833744,
+      "loss": 1.1873,
+      "step": 9688
+    },
+    {
+      "epoch": 0.745250365356511,
+      "learning_rate": 0.00045528386908358153,
+      "loss": 1.0161,
+      "step": 9689
+    },
+    {
+      "epoch": 0.7453272825167295,
+      "learning_rate": 0.0004550238034408889,
+      "loss": 1.0107,
+      "step": 9690
+    },
+    {
+      "epoch": 0.7454041996769479,
+      "learning_rate": 0.000454763798815446,
+      "loss": 0.8771,
+      "step": 9691
+    },
+    {
+      "epoch": 0.7454811168371663,
+      "learning_rate": 0.0004545038552224345,
+      "loss": 1.2641,
+      "step": 9692
+    },
+    {
+      "epoch": 0.7455580339973849,
+      "learning_rate": 0.00045424397267703154,
+      "loss": 1.0957,
+      "step": 9693
+    },
+    {
+      "epoch": 0.7456349511576033,
+      "learning_rate": 0.0004539841511944134,
+      "loss": 1.4517,
+      "step": 9694
+    },
+    {
+      "epoch": 0.7457118683178217,
+      "learning_rate": 0.0004537243907897509,
+      "loss": 1.0689,
+      "step": 9695
+    },
+    {
+      "epoch": 0.7457887854780402,
+      "learning_rate": 0.0004534646914782115,
+      "loss": 1.4997,
+      "step": 9696
+    },
+    {
+      "epoch": 0.7458657026382586,
+      "learning_rate": 0.0004532050532749595,
+      "loss": 1.5104,
+      "step": 9697
+    },
+    {
+      "epoch": 0.7459426197984771,
+      "learning_rate": 0.0004529454761951549,
+      "loss": 0.7274,
+      "step": 9698
+    },
+    {
+      "epoch": 0.7460195369586955,
+      "learning_rate": 0.00045268596025395554,
+      "loss": 1.0045,
+      "step": 9699
+    },
+    {
+      "epoch": 0.7460964541189139,
+      "learning_rate": 0.0004524265054665146,
+      "loss": 1.2495,
+      "step": 9700
+    },
+    {
+      "epoch": 0.7461733712791324,
+      "learning_rate": 0.0004521671118479818,
+      "loss": 1.1395,
+      "step": 9701
+    },
+    {
+      "epoch": 0.7462502884393508,
+      "learning_rate": 0.00045190777941350307,
+      "loss": 1.5124,
+      "step": 9702
+    },
+    {
+      "epoch": 0.7463272055995692,
+      "learning_rate": 0.00045164850817822164,
+      "loss": 1.4483,
+      "step": 9703
+    },
+    {
+      "epoch": 0.7464041227597877,
+      "learning_rate": 0.0004513892981572759,
+      "loss": 1.1922,
+      "step": 9704
+    },
+    {
+      "epoch": 0.7464810399200061,
+      "learning_rate": 0.0004511301493658027,
+      "loss": 0.8614,
+      "step": 9705
+    },
+    {
+      "epoch": 0.7465579570802245,
+      "learning_rate": 0.0004508710618189329,
+      "loss": 0.9824,
+      "step": 9706
+    },
+    {
+      "epoch": 0.7466348742404431,
+      "learning_rate": 0.00045061203553179455,
+      "loss": 0.982,
+      "step": 9707
+    },
+    {
+      "epoch": 0.7467117914006615,
+      "learning_rate": 0.00045035307051951353,
+      "loss": 1.5379,
+      "step": 9708
+    },
+    {
+      "epoch": 0.74678870856088,
+      "learning_rate": 0.00045009416679721076,
+      "loss": 1.152,
+      "step": 9709
+    },
+    {
+      "epoch": 0.7468656257210984,
+      "learning_rate": 0.0004498353243800037,
+      "loss": 1.3142,
+      "step": 9710
+    },
+    {
+      "epoch": 0.7469425428813168,
+      "learning_rate": 0.0004495765432830065,
+      "loss": 1.0859,
+      "step": 9711
+    },
+    {
+      "epoch": 0.7470194600415353,
+      "learning_rate": 0.00044931782352132917,
+      "loss": 1.1589,
+      "step": 9712
+    },
+    {
+      "epoch": 0.7470963772017537,
+      "learning_rate": 0.0004490591651100797,
+      "loss": 1.077,
+      "step": 9713
+    },
+    {
+      "epoch": 0.7471732943619721,
+      "learning_rate": 0.00044880056806436124,
+      "loss": 1.435,
+      "step": 9714
+    },
+    {
+      "epoch": 0.7472502115221906,
+      "learning_rate": 0.0004485420323992722,
+      "loss": 1.2256,
+      "step": 9715
+    },
+    {
+      "epoch": 0.747327128682409,
+      "learning_rate": 0.0004482835581299102,
+      "loss": 1.1679,
+      "step": 9716
+    },
+    {
+      "epoch": 0.7474040458426275,
+      "learning_rate": 0.0004480251452713669,
+      "loss": 1.0517,
+      "step": 9717
+    },
+    {
+      "epoch": 0.7474809630028459,
+      "learning_rate": 0.0004477667938387325,
+      "loss": 1.2843,
+      "step": 9718
+    },
+    {
+      "epoch": 0.7475578801630643,
+      "learning_rate": 0.00044750850384709116,
+      "loss": 1.4952,
+      "step": 9719
+    },
+    {
+      "epoch": 0.7476347973232829,
+      "learning_rate": 0.0004472502753115248,
+      "loss": 1.2533,
+      "step": 9720
+    },
+    {
+      "epoch": 0.7477117144835013,
+      "learning_rate": 0.0004469921082471124,
+      "loss": 1.1568,
+      "step": 9721
+    },
+    {
+      "epoch": 0.7477886316437197,
+      "learning_rate": 0.0004467340026689283,
+      "loss": 1.26,
+      "step": 9722
+    },
+    {
+      "epoch": 0.7478655488039382,
+      "learning_rate": 0.00044647595859204336,
+      "loss": 0.9277,
+      "step": 9723
+    },
+    {
+      "epoch": 0.7479424659641566,
+      "learning_rate": 0.0004462179760315253,
+      "loss": 1.4459,
+      "step": 9724
+    },
+    {
+      "epoch": 0.7480193831243751,
+      "learning_rate": 0.00044596005500243736,
+      "loss": 1.577,
+      "step": 9725
+    },
+    {
+      "epoch": 0.7480963002845935,
+      "learning_rate": 0.00044570219551984063,
+      "loss": 1.0829,
+      "step": 9726
+    },
+    {
+      "epoch": 0.7481732174448119,
+      "learning_rate": 0.00044544439759879197,
+      "loss": 1.2658,
+      "step": 9727
+    },
+    {
+      "epoch": 0.7482501346050304,
+      "learning_rate": 0.0004451866612543431,
+      "loss": 1.3286,
+      "step": 9728
+    },
+    {
+      "epoch": 0.7483270517652488,
+      "learning_rate": 0.0004449289865015448,
+      "loss": 0.9882,
+      "step": 9729
+    },
+    {
+      "epoch": 0.7484039689254672,
+      "learning_rate": 0.00044467137335544264,
+      "loss": 1.5653,
+      "step": 9730
+    },
+    {
+      "epoch": 0.7484808860856857,
+      "learning_rate": 0.0004444138218310787,
+      "loss": 1.2296,
+      "step": 9731
+    },
+    {
+      "epoch": 0.7485578032459042,
+      "learning_rate": 0.0004441563319434919,
+      "loss": 0.8052,
+      "step": 9732
+    },
+    {
+      "epoch": 0.7486347204061226,
+      "learning_rate": 0.00044389890370771714,
+      "loss": 1.3293,
+      "step": 9733
+    },
+    {
+      "epoch": 0.7487116375663411,
+      "learning_rate": 0.00044364153713878564,
+      "loss": 1.0459,
+      "step": 9734
+    },
+    {
+      "epoch": 0.7487885547265595,
+      "learning_rate": 0.00044338423225172627,
+      "loss": 1.6434,
+      "step": 9735
+    },
+    {
+      "epoch": 0.748865471886778,
+      "learning_rate": 0.00044312698906156264,
+      "loss": 1.0665,
+      "step": 9736
+    },
+    {
+      "epoch": 0.7489423890469964,
+      "learning_rate": 0.0004428698075833156,
+      "loss": 1.1757,
+      "step": 9737
+    },
+    {
+      "epoch": 0.7490193062072148,
+      "learning_rate": 0.00044261268783200207,
+      "loss": 1.2172,
+      "step": 9738
+    },
+    {
+      "epoch": 0.7490962233674333,
+      "learning_rate": 0.0004423556298226353,
+      "loss": 0.7575,
+      "step": 9739
+    },
+    {
+      "epoch": 0.7491731405276517,
+      "learning_rate": 0.0004420986335702266,
+      "loss": 1.3136,
+      "step": 9740
+    },
+    {
+      "epoch": 0.7492500576878701,
+      "learning_rate": 0.00044184169908978064,
+      "loss": 1.3724,
+      "step": 9741
+    },
+    {
+      "epoch": 0.7493269748480886,
+      "learning_rate": 0.00044158482639630015,
+      "loss": 0.932,
+      "step": 9742
+    },
+    {
+      "epoch": 0.749403892008307,
+      "learning_rate": 0.00044132801550478523,
+      "loss": 1.5715,
+      "step": 9743
+    },
+    {
+      "epoch": 0.7494808091685256,
+      "learning_rate": 0.00044107126643023087,
+      "loss": 1.2626,
+      "step": 9744
+    },
+    {
+      "epoch": 0.749557726328744,
+      "learning_rate": 0.0004408145791876287,
+      "loss": 1.1827,
+      "step": 9745
+    },
+    {
+      "epoch": 0.7496346434889624,
+      "learning_rate": 0.00044055795379196734,
+      "loss": 1.2491,
+      "step": 9746
+    },
+    {
+      "epoch": 0.7497115606491809,
+      "learning_rate": 0.0004403013902582305,
+      "loss": 0.8887,
+      "step": 9747
+    },
+    {
+      "epoch": 0.7497884778093993,
+      "learning_rate": 0.0004400448886014005,
+      "loss": 1.1813,
+      "step": 9748
+    },
+    {
+      "epoch": 0.7498653949696177,
+      "learning_rate": 0.00043978844883645437,
+      "loss": 1.2962,
+      "step": 9749
+    },
+    {
+      "epoch": 0.7499423121298362,
+      "learning_rate": 0.0004395320709783649,
+      "loss": 1.2228,
+      "step": 9750
+    },
+    {
+      "epoch": 0.7500192292900546,
+      "learning_rate": 0.00043927575504210326,
+      "loss": 0.9294,
+      "step": 9751
+    },
+    {
+      "epoch": 0.750096146450273,
+      "learning_rate": 0.000439019501042636,
+      "loss": 0.7966,
+      "step": 9752
+    },
+    {
+      "epoch": 0.7501730636104915,
+      "learning_rate": 0.0004387633089949256,
+      "loss": 1.073,
+      "step": 9753
+    },
+    {
+      "epoch": 0.7502499807707099,
+      "learning_rate": 0.00043850717891393167,
+      "loss": 0.8838,
+      "step": 9754
+    },
+    {
+      "epoch": 0.7503268979309284,
+      "learning_rate": 0.00043825111081460935,
+      "loss": 1.0305,
+      "step": 9755
+    },
+    {
+      "epoch": 0.7504038150911468,
+      "learning_rate": 0.00043799510471191166,
+      "loss": 1.3044,
+      "step": 9756
+    },
+    {
+      "epoch": 0.7504807322513652,
+      "learning_rate": 0.00043773916062078656,
+      "loss": 0.8356,
+      "step": 9757
+    },
+    {
+      "epoch": 0.7505576494115838,
+      "learning_rate": 0.00043748327855617906,
+      "loss": 0.9336,
+      "step": 9758
+    },
+    {
+      "epoch": 0.7506345665718022,
+      "learning_rate": 0.00043722745853303017,
+      "loss": 1.0255,
+      "step": 9759
+    },
+    {
+      "epoch": 0.7507114837320206,
+      "learning_rate": 0.00043697170056627744,
+      "loss": 1.1667,
+      "step": 9760
+    },
+    {
+      "epoch": 0.7507884008922391,
+      "learning_rate": 0.00043671600467085466,
+      "loss": 1.3942,
+      "step": 9761
+    },
+    {
+      "epoch": 0.7508653180524575,
+      "learning_rate": 0.00043646037086169347,
+      "loss": 1.1702,
+      "step": 9762
+    },
+    {
+      "epoch": 0.750942235212676,
+      "learning_rate": 0.0004362047991537185,
+      "loss": 1.3565,
+      "step": 9763
+    },
+    {
+      "epoch": 0.7510191523728944,
+      "learning_rate": 0.0004359492895618546,
+      "loss": 1.3172,
+      "step": 9764
+    },
+    {
+      "epoch": 0.7510960695331128,
+      "learning_rate": 0.0004356938421010206,
+      "loss": 0.9548,
+      "step": 9765
+    },
+    {
+      "epoch": 0.7511729866933313,
+      "learning_rate": 0.00043543845678613226,
+      "loss": 1.3256,
+      "step": 9766
+    },
+    {
+      "epoch": 0.7512499038535497,
+      "learning_rate": 0.00043518313363210203,
+      "loss": 1.2936,
+      "step": 9767
+    },
+    {
+      "epoch": 0.7513268210137681,
+      "learning_rate": 0.00043492787265383823,
+      "loss": 1.2388,
+      "step": 9768
+    },
+    {
+      "epoch": 0.7514037381739866,
+      "learning_rate": 0.0004346726738662455,
+      "loss": 1.1373,
+      "step": 9769
+    },
+    {
+      "epoch": 0.751480655334205,
+      "learning_rate": 0.0004344175372842261,
+      "loss": 1.0863,
+      "step": 9770
+    },
+    {
+      "epoch": 0.7515575724944235,
+      "learning_rate": 0.00043416246292267717,
+      "loss": 1.2035,
+      "step": 9771
+    },
+    {
+      "epoch": 0.751634489654642,
+      "learning_rate": 0.0004339074507964928,
+      "loss": 1.0746,
+      "step": 9772
+    },
+    {
+      "epoch": 0.7517114068148604,
+      "learning_rate": 0.0004336525009205635,
+      "loss": 1.2591,
+      "step": 9773
+    },
+    {
+      "epoch": 0.7517883239750789,
+      "learning_rate": 0.00043339761330977544,
+      "loss": 0.8002,
+      "step": 9774
+    },
+    {
+      "epoch": 0.7518652411352973,
+      "learning_rate": 0.00043314278797901333,
+      "loss": 0.9797,
+      "step": 9775
+    },
+    {
+      "epoch": 0.7519421582955157,
+      "learning_rate": 0.0004328880249431553,
+      "loss": 0.9351,
+      "step": 9776
+    },
+    {
+      "epoch": 0.7520190754557342,
+      "learning_rate": 0.00043263332421707694,
+      "loss": 0.9714,
+      "step": 9777
+    },
+    {
+      "epoch": 0.7520959926159526,
+      "learning_rate": 0.00043237868581565185,
+      "loss": 1.349,
+      "step": 9778
+    },
+    {
+      "epoch": 0.752172909776171,
+      "learning_rate": 0.00043212410975374785,
+      "loss": 1.3253,
+      "step": 9779
+    },
+    {
+      "epoch": 0.7522498269363895,
+      "learning_rate": 0.00043186959604622993,
+      "loss": 1.2438,
+      "step": 9780
+    },
+    {
+      "epoch": 0.7523267440966079,
+      "learning_rate": 0.0004316151447079595,
+      "loss": 1.2926,
+      "step": 9781
+    },
+    {
+      "epoch": 0.7524036612568265,
+      "learning_rate": 0.00043136075575379374,
+      "loss": 1.4467,
+      "step": 9782
+    },
+    {
+      "epoch": 0.7524805784170449,
+      "learning_rate": 0.00043110642919858764,
+      "loss": 1.8067,
+      "step": 9783
+    },
+    {
+      "epoch": 0.7525574955772633,
+      "learning_rate": 0.00043085216505719124,
+      "loss": 1.134,
+      "step": 9784
+    },
+    {
+      "epoch": 0.7526344127374818,
+      "learning_rate": 0.000430597963344451,
+      "loss": 1.2367,
+      "step": 9785
+    },
+    {
+      "epoch": 0.7527113298977002,
+      "learning_rate": 0.00043034382407521017,
+      "loss": 1.1981,
+      "step": 9786
+    },
+    {
+      "epoch": 0.7527882470579186,
+      "learning_rate": 0.00043008974726430836,
+      "loss": 1.1371,
+      "step": 9787
+    },
+    {
+      "epoch": 0.7528651642181371,
+      "learning_rate": 0.00042983573292658113,
+      "loss": 1.1948,
+      "step": 9788
+    },
+    {
+      "epoch": 0.7529420813783555,
+      "learning_rate": 0.00042958178107686086,
+      "loss": 0.8531,
+      "step": 9789
+    },
+    {
+      "epoch": 0.7530189985385739,
+      "learning_rate": 0.00042932789172997585,
+      "loss": 1.3793,
+      "step": 9790
+    },
+    {
+      "epoch": 0.7530959156987924,
+      "learning_rate": 0.00042907406490075067,
+      "loss": 1.1198,
+      "step": 9791
+    },
+    {
+      "epoch": 0.7531728328590108,
+      "learning_rate": 0.0004288203006040074,
+      "loss": 1.5074,
+      "step": 9792
+    },
+    {
+      "epoch": 0.7532497500192293,
+      "learning_rate": 0.0004285665988545631,
+      "loss": 1.0186,
+      "step": 9793
+    },
+    {
+      "epoch": 0.7533266671794477,
+      "learning_rate": 0.00042831295966723167,
+      "loss": 1.3088,
+      "step": 9794
+    },
+    {
+      "epoch": 0.7534035843396661,
+      "learning_rate": 0.0004280593830568235,
+      "loss": 1.0121,
+      "step": 9795
+    },
+    {
+      "epoch": 0.7534805014998847,
+      "learning_rate": 0.0004278058690381445,
+      "loss": 1.3308,
+      "step": 9796
+    },
+    {
+      "epoch": 0.7535574186601031,
+      "learning_rate": 0.00042755241762599877,
+      "loss": 1.2559,
+      "step": 9797
+    },
+    {
+      "epoch": 0.7536343358203215,
+      "learning_rate": 0.00042729902883518567,
+      "loss": 1.0505,
+      "step": 9798
+    },
+    {
+      "epoch": 0.75371125298054,
+      "learning_rate": 0.00042704570268049905,
+      "loss": 1.103,
+      "step": 9799
+    },
+    {
+      "epoch": 0.7537881701407584,
+      "learning_rate": 0.00042679243917673264,
+      "loss": 0.8859,
+      "step": 9800
+    },
+    {
+      "epoch": 0.7538650873009769,
+      "learning_rate": 0.00042653923833867424,
+      "loss": 1.5809,
+      "step": 9801
+    },
+    {
+      "epoch": 0.7539420044611953,
+      "learning_rate": 0.0004262861001811083,
+      "loss": 1.2575,
+      "step": 9802
+    },
+    {
+      "epoch": 0.7540189216214137,
+      "learning_rate": 0.000426033024718816,
+      "loss": 1.1196,
+      "step": 9803
+    },
+    {
+      "epoch": 0.7540958387816322,
+      "learning_rate": 0.00042578001196657417,
+      "loss": 1.2707,
+      "step": 9804
+    },
+    {
+      "epoch": 0.7541727559418506,
+      "learning_rate": 0.0004255270619391575,
+      "loss": 1.1006,
+      "step": 9805
+    },
+    {
+      "epoch": 0.754249673102069,
+      "learning_rate": 0.00042527417465133556,
+      "loss": 1.1542,
+      "step": 9806
+    },
+    {
+      "epoch": 0.7543265902622875,
+      "learning_rate": 0.00042502135011787445,
+      "loss": 1.5866,
+      "step": 9807
+    },
+    {
+      "epoch": 0.754403507422506,
+      "learning_rate": 0.0004247685883535371,
+      "loss": 1.1532,
+      "step": 9808
+    },
+    {
+      "epoch": 0.7544804245827244,
+      "learning_rate": 0.0004245158893730825,
+      "loss": 1.181,
+      "step": 9809
+    },
+    {
+      "epoch": 0.7545573417429429,
+      "learning_rate": 0.00042426325319126534,
+      "loss": 1.0649,
+      "step": 9810
+    },
+    {
+      "epoch": 0.7546342589031613,
+      "learning_rate": 0.00042401067982283903,
+      "loss": 1.1536,
+      "step": 9811
+    },
+    {
+      "epoch": 0.7547111760633798,
+      "learning_rate": 0.0004237581692825494,
+      "loss": 1.0421,
+      "step": 9812
+    },
+    {
+      "epoch": 0.7547880932235982,
+      "learning_rate": 0.0004235057215851425,
+      "loss": 1.1205,
+      "step": 9813
+    },
+    {
+      "epoch": 0.7548650103838166,
+      "learning_rate": 0.0004232533367453584,
+      "loss": 0.8551,
+      "step": 9814
+    },
+    {
+      "epoch": 0.7549419275440351,
+      "learning_rate": 0.0004230010147779341,
+      "loss": 0.9364,
+      "step": 9815
+    },
+    {
+      "epoch": 0.7550188447042535,
+      "learning_rate": 0.0004227487556976028,
+      "loss": 1.0157,
+      "step": 9816
+    },
+    {
+      "epoch": 0.7550957618644719,
+      "learning_rate": 0.0004224965595190946,
+      "loss": 1.1261,
+      "step": 9817
+    },
+    {
+      "epoch": 0.7551726790246904,
+      "learning_rate": 0.00042224442625713474,
+      "loss": 1.1662,
+      "step": 9818
+    },
+    {
+      "epoch": 0.7552495961849088,
+      "learning_rate": 0.00042199235592644635,
+      "loss": 1.5206,
+      "step": 9819
+    },
+    {
+      "epoch": 0.7553265133451273,
+      "learning_rate": 0.00042174034854174775,
+      "loss": 0.9527,
+      "step": 9820
+    },
+    {
+      "epoch": 0.7554034305053458,
+      "learning_rate": 0.00042148840411775415,
+      "loss": 0.7728,
+      "step": 9821
+    },
+    {
+      "epoch": 0.7554803476655642,
+      "learning_rate": 0.00042123652266917657,
+      "loss": 0.9655,
+      "step": 9822
+    },
+    {
+      "epoch": 0.7555572648257827,
+      "learning_rate": 0.0004209847042107221,
+      "loss": 0.8613,
+      "step": 9823
+    },
+    {
+      "epoch": 0.7556341819860011,
+      "learning_rate": 0.0004207329487570963,
+      "loss": 1.3195,
+      "step": 9824
+    },
+    {
+      "epoch": 0.7557110991462195,
+      "learning_rate": 0.0004204812563229978,
+      "loss": 1.0556,
+      "step": 9825
+    },
+    {
+      "epoch": 0.755788016306438,
+      "learning_rate": 0.0004202296269231235,
+      "loss": 1.2988,
+      "step": 9826
+    },
+    {
+      "epoch": 0.7558649334666564,
+      "learning_rate": 0.00041997806057216707,
+      "loss": 1.2244,
+      "step": 9827
+    },
+    {
+      "epoch": 0.7559418506268748,
+      "learning_rate": 0.0004197265572848172,
+      "loss": 1.0913,
+      "step": 9828
+    },
+    {
+      "epoch": 0.7560187677870933,
+      "learning_rate": 0.00041947511707575934,
+      "loss": 0.9811,
+      "step": 9829
+    },
+    {
+      "epoch": 0.7560956849473117,
+      "learning_rate": 0.0004192237399596757,
+      "loss": 0.7383,
+      "step": 9830
+    },
+    {
+      "epoch": 0.7561726021075302,
+      "learning_rate": 0.00041897242595124366,
+      "loss": 1.1587,
+      "step": 9831
+    },
+    {
+      "epoch": 0.7562495192677486,
+      "learning_rate": 0.00041872117506513886,
+      "loss": 0.8737,
+      "step": 9832
+    },
+    {
+      "epoch": 0.756326436427967,
+      "learning_rate": 0.00041846998731603197,
+      "loss": 0.924,
+      "step": 9833
+    },
+    {
+      "epoch": 0.7564033535881856,
+      "learning_rate": 0.00041821886271858867,
+      "loss": 1.2262,
+      "step": 9834
+    },
+    {
+      "epoch": 0.756480270748404,
+      "learning_rate": 0.00041796780128747386,
+      "loss": 1.234,
+      "step": 9835
+    },
+    {
+      "epoch": 0.7565571879086224,
+      "learning_rate": 0.00041771680303734685,
+      "loss": 0.8367,
+      "step": 9836
+    },
+    {
+      "epoch": 0.7566341050688409,
+      "learning_rate": 0.00041746586798286354,
+      "loss": 0.9922,
+      "step": 9837
+    },
+    {
+      "epoch": 0.7567110222290593,
+      "learning_rate": 0.00041721499613867654,
+      "loss": 1.1963,
+      "step": 9838
+    },
+    {
+      "epoch": 0.7567879393892778,
+      "learning_rate": 0.00041696418751943366,
+      "loss": 1.0821,
+      "step": 9839
+    },
+    {
+      "epoch": 0.7568648565494962,
+      "learning_rate": 0.0004167134421397812,
+      "loss": 1.1669,
+      "step": 9840
+    },
+    {
+      "epoch": 0.7569417737097146,
+      "learning_rate": 0.0004164627600143599,
+      "loss": 1.3612,
+      "step": 9841
+    },
+    {
+      "epoch": 0.7570186908699331,
+      "learning_rate": 0.00041621214115780724,
+      "loss": 1.1686,
+      "step": 9842
+    },
+    {
+      "epoch": 0.7570956080301515,
+      "learning_rate": 0.0004159615855847571,
+      "loss": 0.9845,
+      "step": 9843
+    },
+    {
+      "epoch": 0.7571725251903699,
+      "learning_rate": 0.00041571109330983976,
+      "loss": 1.3389,
+      "step": 9844
+    },
+    {
+      "epoch": 0.7572494423505884,
+      "learning_rate": 0.0004154606643476811,
+      "loss": 1.1986,
+      "step": 9845
+    },
+    {
+      "epoch": 0.7573263595108068,
+      "learning_rate": 0.0004152102987129057,
+      "loss": 1.526,
+      "step": 9846
+    },
+    {
+      "epoch": 0.7574032766710254,
+      "learning_rate": 0.00041495999642013044,
+      "loss": 0.8463,
+      "step": 9847
+    },
+    {
+      "epoch": 0.7574801938312438,
+      "learning_rate": 0.0004147097574839723,
+      "loss": 1.1937,
+      "step": 9848
+    },
+    {
+      "epoch": 0.7575571109914622,
+      "learning_rate": 0.00041445958191904236,
+      "loss": 1.1596,
+      "step": 9849
+    },
+    {
+      "epoch": 0.7576340281516807,
+      "learning_rate": 0.0004142094697399489,
+      "loss": 0.7386,
+      "step": 9850
+    },
+    {
+      "epoch": 0.7577109453118991,
+      "learning_rate": 0.00041395942096129613,
+      "loss": 1.2507,
+      "step": 9851
+    },
+    {
+      "epoch": 0.7577878624721175,
+      "learning_rate": 0.00041370943559768445,
+      "loss": 1.2967,
+      "step": 9852
+    },
+    {
+      "epoch": 0.757864779632336,
+      "learning_rate": 0.0004134595136637106,
+      "loss": 1.1242,
+      "step": 9853
+    },
+    {
+      "epoch": 0.7579416967925544,
+      "learning_rate": 0.00041320965517396846,
+      "loss": 1.9654,
+      "step": 9854
+    },
+    {
+      "epoch": 0.7580186139527728,
+      "learning_rate": 0.00041295986014304773,
+      "loss": 1.4288,
+      "step": 9855
+    },
+    {
+      "epoch": 0.7580955311129913,
+      "learning_rate": 0.0004127101285855326,
+      "loss": 1.4664,
+      "step": 9856
+    },
+    {
+      "epoch": 0.7581724482732097,
+      "learning_rate": 0.00041246046051600675,
+      "loss": 1.5966,
+      "step": 9857
+    },
+    {
+      "epoch": 0.7582493654334282,
+      "learning_rate": 0.00041221085594904763,
+      "loss": 1.1626,
+      "step": 9858
+    },
+    {
+      "epoch": 0.7583262825936467,
+      "learning_rate": 0.00041196131489923117,
+      "loss": 0.9965,
+      "step": 9859
+    },
+    {
+      "epoch": 0.758403199753865,
+      "learning_rate": 0.00041171183738112694,
+      "loss": 1.3445,
+      "step": 9860
+    },
+    {
+      "epoch": 0.7584801169140836,
+      "learning_rate": 0.0004114624234093023,
+      "loss": 1.1673,
+      "step": 9861
+    },
+    {
+      "epoch": 0.758557034074302,
+      "learning_rate": 0.00041121307299832175,
+      "loss": 1.1812,
+      "step": 9862
+    },
+    {
+      "epoch": 0.7586339512345204,
+      "learning_rate": 0.0004109637861627444,
+      "loss": 1.6752,
+      "step": 9863
+    },
+    {
+      "epoch": 0.7587108683947389,
+      "learning_rate": 0.0004107145629171266,
+      "loss": 1.0776,
+      "step": 9864
+    },
+    {
+      "epoch": 0.7587877855549573,
+      "learning_rate": 0.00041046540327602075,
+      "loss": 1.1185,
+      "step": 9865
+    },
+    {
+      "epoch": 0.7588647027151758,
+      "learning_rate": 0.00041021630725397495,
+      "loss": 0.8098,
+      "step": 9866
+    },
+    {
+      "epoch": 0.7589416198753942,
+      "learning_rate": 0.00040996727486553535,
+      "loss": 1.1833,
+      "step": 9867
+    },
+    {
+      "epoch": 0.7590185370356126,
+      "learning_rate": 0.0004097183061252429,
+      "loss": 1.0936,
+      "step": 9868
+    },
+    {
+      "epoch": 0.7590954541958311,
+      "learning_rate": 0.000409469401047634,
+      "loss": 1.0916,
+      "step": 9869
+    },
+    {
+      "epoch": 0.7591723713560495,
+      "learning_rate": 0.0004092205596472436,
+      "loss": 1.0741,
+      "step": 9870
+    },
+    {
+      "epoch": 0.7592492885162679,
+      "learning_rate": 0.0004089717819386016,
+      "loss": 1.1404,
+      "step": 9871
+    },
+    {
+      "epoch": 0.7593262056764865,
+      "learning_rate": 0.00040872306793623416,
+      "loss": 1.2892,
+      "step": 9872
+    },
+    {
+      "epoch": 0.7594031228367049,
+      "learning_rate": 0.0004084744176546642,
+      "loss": 1.1543,
+      "step": 9873
+    },
+    {
+      "epoch": 0.7594800399969233,
+      "learning_rate": 0.0004082258311084106,
+      "loss": 1.1737,
+      "step": 9874
+    },
+    {
+      "epoch": 0.7595569571571418,
+      "learning_rate": 0.0004079773083119881,
+      "loss": 1.3703,
+      "step": 9875
+    },
+    {
+      "epoch": 0.7596338743173602,
+      "learning_rate": 0.00040772884927990904,
+      "loss": 1.1074,
+      "step": 9876
+    },
+    {
+      "epoch": 0.7597107914775787,
+      "learning_rate": 0.0004074804540266807,
+      "loss": 0.8953,
+      "step": 9877
+    },
+    {
+      "epoch": 0.7597877086377971,
+      "learning_rate": 0.0004072321225668075,
+      "loss": 1.2146,
+      "step": 9878
+    },
+    {
+      "epoch": 0.7598646257980155,
+      "learning_rate": 0.0004069838549147892,
+      "loss": 0.986,
+      "step": 9879
+    },
+    {
+      "epoch": 0.759941542958234,
+      "learning_rate": 0.0004067356510851226,
+      "loss": 1.4233,
+      "step": 9880
+    },
+    {
+      "epoch": 0.7600184601184524,
+      "learning_rate": 0.0004064875110923014,
+      "loss": 0.9556,
+      "step": 9881
+    },
+    {
+      "epoch": 0.7600953772786708,
+      "learning_rate": 0.00040623943495081356,
+      "loss": 1.2152,
+      "step": 9882
+    },
+    {
+      "epoch": 0.7601722944388893,
+      "learning_rate": 0.0004059914226751447,
+      "loss": 1.191,
+      "step": 9883
+    },
+    {
+      "epoch": 0.7602492115991077,
+      "learning_rate": 0.00040574347427977726,
+      "loss": 0.8316,
+      "step": 9884
+    },
+    {
+      "epoch": 0.7603261287593263,
+      "learning_rate": 0.0004054955897791888,
+      "loss": 1.2691,
+      "step": 9885
+    },
+    {
+      "epoch": 0.7604030459195447,
+      "learning_rate": 0.00040524776918785335,
+      "loss": 1.0548,
+      "step": 9886
+    },
+    {
+      "epoch": 0.7604799630797631,
+      "learning_rate": 0.0004050000125202417,
+      "loss": 0.7575,
+      "step": 9887
+    },
+    {
+      "epoch": 0.7605568802399816,
+      "learning_rate": 0.00040475231979082,
+      "loss": 0.9693,
+      "step": 9888
+    },
+    {
+      "epoch": 0.7606337974002,
+      "learning_rate": 0.00040450469101405227,
+      "loss": 1.1548,
+      "step": 9889
+    },
+    {
+      "epoch": 0.7607107145604184,
+      "learning_rate": 0.00040425712620439767,
+      "loss": 1.2662,
+      "step": 9890
+    },
+    {
+      "epoch": 0.7607876317206369,
+      "learning_rate": 0.0004040096253763106,
+      "loss": 1.2688,
+      "step": 9891
+    },
+    {
+      "epoch": 0.7608645488808553,
+      "learning_rate": 0.00040376218854424424,
+      "loss": 1.1372,
+      "step": 9892
+    },
+    {
+      "epoch": 0.7609414660410737,
+      "learning_rate": 0.0004035148157226463,
+      "loss": 0.8423,
+      "step": 9893
+    },
+    {
+      "epoch": 0.7610183832012922,
+      "learning_rate": 0.00040326750692596064,
+      "loss": 1.0708,
+      "step": 9894
+    },
+    {
+      "epoch": 0.7610953003615106,
+      "learning_rate": 0.0004030202621686285,
+      "loss": 1.1025,
+      "step": 9895
+    },
+    {
+      "epoch": 0.7611722175217291,
+      "learning_rate": 0.000402773081465086,
+      "loss": 1.1715,
+      "step": 9896
+    },
+    {
+      "epoch": 0.7612491346819475,
+      "learning_rate": 0.00040252596482976706,
+      "loss": 1.245,
+      "step": 9897
+    },
+    {
+      "epoch": 0.761326051842166,
+      "learning_rate": 0.000402278912277101,
+      "loss": 1.3009,
+      "step": 9898
+    },
+    {
+      "epoch": 0.7614029690023845,
+      "learning_rate": 0.00040203192382151317,
+      "loss": 1.3432,
+      "step": 9899
+    },
+    {
+      "epoch": 0.7614798861626029,
+      "learning_rate": 0.00040178499947742566,
+      "loss": 1.1632,
+      "step": 9900
+    },
+    {
+      "epoch": 0.7615568033228213,
+      "learning_rate": 0.00040153813925925653,
+      "loss": 1.2224,
+      "step": 9901
+    },
+    {
+      "epoch": 0.7616337204830398,
+      "learning_rate": 0.0004012913431814196,
+      "loss": 1.1971,
+      "step": 9902
+    },
+    {
+      "epoch": 0.7617106376432582,
+      "learning_rate": 0.0004010446112583274,
+      "loss": 1.3722,
+      "step": 9903
+    },
+    {
+      "epoch": 0.7617875548034767,
+      "learning_rate": 0.00040079794350438453,
+      "loss": 0.9624,
+      "step": 9904
+    },
+    {
+      "epoch": 0.7618644719636951,
+      "learning_rate": 0.0004005513399339958,
+      "loss": 1.1057,
+      "step": 9905
+    },
+    {
+      "epoch": 0.7619413891239135,
+      "learning_rate": 0.00040030480056156017,
+      "loss": 1.4653,
+      "step": 9906
+    },
+    {
+      "epoch": 0.762018306284132,
+      "learning_rate": 0.0004000583254014733,
+      "loss": 1.0507,
+      "step": 9907
+    },
+    {
+      "epoch": 0.7620952234443504,
+      "learning_rate": 0.0003998119144681272,
+      "loss": 1.3143,
+      "step": 9908
+    },
+    {
+      "epoch": 0.7621721406045688,
+      "learning_rate": 0.00039956556777591,
+      "loss": 1.2349,
+      "step": 9909
+    },
+    {
+      "epoch": 0.7622490577647874,
+      "learning_rate": 0.0003993192853392056,
+      "loss": 1.2431,
+      "step": 9910
+    },
+    {
+      "epoch": 0.7623259749250058,
+      "learning_rate": 0.00039907306717239574,
+      "loss": 1.3292,
+      "step": 9911
+    },
+    {
+      "epoch": 0.7624028920852242,
+      "learning_rate": 0.0003988269132898573,
+      "loss": 1.0823,
+      "step": 9912
+    },
+    {
+      "epoch": 0.7624798092454427,
+      "learning_rate": 0.00039858082370596205,
+      "loss": 1.2541,
+      "step": 9913
+    },
+    {
+      "epoch": 0.7625567264056611,
+      "learning_rate": 0.00039833479843508093,
+      "loss": 0.9036,
+      "step": 9914
+    },
+    {
+      "epoch": 0.7626336435658796,
+      "learning_rate": 0.00039808883749157853,
+      "loss": 1.2089,
+      "step": 9915
+    },
+    {
+      "epoch": 0.762710560726098,
+      "learning_rate": 0.00039784294088981824,
+      "loss": 1.0709,
+      "step": 9916
+    },
+    {
+      "epoch": 0.7627874778863164,
+      "learning_rate": 0.0003975971086441568,
+      "loss": 1.2823,
+      "step": 9917
+    },
+    {
+      "epoch": 0.7628643950465349,
+      "learning_rate": 0.0003973513407689486,
+      "loss": 1.1153,
+      "step": 9918
+    },
+    {
+      "epoch": 0.7629413122067533,
+      "learning_rate": 0.00039710563727854523,
+      "loss": 1.0983,
+      "step": 9919
+    },
+    {
+      "epoch": 0.7630182293669717,
+      "learning_rate": 0.0003968599981872932,
+      "loss": 1.4162,
+      "step": 9920
+    },
+    {
+      "epoch": 0.7630951465271902,
+      "learning_rate": 0.0003966144235095354,
+      "loss": 1.1128,
+      "step": 9921
+    },
+    {
+      "epoch": 0.7631720636874086,
+      "learning_rate": 0.00039636891325961146,
+      "loss": 1.2088,
+      "step": 9922
+    },
+    {
+      "epoch": 0.7632489808476272,
+      "learning_rate": 0.00039612346745185626,
+      "loss": 1.3528,
+      "step": 9923
+    },
+    {
+      "epoch": 0.7633258980078456,
+      "learning_rate": 0.0003958780861006028,
+      "loss": 0.713,
+      "step": 9924
+    },
+    {
+      "epoch": 0.763402815168064,
+      "learning_rate": 0.0003956327692201789,
+      "loss": 1.2216,
+      "step": 9925
+    },
+    {
+      "epoch": 0.7634797323282825,
+      "learning_rate": 0.0003953875168249078,
+      "loss": 0.7214,
+      "step": 9926
+    },
+    {
+      "epoch": 0.7635566494885009,
+      "learning_rate": 0.00039514232892911113,
+      "loss": 0.9942,
+      "step": 9927
+    },
+    {
+      "epoch": 0.7636335666487193,
+      "learning_rate": 0.0003948972055471054,
+      "loss": 0.9765,
+      "step": 9928
+    },
+    {
+      "epoch": 0.7637104838089378,
+      "learning_rate": 0.00039465214669320354,
+      "loss": 1.1006,
+      "step": 9929
+    },
+    {
+      "epoch": 0.7637874009691562,
+      "learning_rate": 0.00039440715238171477,
+      "loss": 1.1188,
+      "step": 9930
+    },
+    {
+      "epoch": 0.7638643181293746,
+      "learning_rate": 0.00039416222262694464,
+      "loss": 1.2701,
+      "step": 9931
+    },
+    {
+      "epoch": 0.7639412352895931,
+      "learning_rate": 0.0003939173574431943,
+      "loss": 1.4445,
+      "step": 9932
+    },
+    {
+      "epoch": 0.7640181524498115,
+      "learning_rate": 0.00039367255684476265,
+      "loss": 1.0762,
+      "step": 9933
+    },
+    {
+      "epoch": 0.76409506961003,
+      "learning_rate": 0.0003934278208459435,
+      "loss": 1.1465,
+      "step": 9934
+    },
+    {
+      "epoch": 0.7641719867702484,
+      "learning_rate": 0.000393183149461027,
+      "loss": 0.9773,
+      "step": 9935
+    },
+    {
+      "epoch": 0.7642489039304668,
+      "learning_rate": 0.0003929385427043,
+      "loss": 1.2645,
+      "step": 9936
+    },
+    {
+      "epoch": 0.7643258210906854,
+      "learning_rate": 0.00039269400059004474,
+      "loss": 1.2645,
+      "step": 9937
+    },
+    {
+      "epoch": 0.7644027382509038,
+      "learning_rate": 0.0003924495231325416,
+      "loss": 0.9997,
+      "step": 9938
+    },
+    {
+      "epoch": 0.7644796554111222,
+      "learning_rate": 0.00039220511034606477,
+      "loss": 1.63,
+      "step": 9939
+    },
+    {
+      "epoch": 0.7645565725713407,
+      "learning_rate": 0.0003919607622448856,
+      "loss": 1.1389,
+      "step": 9940
+    },
+    {
+      "epoch": 0.7646334897315591,
+      "learning_rate": 0.0003917164788432726,
+      "loss": 1.1261,
+      "step": 9941
+    },
+    {
+      "epoch": 0.7647104068917776,
+      "learning_rate": 0.0003914722601554894,
+      "loss": 1.0346,
+      "step": 9942
+    },
+    {
+      "epoch": 0.764787324051996,
+      "learning_rate": 0.0003912281061957965,
+      "loss": 1.2263,
+      "step": 9943
+    },
+    {
+      "epoch": 0.7648642412122144,
+      "learning_rate": 0.0003909840169784498,
+      "loss": 1.0811,
+      "step": 9944
+    },
+    {
+      "epoch": 0.7649411583724329,
+      "learning_rate": 0.00039073999251770173,
+      "loss": 1.1438,
+      "step": 9945
+    },
+    {
+      "epoch": 0.7650180755326513,
+      "learning_rate": 0.0003904960328278019,
+      "loss": 1.187,
+      "step": 9946
+    },
+    {
+      "epoch": 0.7650949926928697,
+      "learning_rate": 0.00039025213792299513,
+      "loss": 1.1532,
+      "step": 9947
+    },
+    {
+      "epoch": 0.7651719098530883,
+      "learning_rate": 0.0003900083078175226,
+      "loss": 1.0514,
+      "step": 9948
+    },
+    {
+      "epoch": 0.7652488270133067,
+      "learning_rate": 0.0003897645425256216,
+      "loss": 0.8389,
+      "step": 9949
+    },
+    {
+      "epoch": 0.7653257441735252,
+      "learning_rate": 0.00038952084206152573,
+      "loss": 1.209,
+      "step": 9950
+    },
+    {
+      "epoch": 0.7654026613337436,
+      "learning_rate": 0.0003892772064394657,
+      "loss": 0.9865,
+      "step": 9951
+    },
+    {
+      "epoch": 0.765479578493962,
+      "learning_rate": 0.00038903363567366746,
+      "loss": 1.1961,
+      "step": 9952
+    },
+    {
+      "epoch": 0.7655564956541805,
+      "learning_rate": 0.0003887901297783523,
+      "loss": 1.4159,
+      "step": 9953
+    },
+    {
+      "epoch": 0.7656334128143989,
+      "learning_rate": 0.0003885466887677398,
+      "loss": 1.653,
+      "step": 9954
+    },
+    {
+      "epoch": 0.7657103299746173,
+      "learning_rate": 0.00038830331265604466,
+      "loss": 0.9447,
+      "step": 9955
+    },
+    {
+      "epoch": 0.7657872471348358,
+      "learning_rate": 0.0003880600014574776,
+      "loss": 1.2132,
+      "step": 9956
+    },
+    {
+      "epoch": 0.7658641642950542,
+      "learning_rate": 0.00038781675518624587,
+      "loss": 1.2148,
+      "step": 9957
+    },
+    {
+      "epoch": 0.7659410814552726,
+      "learning_rate": 0.00038757357385655293,
+      "loss": 1.2414,
+      "step": 9958
+    },
+    {
+      "epoch": 0.7660179986154911,
+      "learning_rate": 0.00038733045748259783,
+      "loss": 1.5285,
+      "step": 9959
+    },
+    {
+      "epoch": 0.7660949157757095,
+      "learning_rate": 0.00038708740607857744,
+      "loss": 1.4117,
+      "step": 9960
+    },
+    {
+      "epoch": 0.766171832935928,
+      "learning_rate": 0.00038684441965868336,
+      "loss": 1.2747,
+      "step": 9961
+    },
+    {
+      "epoch": 0.7662487500961465,
+      "learning_rate": 0.00038660149823710346,
+      "loss": 0.9704,
+      "step": 9962
+    },
+    {
+      "epoch": 0.7663256672563649,
+      "learning_rate": 0.0003863586418280226,
+      "loss": 1.1634,
+      "step": 9963
+    },
+    {
+      "epoch": 0.7664025844165834,
+      "learning_rate": 0.0003861158504456208,
+      "loss": 0.9502,
+      "step": 9964
+    },
+    {
+      "epoch": 0.7664795015768018,
+      "learning_rate": 0.0003858731241040762,
+      "loss": 1.151,
+      "step": 9965
+    },
+    {
+      "epoch": 0.7665564187370202,
+      "learning_rate": 0.0003856304628175606,
+      "loss": 1.2631,
+      "step": 9966
+    },
+    {
+      "epoch": 0.7666333358972387,
+      "learning_rate": 0.0003853878666002431,
+      "loss": 1.0425,
+      "step": 9967
+    },
+    {
+      "epoch": 0.7667102530574571,
+      "learning_rate": 0.00038514533546629026,
+      "loss": 0.8835,
+      "step": 9968
+    },
+    {
+      "epoch": 0.7667871702176756,
+      "learning_rate": 0.00038490286942986313,
+      "loss": 1.1304,
+      "step": 9969
+    },
+    {
+      "epoch": 0.766864087377894,
+      "learning_rate": 0.00038466046850511957,
+      "loss": 1.0838,
+      "step": 9970
+    },
+    {
+      "epoch": 0.7669410045381124,
+      "learning_rate": 0.00038441813270621367,
+      "loss": 0.9887,
+      "step": 9971
+    },
+    {
+      "epoch": 0.7670179216983309,
+      "learning_rate": 0.00038417586204729504,
+      "loss": 1.152,
+      "step": 9972
+    },
+    {
+      "epoch": 0.7670948388585493,
+      "learning_rate": 0.0003839336565425113,
+      "loss": 1.3359,
+      "step": 9973
+    },
+    {
+      "epoch": 0.7671717560187677,
+      "learning_rate": 0.0003836915162060049,
+      "loss": 1.1356,
+      "step": 9974
+    },
+    {
+      "epoch": 0.7672486731789863,
+      "learning_rate": 0.0003834494410519134,
+      "loss": 0.9708,
+      "step": 9975
+    },
+    {
+      "epoch": 0.7673255903392047,
+      "learning_rate": 0.00038320743109437314,
+      "loss": 1.2937,
+      "step": 9976
+    },
+    {
+      "epoch": 0.7674025074994231,
+      "learning_rate": 0.00038296548634751465,
+      "loss": 1.3575,
+      "step": 9977
+    },
+    {
+      "epoch": 0.7674794246596416,
+      "learning_rate": 0.00038272360682546576,
+      "loss": 1.1872,
+      "step": 9978
+    },
+    {
+      "epoch": 0.76755634181986,
+      "learning_rate": 0.0003824817925423498,
+      "loss": 1.1093,
+      "step": 9979
+    },
+    {
+      "epoch": 0.7676332589800785,
+      "learning_rate": 0.0003822400435122863,
+      "loss": 1.3454,
+      "step": 9980
+    },
+    {
+      "epoch": 0.7677101761402969,
+      "learning_rate": 0.0003819983597493919,
+      "loss": 1.2304,
+      "step": 9981
+    },
+    {
+      "epoch": 0.7677870933005153,
+      "learning_rate": 0.0003817567412677784,
+      "loss": 1.2618,
+      "step": 9982
+    },
+    {
+      "epoch": 0.7678640104607338,
+      "learning_rate": 0.0003815151880815542,
+      "loss": 0.9594,
+      "step": 9983
+    },
+    {
+      "epoch": 0.7679409276209522,
+      "learning_rate": 0.0003812737002048238,
+      "loss": 1.2799,
+      "step": 9984
+    },
+    {
+      "epoch": 0.7680178447811706,
+      "learning_rate": 0.0003810322776516881,
+      "loss": 1.0407,
+      "step": 9985
+    },
+    {
+      "epoch": 0.7680947619413891,
+      "learning_rate": 0.0003807909204362432,
+      "loss": 1.2365,
+      "step": 9986
+    },
+    {
+      "epoch": 0.7681716791016076,
+      "learning_rate": 0.0003805496285725839,
+      "loss": 1.1432,
+      "step": 9987
+    },
+    {
+      "epoch": 0.7682485962618261,
+      "learning_rate": 0.00038030840207479766,
+      "loss": 1.511,
+      "step": 9988
+    },
+    {
+      "epoch": 0.7683255134220445,
+      "learning_rate": 0.00038006724095697106,
+      "loss": 1.3152,
+      "step": 9989
+    },
+    {
+      "epoch": 0.7684024305822629,
+      "learning_rate": 0.00037982614523318544,
+      "loss": 1.0888,
+      "step": 9990
+    },
+    {
+      "epoch": 0.7684793477424814,
+      "learning_rate": 0.0003795851149175186,
+      "loss": 1.0369,
+      "step": 9991
+    },
+    {
+      "epoch": 0.7685562649026998,
+      "learning_rate": 0.00037934415002404456,
+      "loss": 1.7193,
+      "step": 9992
+    },
+    {
+      "epoch": 0.7686331820629182,
+      "learning_rate": 0.00037910325056683344,
+      "loss": 0.9679,
+      "step": 9993
+    },
+    {
+      "epoch": 0.7687100992231367,
+      "learning_rate": 0.0003788624165599513,
+      "loss": 0.9817,
+      "step": 9994
+    },
+    {
+      "epoch": 0.7687870163833551,
+      "learning_rate": 0.00037862164801746157,
+      "loss": 1.168,
+      "step": 9995
+    },
+    {
+      "epoch": 0.7688639335435735,
+      "learning_rate": 0.000378380944953423,
+      "loss": 1.3463,
+      "step": 9996
+    },
+    {
+      "epoch": 0.768940850703792,
+      "learning_rate": 0.00037814030738188905,
+      "loss": 1.5224,
+      "step": 9997
+    },
+    {
+      "epoch": 0.7690177678640104,
+      "learning_rate": 0.0003778997353169121,
+      "loss": 1.3013,
+      "step": 9998
+    },
+    {
+      "epoch": 0.769094685024229,
+      "learning_rate": 0.00037765922877253853,
+      "loss": 1.2762,
+      "step": 9999
+    },
+    {
+      "epoch": 0.7691716021844474,
+      "learning_rate": 0.00037741878776281323,
+      "loss": 1.1189,
+      "step": 10000
+    },
+    {
+      "epoch": 0.7692485193446658,
+      "learning_rate": 0.0003771784123017744,
+      "loss": 0.9666,
+      "step": 10001
+    },
+    {
+      "epoch": 0.7693254365048843,
+      "learning_rate": 0.00037693810240345784,
+      "loss": 1.1382,
+      "step": 10002
+    },
+    {
+      "epoch": 0.7694023536651027,
+      "learning_rate": 0.00037669785808189645,
+      "loss": 1.1876,
+      "step": 10003
+    },
+    {
+      "epoch": 0.7694792708253211,
+      "learning_rate": 0.00037645767935111753,
+      "loss": 1.2285,
+      "step": 10004
+    },
+    {
+      "epoch": 0.7695561879855396,
+      "learning_rate": 0.00037621756622514585,
+      "loss": 1.4077,
+      "step": 10005
+    },
+    {
+      "epoch": 0.769633105145758,
+      "learning_rate": 0.00037597751871800154,
+      "loss": 0.8036,
+      "step": 10006
+    },
+    {
+      "epoch": 0.7697100223059765,
+      "learning_rate": 0.0003757375368437011,
+      "loss": 1.0401,
+      "step": 10007
+    },
+    {
+      "epoch": 0.7697869394661949,
+      "learning_rate": 0.00037549762061625774,
+      "loss": 1.4036,
+      "step": 10008
+    },
+    {
+      "epoch": 0.7698638566264133,
+      "learning_rate": 0.0003752577700496808,
+      "loss": 1.0638,
+      "step": 10009
+    },
+    {
+      "epoch": 0.7699407737866318,
+      "learning_rate": 0.0003750179851579741,
+      "loss": 1.105,
+      "step": 10010
+    },
+    {
+      "epoch": 0.7700176909468502,
+      "learning_rate": 0.00037477826595513984,
+      "loss": 1.0871,
+      "step": 10011
+    },
+    {
+      "epoch": 0.7700946081070686,
+      "learning_rate": 0.00037453861245517544,
+      "loss": 0.9914,
+      "step": 10012
+    },
+    {
+      "epoch": 0.7701715252672872,
+      "learning_rate": 0.00037429902467207447,
+      "loss": 0.9106,
+      "step": 10013
+    },
+    {
+      "epoch": 0.7702484424275056,
+      "learning_rate": 0.00037405950261982645,
+      "loss": 1.1255,
+      "step": 10014
+    },
+    {
+      "epoch": 0.770325359587724,
+      "learning_rate": 0.0003738200463124177,
+      "loss": 1.3214,
+      "step": 10015
+    },
+    {
+      "epoch": 0.7704022767479425,
+      "learning_rate": 0.0003735806557638296,
+      "loss": 1.2187,
+      "step": 10016
+    },
+    {
+      "epoch": 0.7704791939081609,
+      "learning_rate": 0.00037334133098804124,
+      "loss": 1.4611,
+      "step": 10017
+    },
+    {
+      "epoch": 0.7705561110683794,
+      "learning_rate": 0.0003731020719990269,
+      "loss": 1.0869,
+      "step": 10018
+    },
+    {
+      "epoch": 0.7706330282285978,
+      "learning_rate": 0.0003728628788107569,
+      "loss": 1.4927,
+      "step": 10019
+    },
+    {
+      "epoch": 0.7707099453888162,
+      "learning_rate": 0.0003726237514371983,
+      "loss": 1.0439,
+      "step": 10020
+    },
+    {
+      "epoch": 0.7707868625490347,
+      "learning_rate": 0.00037238468989231316,
+      "loss": 0.9544,
+      "step": 10021
+    },
+    {
+      "epoch": 0.7708637797092531,
+      "learning_rate": 0.00037214569419006204,
+      "loss": 1.4451,
+      "step": 10022
+    },
+    {
+      "epoch": 0.7709406968694715,
+      "learning_rate": 0.0003719067643443988,
+      "loss": 1.0804,
+      "step": 10023
+    },
+    {
+      "epoch": 0.77101761402969,
+      "learning_rate": 0.00037166790036927475,
+      "loss": 1.1747,
+      "step": 10024
+    },
+    {
+      "epoch": 0.7710945311899084,
+      "learning_rate": 0.0003714291022786385,
+      "loss": 1.2461,
+      "step": 10025
+    },
+    {
+      "epoch": 0.771171448350127,
+      "learning_rate": 0.00037119037008643305,
+      "loss": 1.1186,
+      "step": 10026
+    },
+    {
+      "epoch": 0.7712483655103454,
+      "learning_rate": 0.0003709517038065984,
+      "loss": 0.9967,
+      "step": 10027
+    },
+    {
+      "epoch": 0.7713252826705638,
+      "learning_rate": 0.00037071310345307044,
+      "loss": 1.1569,
+      "step": 10028
+    },
+    {
+      "epoch": 0.7714021998307823,
+      "learning_rate": 0.00037047456903978066,
+      "loss": 1.2187,
+      "step": 10029
+    },
+    {
+      "epoch": 0.7714791169910007,
+      "learning_rate": 0.0003702361005806585,
+      "loss": 1.0911,
+      "step": 10030
+    },
+    {
+      "epoch": 0.7715560341512191,
+      "learning_rate": 0.0003699976980896284,
+      "loss": 1.1432,
+      "step": 10031
+    },
+    {
+      "epoch": 0.7716329513114376,
+      "learning_rate": 0.0003697593615806093,
+      "loss": 1.4127,
+      "step": 10032
+    },
+    {
+      "epoch": 0.771709868471656,
+      "learning_rate": 0.00036952109106751956,
+      "loss": 0.9756,
+      "step": 10033
+    },
+    {
+      "epoch": 0.7717867856318744,
+      "learning_rate": 0.00036928288656427145,
+      "loss": 0.9238,
+      "step": 10034
+    },
+    {
+      "epoch": 0.7718637027920929,
+      "learning_rate": 0.00036904474808477394,
+      "loss": 1.3014,
+      "step": 10035
+    },
+    {
+      "epoch": 0.7719406199523113,
+      "learning_rate": 0.0003688066756429323,
+      "loss": 0.9536,
+      "step": 10036
+    },
+    {
+      "epoch": 0.7720175371125299,
+      "learning_rate": 0.0003685686692526472,
+      "loss": 1.2614,
+      "step": 10037
+    },
+    {
+      "epoch": 0.7720944542727483,
+      "learning_rate": 0.00036833072892781724,
+      "loss": 1.1682,
+      "step": 10038
+    },
+    {
+      "epoch": 0.7721713714329667,
+      "learning_rate": 0.00036809285468233536,
+      "loss": 1.233,
+      "step": 10039
+    },
+    {
+      "epoch": 0.7722482885931852,
+      "learning_rate": 0.0003678550465300913,
+      "loss": 1.2303,
+      "step": 10040
+    },
+    {
+      "epoch": 0.7723252057534036,
+      "learning_rate": 0.0003676173044849711,
+      "loss": 1.0965,
+      "step": 10041
+    },
+    {
+      "epoch": 0.772402122913622,
+      "learning_rate": 0.00036737962856085645,
+      "loss": 1.1599,
+      "step": 10042
+    },
+    {
+      "epoch": 0.7724790400738405,
+      "learning_rate": 0.00036714201877162536,
+      "loss": 0.7442,
+      "step": 10043
+    },
+    {
+      "epoch": 0.7725559572340589,
+      "learning_rate": 0.00036690447513115337,
+      "loss": 1.0247,
+      "step": 10044
+    },
+    {
+      "epoch": 0.7726328743942774,
+      "learning_rate": 0.00036666699765330895,
+      "loss": 1.0928,
+      "step": 10045
+    },
+    {
+      "epoch": 0.7727097915544958,
+      "learning_rate": 0.0003664295863519602,
+      "loss": 1.065,
+      "step": 10046
+    },
+    {
+      "epoch": 0.7727867087147142,
+      "learning_rate": 0.00036619224124096925,
+      "loss": 1.0316,
+      "step": 10047
+    },
+    {
+      "epoch": 0.7728636258749327,
+      "learning_rate": 0.0003659549623341949,
+      "loss": 1.351,
+      "step": 10048
+    },
+    {
+      "epoch": 0.7729405430351511,
+      "learning_rate": 0.00036571774964549227,
+      "loss": 1.4493,
+      "step": 10049
+    },
+    {
+      "epoch": 0.7730174601953695,
+      "learning_rate": 0.00036548060318871225,
+      "loss": 1.2986,
+      "step": 10050
+    },
+    {
+      "epoch": 0.7730943773555881,
+      "learning_rate": 0.00036524352297770174,
+      "loss": 1.2326,
+      "step": 10051
+    },
+    {
+      "epoch": 0.7731712945158065,
+      "learning_rate": 0.00036500650902630507,
+      "loss": 1.1512,
+      "step": 10052
+    },
+    {
+      "epoch": 0.7732482116760249,
+      "learning_rate": 0.0003647695613483616,
+      "loss": 1.2492,
+      "step": 10053
+    },
+    {
+      "epoch": 0.7733251288362434,
+      "learning_rate": 0.00036453267995770563,
+      "loss": 1.3303,
+      "step": 10054
+    },
+    {
+      "epoch": 0.7734020459964618,
+      "learning_rate": 0.0003642958648681704,
+      "loss": 1.3181,
+      "step": 10055
+    },
+    {
+      "epoch": 0.7734789631566803,
+      "learning_rate": 0.00036405911609358274,
+      "loss": 1.6058,
+      "step": 10056
+    },
+    {
+      "epoch": 0.7735558803168987,
+      "learning_rate": 0.00036382243364776784,
+      "loss": 1.1234,
+      "step": 10057
+    },
+    {
+      "epoch": 0.7736327974771171,
+      "learning_rate": 0.0003635858175445449,
+      "loss": 0.9767,
+      "step": 10058
+    },
+    {
+      "epoch": 0.7737097146373356,
+      "learning_rate": 0.00036334926779772996,
+      "loss": 1.3137,
+      "step": 10059
+    },
+    {
+      "epoch": 0.773786631797554,
+      "learning_rate": 0.0003631127844211362,
+      "loss": 1.3275,
+      "step": 10060
+    },
+    {
+      "epoch": 0.7738635489577724,
+      "learning_rate": 0.0003628763674285719,
+      "loss": 1.0674,
+      "step": 10061
+    },
+    {
+      "epoch": 0.7739404661179909,
+      "learning_rate": 0.00036264001683384146,
+      "loss": 1.3303,
+      "step": 10062
+    },
+    {
+      "epoch": 0.7740173832782093,
+      "learning_rate": 0.00036240373265074586,
+      "loss": 1.3485,
+      "step": 10063
+    },
+    {
+      "epoch": 0.7740943004384279,
+      "learning_rate": 0.00036216751489308145,
+      "loss": 1.1655,
+      "step": 10064
+    },
+    {
+      "epoch": 0.7741712175986463,
+      "learning_rate": 0.00036193136357464203,
+      "loss": 1.159,
+      "step": 10065
+    },
+    {
+      "epoch": 0.7742481347588647,
+      "learning_rate": 0.0003616952787092168,
+      "loss": 1.4024,
+      "step": 10066
+    },
+    {
+      "epoch": 0.7743250519190832,
+      "learning_rate": 0.0003614592603105898,
+      "loss": 0.7212,
+      "step": 10067
+    },
+    {
+      "epoch": 0.7744019690793016,
+      "learning_rate": 0.0003612233083925436,
+      "loss": 1.4691,
+      "step": 10068
+    },
+    {
+      "epoch": 0.77447888623952,
+      "learning_rate": 0.000360987422968855,
+      "loss": 0.9413,
+      "step": 10069
+    },
+    {
+      "epoch": 0.7745558033997385,
+      "learning_rate": 0.0003607516040532981,
+      "loss": 1.1579,
+      "step": 10070
+    },
+    {
+      "epoch": 0.7746327205599569,
+      "learning_rate": 0.00036051585165964217,
+      "loss": 1.2264,
+      "step": 10071
+    },
+    {
+      "epoch": 0.7747096377201754,
+      "learning_rate": 0.0003602801658016527,
+      "loss": 1.645,
+      "step": 10072
+    },
+    {
+      "epoch": 0.7747865548803938,
+      "learning_rate": 0.00036004454649309276,
+      "loss": 1.1012,
+      "step": 10073
+    },
+    {
+      "epoch": 0.7748634720406122,
+      "learning_rate": 0.00035980899374771957,
+      "loss": 1.0654,
+      "step": 10074
+    },
+    {
+      "epoch": 0.7749403892008307,
+      "learning_rate": 0.0003595735075792877,
+      "loss": 1.3011,
+      "step": 10075
+    },
+    {
+      "epoch": 0.7750173063610492,
+      "learning_rate": 0.00035933808800154725,
+      "loss": 0.8951,
+      "step": 10076
+    },
+    {
+      "epoch": 0.7750942235212676,
+      "learning_rate": 0.00035910273502824447,
+      "loss": 1.1847,
+      "step": 10077
+    },
+    {
+      "epoch": 0.7751711406814861,
+      "learning_rate": 0.0003588674486731217,
+      "loss": 1.1407,
+      "step": 10078
+    },
+    {
+      "epoch": 0.7752480578417045,
+      "learning_rate": 0.00035863222894991865,
+      "loss": 0.9388,
+      "step": 10079
+    },
+    {
+      "epoch": 0.7753249750019229,
+      "learning_rate": 0.0003583970758723688,
+      "loss": 1.1442,
+      "step": 10080
+    },
+    {
+      "epoch": 0.7754018921621414,
+      "learning_rate": 0.00035816198945420316,
+      "loss": 1.0684,
+      "step": 10081
+    },
+    {
+      "epoch": 0.7754788093223598,
+      "learning_rate": 0.0003579269697091492,
+      "loss": 1.2297,
+      "step": 10082
+    },
+    {
+      "epoch": 0.7755557264825783,
+      "learning_rate": 0.00035769201665092977,
+      "loss": 0.8241,
+      "step": 10083
+    },
+    {
+      "epoch": 0.7756326436427967,
+      "learning_rate": 0.0003574571302932641,
+      "loss": 1.2009,
+      "step": 10084
+    },
+    {
+      "epoch": 0.7757095608030151,
+      "learning_rate": 0.0003572223106498672,
+      "loss": 1.2966,
+      "step": 10085
+    },
+    {
+      "epoch": 0.7757864779632336,
+      "learning_rate": 0.0003569875577344501,
+      "loss": 1.3863,
+      "step": 10086
+    },
+    {
+      "epoch": 0.775863395123452,
+      "learning_rate": 0.0003567528715607212,
+      "loss": 1.1067,
+      "step": 10087
+    },
+    {
+      "epoch": 0.7759403122836704,
+      "learning_rate": 0.0003565182521423837,
+      "loss": 1.3285,
+      "step": 10088
+    },
+    {
+      "epoch": 0.776017229443889,
+      "learning_rate": 0.00035628369949313696,
+      "loss": 1.2468,
+      "step": 10089
+    },
+    {
+      "epoch": 0.7760941466041074,
+      "learning_rate": 0.00035604921362667716,
+      "loss": 0.8892,
+      "step": 10090
+    },
+    {
+      "epoch": 0.7761710637643259,
+      "learning_rate": 0.0003558147945566954,
+      "loss": 1.2645,
+      "step": 10091
+    },
+    {
+      "epoch": 0.7762479809245443,
+      "learning_rate": 0.0003555804422968811,
+      "loss": 1.2205,
+      "step": 10092
+    },
+    {
+      "epoch": 0.7763248980847627,
+      "learning_rate": 0.00035534615686091703,
+      "loss": 1.3176,
+      "step": 10093
+    },
+    {
+      "epoch": 0.7764018152449812,
+      "learning_rate": 0.00035511193826248346,
+      "loss": 0.812,
+      "step": 10094
+    },
+    {
+      "epoch": 0.7764787324051996,
+      "learning_rate": 0.00035487778651525724,
+      "loss": 1.1748,
+      "step": 10095
+    },
+    {
+      "epoch": 0.776555649565418,
+      "learning_rate": 0.0003546437016329107,
+      "loss": 1.0933,
+      "step": 10096
+    },
+    {
+      "epoch": 0.7766325667256365,
+      "learning_rate": 0.000354409683629112,
+      "loss": 1.075,
+      "step": 10097
+    },
+    {
+      "epoch": 0.7767094838858549,
+      "learning_rate": 0.0003541757325175258,
+      "loss": 1.2219,
+      "step": 10098
+    },
+    {
+      "epoch": 0.7767864010460733,
+      "learning_rate": 0.00035394184831181254,
+      "loss": 1.5471,
+      "step": 10099
+    },
+    {
+      "epoch": 0.7768633182062918,
+      "learning_rate": 0.0003537080310256289,
+      "loss": 1.1618,
+      "step": 10100
+    },
+    {
+      "epoch": 0.7769402353665102,
+      "learning_rate": 0.00035347428067262837,
+      "loss": 1.2653,
+      "step": 10101
+    },
+    {
+      "epoch": 0.7770171525267288,
+      "learning_rate": 0.0003532405972664595,
+      "loss": 1.2221,
+      "step": 10102
+    },
+    {
+      "epoch": 0.7770940696869472,
+      "learning_rate": 0.0003530069808207672,
+      "loss": 0.8411,
+      "step": 10103
+    },
+    {
+      "epoch": 0.7771709868471656,
+      "learning_rate": 0.0003527734313491927,
+      "loss": 1.1979,
+      "step": 10104
+    },
+    {
+      "epoch": 0.7772479040073841,
+      "learning_rate": 0.00035253994886537307,
+      "loss": 1.4187,
+      "step": 10105
+    },
+    {
+      "epoch": 0.7773248211676025,
+      "learning_rate": 0.00035230653338294186,
+      "loss": 1.0576,
+      "step": 10106
+    },
+    {
+      "epoch": 0.7774017383278209,
+      "learning_rate": 0.00035207318491552815,
+      "loss": 1.108,
+      "step": 10107
+    },
+    {
+      "epoch": 0.7774786554880394,
+      "learning_rate": 0.0003518399034767571,
+      "loss": 0.9282,
+      "step": 10108
+    },
+    {
+      "epoch": 0.7775555726482578,
+      "learning_rate": 0.0003516066890802511,
+      "loss": 1.2714,
+      "step": 10109
+    },
+    {
+      "epoch": 0.7776324898084763,
+      "learning_rate": 0.0003513735417396273,
+      "loss": 1.3496,
+      "step": 10110
+    },
+    {
+      "epoch": 0.7777094069686947,
+      "learning_rate": 0.0003511404614684995,
+      "loss": 1.0805,
+      "step": 10111
+    },
+    {
+      "epoch": 0.7777863241289131,
+      "learning_rate": 0.00035090744828047754,
+      "loss": 0.7998,
+      "step": 10112
+    },
+    {
+      "epoch": 0.7778632412891316,
+      "learning_rate": 0.0003506745021891668,
+      "loss": 1.0188,
+      "step": 10113
+    },
+    {
+      "epoch": 0.77794015844935,
+      "learning_rate": 0.00035044162320817,
+      "loss": 0.9875,
+      "step": 10114
+    },
+    {
+      "epoch": 0.7780170756095685,
+      "learning_rate": 0.00035020881135108535,
+      "loss": 1.2888,
+      "step": 10115
+    },
+    {
+      "epoch": 0.778093992769787,
+      "learning_rate": 0.00034997606663150554,
+      "loss": 1.2168,
+      "step": 10116
+    },
+    {
+      "epoch": 0.7781709099300054,
+      "learning_rate": 0.00034974338906302227,
+      "loss": 1.3169,
+      "step": 10117
+    },
+    {
+      "epoch": 0.7782478270902238,
+      "learning_rate": 0.00034951077865922134,
+      "loss": 1.2438,
+      "step": 10118
+    },
+    {
+      "epoch": 0.7783247442504423,
+      "learning_rate": 0.0003492782354336848,
+      "loss": 1.2502,
+      "step": 10119
+    },
+    {
+      "epoch": 0.7784016614106607,
+      "learning_rate": 0.00034904575939999154,
+      "loss": 1.1903,
+      "step": 10120
+    },
+    {
+      "epoch": 0.7784785785708792,
+      "learning_rate": 0.00034881335057171555,
+      "loss": 1.1976,
+      "step": 10121
+    },
+    {
+      "epoch": 0.7785554957310976,
+      "learning_rate": 0.00034858100896242806,
+      "loss": 1.09,
+      "step": 10122
+    },
+    {
+      "epoch": 0.778632412891316,
+      "learning_rate": 0.00034834873458569545,
+      "loss": 1.3113,
+      "step": 10123
+    },
+    {
+      "epoch": 0.7787093300515345,
+      "learning_rate": 0.0003481165274550803,
+      "loss": 0.787,
+      "step": 10124
+    },
+    {
+      "epoch": 0.7787862472117529,
+      "learning_rate": 0.0003478843875841417,
+      "loss": 1.4309,
+      "step": 10125
+    },
+    {
+      "epoch": 0.7788631643719713,
+      "learning_rate": 0.0003476523149864344,
+      "loss": 1.1621,
+      "step": 10126
+    },
+    {
+      "epoch": 0.7789400815321899,
+      "learning_rate": 0.00034742030967550906,
+      "loss": 1.36,
+      "step": 10127
+    },
+    {
+      "epoch": 0.7790169986924083,
+      "learning_rate": 0.00034718837166491375,
+      "loss": 1.4446,
+      "step": 10128
+    },
+    {
+      "epoch": 0.7790939158526268,
+      "learning_rate": 0.00034695650096819004,
+      "loss": 1.1737,
+      "step": 10129
+    },
+    {
+      "epoch": 0.7791708330128452,
+      "learning_rate": 0.0003467246975988785,
+      "loss": 1.2886,
+      "step": 10130
+    },
+    {
+      "epoch": 0.7792477501730636,
+      "learning_rate": 0.0003464929615705136,
+      "loss": 1.0131,
+      "step": 10131
+    },
+    {
+      "epoch": 0.7793246673332821,
+      "learning_rate": 0.0003462612928966268,
+      "loss": 1.0757,
+      "step": 10132
+    },
+    {
+      "epoch": 0.7794015844935005,
+      "learning_rate": 0.0003460296915907456,
+      "loss": 1.1423,
+      "step": 10133
+    },
+    {
+      "epoch": 0.7794785016537189,
+      "learning_rate": 0.0003457981576663932,
+      "loss": 1.2314,
+      "step": 10134
+    },
+    {
+      "epoch": 0.7795554188139374,
+      "learning_rate": 0.000345566691137089,
+      "loss": 1.246,
+      "step": 10135
+    },
+    {
+      "epoch": 0.7796323359741558,
+      "learning_rate": 0.0003453352920163491,
+      "loss": 1.1795,
+      "step": 10136
+    },
+    {
+      "epoch": 0.7797092531343742,
+      "learning_rate": 0.00034510396031768543,
+      "loss": 0.9148,
+      "step": 10137
+    },
+    {
+      "epoch": 0.7797861702945927,
+      "learning_rate": 0.0003448726960546042,
+      "loss": 1.1476,
+      "step": 10138
+    },
+    {
+      "epoch": 0.7798630874548111,
+      "learning_rate": 0.0003446414992406106,
+      "loss": 1.1403,
+      "step": 10139
+    },
+    {
+      "epoch": 0.7799400046150297,
+      "learning_rate": 0.0003444103698892034,
+      "loss": 1.257,
+      "step": 10140
+    },
+    {
+      "epoch": 0.7800169217752481,
+      "learning_rate": 0.0003441793080138799,
+      "loss": 1.0433,
+      "step": 10141
+    },
+    {
+      "epoch": 0.7800938389354665,
+      "learning_rate": 0.0003439483136281308,
+      "loss": 1.2958,
+      "step": 10142
+    },
+    {
+      "epoch": 0.780170756095685,
+      "learning_rate": 0.000343717386745444,
+      "loss": 1.4031,
+      "step": 10143
+    },
+    {
+      "epoch": 0.7802476732559034,
+      "learning_rate": 0.00034348652737930456,
+      "loss": 1.0079,
+      "step": 10144
+    },
+    {
+      "epoch": 0.7803245904161218,
+      "learning_rate": 0.00034325573554319205,
+      "loss": 1.8904,
+      "step": 10145
+    },
+    {
+      "epoch": 0.7804015075763403,
+      "learning_rate": 0.0003430250112505825,
+      "loss": 1.1857,
+      "step": 10146
+    },
+    {
+      "epoch": 0.7804784247365587,
+      "learning_rate": 0.00034279435451494843,
+      "loss": 1.5649,
+      "step": 10147
+    },
+    {
+      "epoch": 0.7805553418967772,
+      "learning_rate": 0.00034256376534975763,
+      "loss": 1.0356,
+      "step": 10148
+    },
+    {
+      "epoch": 0.7806322590569956,
+      "learning_rate": 0.0003423332437684753,
+      "loss": 1.4342,
+      "step": 10149
+    },
+    {
+      "epoch": 0.780709176217214,
+      "learning_rate": 0.00034210278978456164,
+      "loss": 1.1224,
+      "step": 10150
+    },
+    {
+      "epoch": 0.7807860933774325,
+      "learning_rate": 0.00034187240341147215,
+      "loss": 1.1346,
+      "step": 10151
+    },
+    {
+      "epoch": 0.780863010537651,
+      "learning_rate": 0.0003416420846626601,
+      "loss": 1.0649,
+      "step": 10152
+    },
+    {
+      "epoch": 0.7809399276978694,
+      "learning_rate": 0.0003414118335515742,
+      "loss": 1.3088,
+      "step": 10153
+    },
+    {
+      "epoch": 0.7810168448580879,
+      "learning_rate": 0.00034118165009165875,
+      "loss": 1.2056,
+      "step": 10154
+    },
+    {
+      "epoch": 0.7810937620183063,
+      "learning_rate": 0.0003409515342963544,
+      "loss": 1.0821,
+      "step": 10155
+    },
+    {
+      "epoch": 0.7811706791785247,
+      "learning_rate": 0.0003407214861790979,
+      "loss": 1.1166,
+      "step": 10156
+    },
+    {
+      "epoch": 0.7812475963387432,
+      "learning_rate": 0.0003404915057533216,
+      "loss": 1.2853,
+      "step": 10157
+    },
+    {
+      "epoch": 0.7813245134989616,
+      "learning_rate": 0.0003402615930324552,
+      "loss": 0.8062,
+      "step": 10158
+    },
+    {
+      "epoch": 0.7814014306591801,
+      "learning_rate": 0.00034003174802992286,
+      "loss": 0.9184,
+      "step": 10159
+    },
+    {
+      "epoch": 0.7814783478193985,
+      "learning_rate": 0.00033980197075914574,
+      "loss": 1.4058,
+      "step": 10160
+    },
+    {
+      "epoch": 0.7815552649796169,
+      "learning_rate": 0.0003395722612335406,
+      "loss": 1.0509,
+      "step": 10161
+    },
+    {
+      "epoch": 0.7816321821398354,
+      "learning_rate": 0.0003393426194665201,
+      "loss": 1.1489,
+      "step": 10162
+    },
+    {
+      "epoch": 0.7817090993000538,
+      "learning_rate": 0.0003391130454714945,
+      "loss": 0.9629,
+      "step": 10163
+    },
+    {
+      "epoch": 0.7817860164602722,
+      "learning_rate": 0.0003388835392618676,
+      "loss": 0.9332,
+      "step": 10164
+    },
+    {
+      "epoch": 0.7818629336204908,
+      "learning_rate": 0.0003386541008510406,
+      "loss": 0.8313,
+      "step": 10165
+    },
+    {
+      "epoch": 0.7819398507807092,
+      "learning_rate": 0.00033842473025241126,
+      "loss": 1.1998,
+      "step": 10166
+    },
+    {
+      "epoch": 0.7820167679409277,
+      "learning_rate": 0.0003381954274793726,
+      "loss": 1.1993,
+      "step": 10167
+    },
+    {
+      "epoch": 0.7820936851011461,
+      "learning_rate": 0.0003379661925453137,
+      "loss": 1.1401,
+      "step": 10168
+    },
+    {
+      "epoch": 0.7821706022613645,
+      "learning_rate": 0.00033773702546361983,
+      "loss": 1.111,
+      "step": 10169
+    },
+    {
+      "epoch": 0.782247519421583,
+      "learning_rate": 0.00033750792624767184,
+      "loss": 1.2488,
+      "step": 10170
+    },
+    {
+      "epoch": 0.7823244365818014,
+      "learning_rate": 0.00033727889491084816,
+      "loss": 1.0438,
+      "step": 10171
+    },
+    {
+      "epoch": 0.7824013537420198,
+      "learning_rate": 0.00033704993146652177,
+      "loss": 1.3586,
+      "step": 10172
+    },
+    {
+      "epoch": 0.7824782709022383,
+      "learning_rate": 0.0003368210359280612,
+      "loss": 1.3086,
+      "step": 10173
+    },
+    {
+      "epoch": 0.7825551880624567,
+      "learning_rate": 0.00033659220830883294,
+      "loss": 1.0261,
+      "step": 10174
+    },
+    {
+      "epoch": 0.7826321052226752,
+      "learning_rate": 0.0003363634486221978,
+      "loss": 1.0639,
+      "step": 10175
+    },
+    {
+      "epoch": 0.7827090223828936,
+      "learning_rate": 0.00033613475688151436,
+      "loss": 1.049,
+      "step": 10176
+    },
+    {
+      "epoch": 0.782785939543112,
+      "learning_rate": 0.0003359061331001349,
+      "loss": 0.6166,
+      "step": 10177
+    },
+    {
+      "epoch": 0.7828628567033306,
+      "learning_rate": 0.0003356775772914091,
+      "loss": 1.0824,
+      "step": 10178
+    },
+    {
+      "epoch": 0.782939773863549,
+      "learning_rate": 0.0003354490894686835,
+      "loss": 1.1148,
+      "step": 10179
+    },
+    {
+      "epoch": 0.7830166910237674,
+      "learning_rate": 0.00033522066964529923,
+      "loss": 1.1823,
+      "step": 10180
+    },
+    {
+      "epoch": 0.7830936081839859,
+      "learning_rate": 0.0003349923178345941,
+      "loss": 1.5071,
+      "step": 10181
+    },
+    {
+      "epoch": 0.7831705253442043,
+      "learning_rate": 0.00033476403404990144,
+      "loss": 1.1846,
+      "step": 10182
+    },
+    {
+      "epoch": 0.7832474425044227,
+      "learning_rate": 0.0003345358183045513,
+      "loss": 1.2911,
+      "step": 10183
+    },
+    {
+      "epoch": 0.7833243596646412,
+      "learning_rate": 0.000334307670611869,
+      "loss": 1.3045,
+      "step": 10184
+    },
+    {
+      "epoch": 0.7834012768248596,
+      "learning_rate": 0.0003340795909851774,
+      "loss": 1.4123,
+      "step": 10185
+    },
+    {
+      "epoch": 0.7834781939850781,
+      "learning_rate": 0.00033385157943779275,
+      "loss": 0.7913,
+      "step": 10186
+    },
+    {
+      "epoch": 0.7835551111452965,
+      "learning_rate": 0.0003336236359830302,
+      "loss": 1.3126,
+      "step": 10187
+    },
+    {
+      "epoch": 0.7836320283055149,
+      "learning_rate": 0.000333395760634199,
+      "loss": 1.0296,
+      "step": 10188
+    },
+    {
+      "epoch": 0.7837089454657334,
+      "learning_rate": 0.00033316795340460525,
+      "loss": 1.1184,
+      "step": 10189
+    },
+    {
+      "epoch": 0.7837858626259518,
+      "learning_rate": 0.0003329402143075506,
+      "loss": 0.9802,
+      "step": 10190
+    },
+    {
+      "epoch": 0.7838627797861702,
+      "learning_rate": 0.00033271254335633327,
+      "loss": 1.1399,
+      "step": 10191
+    },
+    {
+      "epoch": 0.7839396969463888,
+      "learning_rate": 0.00033248494056424663,
+      "loss": 1.2731,
+      "step": 10192
+    },
+    {
+      "epoch": 0.7840166141066072,
+      "learning_rate": 0.00033225740594458147,
+      "loss": 0.9823,
+      "step": 10193
+    },
+    {
+      "epoch": 0.7840935312668257,
+      "learning_rate": 0.0003320299395106233,
+      "loss": 1.3699,
+      "step": 10194
+    },
+    {
+      "epoch": 0.7841704484270441,
+      "learning_rate": 0.0003318025412756542,
+      "loss": 1.0667,
+      "step": 10195
+    },
+    {
+      "epoch": 0.7842473655872625,
+      "learning_rate": 0.0003315752112529523,
+      "loss": 1.3758,
+      "step": 10196
+    },
+    {
+      "epoch": 0.784324282747481,
+      "learning_rate": 0.0003313479494557909,
+      "loss": 1.1681,
+      "step": 10197
+    },
+    {
+      "epoch": 0.7844011999076994,
+      "learning_rate": 0.0003311207558974416,
+      "loss": 1.1172,
+      "step": 10198
+    },
+    {
+      "epoch": 0.7844781170679178,
+      "learning_rate": 0.00033089363059116914,
+      "loss": 1.0151,
+      "step": 10199
+    },
+    {
+      "epoch": 0.7845550342281363,
+      "learning_rate": 0.0003306665735502356,
+      "loss": 1.2513,
+      "step": 10200
+    },
+    {
+      "epoch": 0.7846319513883547,
+      "learning_rate": 0.0003304395847878998,
+      "loss": 1.4673,
+      "step": 10201
+    },
+    {
+      "epoch": 0.7847088685485731,
+      "learning_rate": 0.0003302126643174157,
+      "loss": 1.2164,
+      "step": 10202
+    },
+    {
+      "epoch": 0.7847857857087916,
+      "learning_rate": 0.00032998581215203316,
+      "loss": 1.2011,
+      "step": 10203
+    },
+    {
+      "epoch": 0.78486270286901,
+      "learning_rate": 0.0003297590283049984,
+      "loss": 1.0943,
+      "step": 10204
+    },
+    {
+      "epoch": 0.7849396200292286,
+      "learning_rate": 0.0003295323127895531,
+      "loss": 1.1943,
+      "step": 10205
+    },
+    {
+      "epoch": 0.785016537189447,
+      "learning_rate": 0.0003293056656189364,
+      "loss": 1.2752,
+      "step": 10206
+    },
+    {
+      "epoch": 0.7850934543496654,
+      "learning_rate": 0.00032907908680638237,
+      "loss": 1.6692,
+      "step": 10207
+    },
+    {
+      "epoch": 0.7851703715098839,
+      "learning_rate": 0.00032885257636511983,
+      "loss": 1.376,
+      "step": 10208
+    },
+    {
+      "epoch": 0.7852472886701023,
+      "learning_rate": 0.00032862613430837653,
+      "loss": 1.1115,
+      "step": 10209
+    },
+    {
+      "epoch": 0.7853242058303207,
+      "learning_rate": 0.00032839976064937383,
+      "loss": 1.011,
+      "step": 10210
+    },
+    {
+      "epoch": 0.7854011229905392,
+      "learning_rate": 0.0003281734554013302,
+      "loss": 1.2191,
+      "step": 10211
+    },
+    {
+      "epoch": 0.7854780401507576,
+      "learning_rate": 0.0003279472185774597,
+      "loss": 1.3628,
+      "step": 10212
+    },
+    {
+      "epoch": 0.7855549573109761,
+      "learning_rate": 0.0003277210501909722,
+      "loss": 0.9716,
+      "step": 10213
+    },
+    {
+      "epoch": 0.7856318744711945,
+      "learning_rate": 0.0003274949502550747,
+      "loss": 1.1441,
+      "step": 10214
+    },
+    {
+      "epoch": 0.7857087916314129,
+      "learning_rate": 0.000327268918782969,
+      "loss": 0.9489,
+      "step": 10215
+    },
+    {
+      "epoch": 0.7857857087916315,
+      "learning_rate": 0.00032704295578785337,
+      "loss": 0.8836,
+      "step": 10216
+    },
+    {
+      "epoch": 0.7858626259518499,
+      "learning_rate": 0.00032681706128292196,
+      "loss": 0.9046,
+      "step": 10217
+    },
+    {
+      "epoch": 0.7859395431120683,
+      "learning_rate": 0.00032659123528136497,
+      "loss": 1.2939,
+      "step": 10218
+    },
+    {
+      "epoch": 0.7860164602722868,
+      "learning_rate": 0.0003263654777963683,
+      "loss": 1.0273,
+      "step": 10219
+    },
+    {
+      "epoch": 0.7860933774325052,
+      "learning_rate": 0.0003261397888411153,
+      "loss": 1.094,
+      "step": 10220
+    },
+    {
+      "epoch": 0.7861702945927236,
+      "learning_rate": 0.000325914168428783,
+      "loss": 0.9821,
+      "step": 10221
+    },
+    {
+      "epoch": 0.7862472117529421,
+      "learning_rate": 0.0003256886165725456,
+      "loss": 1.2025,
+      "step": 10222
+    },
+    {
+      "epoch": 0.7863241289131605,
+      "learning_rate": 0.0003254631332855742,
+      "loss": 0.9703,
+      "step": 10223
+    },
+    {
+      "epoch": 0.786401046073379,
+      "learning_rate": 0.00032523771858103433,
+      "loss": 0.7507,
+      "step": 10224
+    },
+    {
+      "epoch": 0.7864779632335974,
+      "learning_rate": 0.00032501237247208856,
+      "loss": 1.0983,
+      "step": 10225
+    },
+    {
+      "epoch": 0.7865548803938158,
+      "learning_rate": 0.00032478709497189476,
+      "loss": 1.1567,
+      "step": 10226
+    },
+    {
+      "epoch": 0.7866317975540343,
+      "learning_rate": 0.0003245618860936069,
+      "loss": 1.1132,
+      "step": 10227
+    },
+    {
+      "epoch": 0.7867087147142527,
+      "learning_rate": 0.0003243367458503761,
+      "loss": 1.0461,
+      "step": 10228
+    },
+    {
+      "epoch": 0.7867856318744711,
+      "learning_rate": 0.0003241116742553482,
+      "loss": 1.5757,
+      "step": 10229
+    },
+    {
+      "epoch": 0.7868625490346897,
+      "learning_rate": 0.0003238866713216643,
+      "loss": 1.1834,
+      "step": 10230
+    },
+    {
+      "epoch": 0.7869394661949081,
+      "learning_rate": 0.0003236617370624638,
+      "loss": 1.2352,
+      "step": 10231
+    },
+    {
+      "epoch": 0.7870163833551266,
+      "learning_rate": 0.00032343687149088003,
+      "loss": 2.163,
+      "step": 10232
+    },
+    {
+      "epoch": 0.787093300515345,
+      "learning_rate": 0.0003232120746200444,
+      "loss": 1.071,
+      "step": 10233
+    },
+    {
+      "epoch": 0.7871702176755634,
+      "learning_rate": 0.00032298734646308164,
+      "loss": 0.7298,
+      "step": 10234
+    },
+    {
+      "epoch": 0.7872471348357819,
+      "learning_rate": 0.00032276268703311403,
+      "loss": 1.3017,
+      "step": 10235
+    },
+    {
+      "epoch": 0.7873240519960003,
+      "learning_rate": 0.00032253809634326034,
+      "loss": 1.0415,
+      "step": 10236
+    },
+    {
+      "epoch": 0.7874009691562187,
+      "learning_rate": 0.0003223135744066344,
+      "loss": 1.0901,
+      "step": 10237
+    },
+    {
+      "epoch": 0.7874778863164372,
+      "learning_rate": 0.0003220891212363461,
+      "loss": 1.1322,
+      "step": 10238
+    },
+    {
+      "epoch": 0.7875548034766556,
+      "learning_rate": 0.00032186473684550176,
+      "loss": 1.0514,
+      "step": 10239
+    },
+    {
+      "epoch": 0.787631720636874,
+      "learning_rate": 0.0003216404212472031,
+      "loss": 1.2795,
+      "step": 10240
+    },
+    {
+      "epoch": 0.7877086377970925,
+      "learning_rate": 0.00032141617445454796,
+      "loss": 1.1256,
+      "step": 10241
+    },
+    {
+      "epoch": 0.787785554957311,
+      "learning_rate": 0.0003211919964806312,
+      "loss": 1.6298,
+      "step": 10242
+    },
+    {
+      "epoch": 0.7878624721175295,
+      "learning_rate": 0.0003209678873385421,
+      "loss": 1.0925,
+      "step": 10243
+    },
+    {
+      "epoch": 0.7879393892777479,
+      "learning_rate": 0.0003207438470413669,
+      "loss": 1.3912,
+      "step": 10244
+    },
+    {
+      "epoch": 0.7880163064379663,
+      "learning_rate": 0.0003205198756021876,
+      "loss": 1.4718,
+      "step": 10245
+    },
+    {
+      "epoch": 0.7880932235981848,
+      "learning_rate": 0.0003202959730340818,
+      "loss": 1.2329,
+      "step": 10246
+    },
+    {
+      "epoch": 0.7881701407584032,
+      "learning_rate": 0.0003200721393501236,
+      "loss": 1.2347,
+      "step": 10247
+    },
+    {
+      "epoch": 0.7882470579186216,
+      "learning_rate": 0.00031984837456338293,
+      "loss": 0.858,
+      "step": 10248
+    },
+    {
+      "epoch": 0.7883239750788401,
+      "learning_rate": 0.0003196246786869253,
+      "loss": 1.2483,
+      "step": 10249
+    },
+    {
+      "epoch": 0.7884008922390585,
+      "learning_rate": 0.000319401051733813,
+      "loss": 1.0947,
+      "step": 10250
+    },
+    {
+      "epoch": 0.788477809399277,
+      "learning_rate": 0.0003191774937171038,
+      "loss": 0.9954,
+      "step": 10251
+    },
+    {
+      "epoch": 0.7885547265594954,
+      "learning_rate": 0.0003189540046498514,
+      "loss": 0.6918,
+      "step": 10252
+    },
+    {
+      "epoch": 0.7886316437197138,
+      "learning_rate": 0.00031873058454510555,
+      "loss": 0.8014,
+      "step": 10253
+    },
+    {
+      "epoch": 0.7887085608799324,
+      "learning_rate": 0.0003185072334159115,
+      "loss": 1.3529,
+      "step": 10254
+    },
+    {
+      "epoch": 0.7887854780401508,
+      "learning_rate": 0.0003182839512753119,
+      "loss": 1.1859,
+      "step": 10255
+    },
+    {
+      "epoch": 0.7888623952003692,
+      "learning_rate": 0.0003180607381363442,
+      "loss": 1.2967,
+      "step": 10256
+    },
+    {
+      "epoch": 0.7889393123605877,
+      "learning_rate": 0.00031783759401204107,
+      "loss": 1.4974,
+      "step": 10257
+    },
+    {
+      "epoch": 0.7890162295208061,
+      "learning_rate": 0.00031761451891543327,
+      "loss": 0.9802,
+      "step": 10258
+    },
+    {
+      "epoch": 0.7890931466810245,
+      "learning_rate": 0.00031739151285954594,
+      "loss": 0.807,
+      "step": 10259
+    },
+    {
+      "epoch": 0.789170063841243,
+      "learning_rate": 0.0003171685758574005,
+      "loss": 0.7691,
+      "step": 10260
+    },
+    {
+      "epoch": 0.7892469810014614,
+      "learning_rate": 0.00031694570792201485,
+      "loss": 1.1775,
+      "step": 10261
+    },
+    {
+      "epoch": 0.7893238981616799,
+      "learning_rate": 0.0003167229090664018,
+      "loss": 1.1752,
+      "step": 10262
+    },
+    {
+      "epoch": 0.7894008153218983,
+      "learning_rate": 0.0003165001793035716,
+      "loss": 1.0563,
+      "step": 10263
+    },
+    {
+      "epoch": 0.7894777324821167,
+      "learning_rate": 0.0003162775186465295,
+      "loss": 1.1318,
+      "step": 10264
+    },
+    {
+      "epoch": 0.7895546496423352,
+      "learning_rate": 0.00031605492710827656,
+      "loss": 1.0815,
+      "step": 10265
+    },
+    {
+      "epoch": 0.7896315668025536,
+      "learning_rate": 0.0003158324047018104,
+      "loss": 1.6239,
+      "step": 10266
+    },
+    {
+      "epoch": 0.789708483962772,
+      "learning_rate": 0.0003156099514401242,
+      "loss": 1.1011,
+      "step": 10267
+    },
+    {
+      "epoch": 0.7897854011229906,
+      "learning_rate": 0.0003153875673362069,
+      "loss": 1.0138,
+      "step": 10268
+    },
+    {
+      "epoch": 0.789862318283209,
+      "learning_rate": 0.0003151652524030448,
+      "loss": 1.3059,
+      "step": 10269
+    },
+    {
+      "epoch": 0.7899392354434275,
+      "learning_rate": 0.0003149430066536178,
+      "loss": 0.821,
+      "step": 10270
+    },
+    {
+      "epoch": 0.7900161526036459,
+      "learning_rate": 0.00031472083010090397,
+      "loss": 1.2222,
+      "step": 10271
+    },
+    {
+      "epoch": 0.7900930697638643,
+      "learning_rate": 0.00031449872275787625,
+      "loss": 1.3313,
+      "step": 10272
+    },
+    {
+      "epoch": 0.7901699869240828,
+      "learning_rate": 0.00031427668463750353,
+      "loss": 1.3168,
+      "step": 10273
+    },
+    {
+      "epoch": 0.7902469040843012,
+      "learning_rate": 0.0003140547157527509,
+      "loss": 1.4956,
+      "step": 10274
+    },
+    {
+      "epoch": 0.7903238212445196,
+      "learning_rate": 0.00031383281611657937,
+      "loss": 1.1706,
+      "step": 10275
+    },
+    {
+      "epoch": 0.7904007384047381,
+      "learning_rate": 0.00031361098574194556,
+      "loss": 1.2118,
+      "step": 10276
+    },
+    {
+      "epoch": 0.7904776555649565,
+      "learning_rate": 0.00031338922464180306,
+      "loss": 0.9576,
+      "step": 10277
+    },
+    {
+      "epoch": 0.7905545727251749,
+      "learning_rate": 0.0003131675328291009,
+      "loss": 1.2488,
+      "step": 10278
+    },
+    {
+      "epoch": 0.7906314898853934,
+      "learning_rate": 0.0003129459103167824,
+      "loss": 1.4225,
+      "step": 10279
+    },
+    {
+      "epoch": 0.7907084070456118,
+      "learning_rate": 0.0003127243571177899,
+      "loss": 1.0132,
+      "step": 10280
+    },
+    {
+      "epoch": 0.7907853242058304,
+      "learning_rate": 0.0003125028732450591,
+      "loss": 1.1335,
+      "step": 10281
+    },
+    {
+      "epoch": 0.7908622413660488,
+      "learning_rate": 0.00031228145871152387,
+      "loss": 0.9534,
+      "step": 10282
+    },
+    {
+      "epoch": 0.7909391585262672,
+      "learning_rate": 0.00031206011353011165,
+      "loss": 1.1116,
+      "step": 10283
+    },
+    {
+      "epoch": 0.7910160756864857,
+      "learning_rate": 0.00031183883771374716,
+      "loss": 1.123,
+      "step": 10284
+    },
+    {
+      "epoch": 0.7910929928467041,
+      "learning_rate": 0.0003116176312753515,
+      "loss": 1.1737,
+      "step": 10285
+    },
+    {
+      "epoch": 0.7911699100069225,
+      "learning_rate": 0.00031139649422784075,
+      "loss": 1.0924,
+      "step": 10286
+    },
+    {
+      "epoch": 0.791246827167141,
+      "learning_rate": 0.00031117542658412774,
+      "loss": 1.2451,
+      "step": 10287
+    },
+    {
+      "epoch": 0.7913237443273594,
+      "learning_rate": 0.00031095442835712023,
+      "loss": 0.9532,
+      "step": 10288
+    },
+    {
+      "epoch": 0.7914006614875779,
+      "learning_rate": 0.00031073349955972263,
+      "loss": 1.1028,
+      "step": 10289
+    },
+    {
+      "epoch": 0.7914775786477963,
+      "learning_rate": 0.0003105126402048358,
+      "loss": 1.0861,
+      "step": 10290
+    },
+    {
+      "epoch": 0.7915544958080147,
+      "learning_rate": 0.00031029185030535583,
+      "loss": 1.0757,
+      "step": 10291
+    },
+    {
+      "epoch": 0.7916314129682333,
+      "learning_rate": 0.00031007112987417384,
+      "loss": 0.8651,
+      "step": 10292
+    },
+    {
+      "epoch": 0.7917083301284517,
+      "learning_rate": 0.00030985047892417915,
+      "loss": 1.0436,
+      "step": 10293
+    },
+    {
+      "epoch": 0.7917852472886701,
+      "learning_rate": 0.0003096298974682553,
+      "loss": 1.1169,
+      "step": 10294
+    },
+    {
+      "epoch": 0.7918621644488886,
+      "learning_rate": 0.0003094093855192823,
+      "loss": 0.919,
+      "step": 10295
+    },
+    {
+      "epoch": 0.791939081609107,
+      "learning_rate": 0.0003091889430901362,
+      "loss": 1.201,
+      "step": 10296
+    },
+    {
+      "epoch": 0.7920159987693255,
+      "learning_rate": 0.00030896857019368823,
+      "loss": 1.2314,
+      "step": 10297
+    },
+    {
+      "epoch": 0.7920929159295439,
+      "learning_rate": 0.0003087482668428072,
+      "loss": 0.8611,
+      "step": 10298
+    },
+    {
+      "epoch": 0.7921698330897623,
+      "learning_rate": 0.00030852803305035637,
+      "loss": 1.1182,
+      "step": 10299
+    },
+    {
+      "epoch": 0.7922467502499808,
+      "learning_rate": 0.00030830786882919524,
+      "loss": 1.2974,
+      "step": 10300
+    },
+    {
+      "epoch": 0.7923236674101992,
+      "learning_rate": 0.0003080877741921798,
+      "loss": 0.9761,
+      "step": 10301
+    },
+    {
+      "epoch": 0.7924005845704176,
+      "learning_rate": 0.0003078677491521613,
+      "loss": 1.2451,
+      "step": 10302
+    },
+    {
+      "epoch": 0.7924775017306361,
+      "learning_rate": 0.000307647793721987,
+      "loss": 1.1248,
+      "step": 10303
+    },
+    {
+      "epoch": 0.7925544188908545,
+      "learning_rate": 0.0003074279079145013,
+      "loss": 1.1212,
+      "step": 10304
+    },
+    {
+      "epoch": 0.7926313360510729,
+      "learning_rate": 0.00030720809174254266,
+      "loss": 1.3907,
+      "step": 10305
+    },
+    {
+      "epoch": 0.7927082532112915,
+      "learning_rate": 0.00030698834521894627,
+      "loss": 1.2372,
+      "step": 10306
+    },
+    {
+      "epoch": 0.7927851703715099,
+      "learning_rate": 0.00030676866835654417,
+      "loss": 1.4555,
+      "step": 10307
+    },
+    {
+      "epoch": 0.7928620875317284,
+      "learning_rate": 0.00030654906116816296,
+      "loss": 0.7656,
+      "step": 10308
+    },
+    {
+      "epoch": 0.7929390046919468,
+      "learning_rate": 0.0003063295236666259,
+      "loss": 1.0952,
+      "step": 10309
+    },
+    {
+      "epoch": 0.7930159218521652,
+      "learning_rate": 0.0003061100558647518,
+      "loss": 1.3046,
+      "step": 10310
+    },
+    {
+      "epoch": 0.7930928390123837,
+      "learning_rate": 0.0003058906577753556,
+      "loss": 0.8801,
+      "step": 10311
+    },
+    {
+      "epoch": 0.7931697561726021,
+      "learning_rate": 0.0003056713294112487,
+      "loss": 1.1586,
+      "step": 10312
+    },
+    {
+      "epoch": 0.7932466733328205,
+      "learning_rate": 0.00030545207078523797,
+      "loss": 1.227,
+      "step": 10313
+    },
+    {
+      "epoch": 0.793323590493039,
+      "learning_rate": 0.00030523288191012496,
+      "loss": 1.1585,
+      "step": 10314
+    },
+    {
+      "epoch": 0.7934005076532574,
+      "learning_rate": 0.0003050137627987096,
+      "loss": 1.5249,
+      "step": 10315
+    },
+    {
+      "epoch": 0.7934774248134759,
+      "learning_rate": 0.00030479471346378544,
+      "loss": 1.1025,
+      "step": 10316
+    },
+    {
+      "epoch": 0.7935543419736943,
+      "learning_rate": 0.00030457573391814454,
+      "loss": 0.9091,
+      "step": 10317
+    },
+    {
+      "epoch": 0.7936312591339127,
+      "learning_rate": 0.0003043568241745719,
+      "loss": 1.1504,
+      "step": 10318
+    },
+    {
+      "epoch": 0.7937081762941313,
+      "learning_rate": 0.00030413798424585,
+      "loss": 1.0784,
+      "step": 10319
+    },
+    {
+      "epoch": 0.7937850934543497,
+      "learning_rate": 0.0003039192141447579,
+      "loss": 1.3788,
+      "step": 10320
+    },
+    {
+      "epoch": 0.7938620106145681,
+      "learning_rate": 0.0003037005138840693,
+      "loss": 1.3925,
+      "step": 10321
+    },
+    {
+      "epoch": 0.7939389277747866,
+      "learning_rate": 0.00030348188347655464,
+      "loss": 1.193,
+      "step": 10322
+    },
+    {
+      "epoch": 0.794015844935005,
+      "learning_rate": 0.00030326332293497966,
+      "loss": 0.8348,
+      "step": 10323
+    },
+    {
+      "epoch": 0.7940927620952234,
+      "learning_rate": 0.0003030448322721065,
+      "loss": 1.3422,
+      "step": 10324
+    },
+    {
+      "epoch": 0.7941696792554419,
+      "learning_rate": 0.00030282641150069264,
+      "loss": 1.512,
+      "step": 10325
+    },
+    {
+      "epoch": 0.7942465964156603,
+      "learning_rate": 0.000302608060633493,
+      "loss": 1.129,
+      "step": 10326
+    },
+    {
+      "epoch": 0.7943235135758788,
+      "learning_rate": 0.0003023897796832559,
+      "loss": 1.0456,
+      "step": 10327
+    },
+    {
+      "epoch": 0.7944004307360972,
+      "learning_rate": 0.000302171568662728,
+      "loss": 1.1939,
+      "step": 10328
+    },
+    {
+      "epoch": 0.7944773478963156,
+      "learning_rate": 0.0003019534275846505,
+      "loss": 1.1195,
+      "step": 10329
+    },
+    {
+      "epoch": 0.7945542650565341,
+      "learning_rate": 0.00030173535646176107,
+      "loss": 1.2152,
+      "step": 10330
+    },
+    {
+      "epoch": 0.7946311822167526,
+      "learning_rate": 0.00030151735530679295,
+      "loss": 1.0945,
+      "step": 10331
+    },
+    {
+      "epoch": 0.794708099376971,
+      "learning_rate": 0.00030129942413247544,
+      "loss": 1.1961,
+      "step": 10332
+    },
+    {
+      "epoch": 0.7947850165371895,
+      "learning_rate": 0.0003010815629515334,
+      "loss": 1.0837,
+      "step": 10333
+    },
+    {
+      "epoch": 0.7948619336974079,
+      "learning_rate": 0.0003008637717766889,
+      "loss": 1.3923,
+      "step": 10334
+    },
+    {
+      "epoch": 0.7949388508576264,
+      "learning_rate": 0.0003006460506206583,
+      "loss": 0.8961,
+      "step": 10335
+    },
+    {
+      "epoch": 0.7950157680178448,
+      "learning_rate": 0.0003004283994961548,
+      "loss": 1.2444,
+      "step": 10336
+    },
+    {
+      "epoch": 0.7950926851780632,
+      "learning_rate": 0.0003002108184158872,
+      "loss": 1.0956,
+      "step": 10337
+    },
+    {
+      "epoch": 0.7951696023382817,
+      "learning_rate": 0.00029999330739256005,
+      "loss": 1.4607,
+      "step": 10338
+    },
+    {
+      "epoch": 0.7952465194985001,
+      "learning_rate": 0.0002997758664388749,
+      "loss": 1.1263,
+      "step": 10339
+    },
+    {
+      "epoch": 0.7953234366587185,
+      "learning_rate": 0.00029955849556752745,
+      "loss": 1.27,
+      "step": 10340
+    },
+    {
+      "epoch": 0.795400353818937,
+      "learning_rate": 0.0002993411947912101,
+      "loss": 1.7407,
+      "step": 10341
+    },
+    {
+      "epoch": 0.7954772709791554,
+      "learning_rate": 0.0002991239641226122,
+      "loss": 1.254,
+      "step": 10342
+    },
+    {
+      "epoch": 0.7955541881393738,
+      "learning_rate": 0.0002989068035744174,
+      "loss": 1.2726,
+      "step": 10343
+    },
+    {
+      "epoch": 0.7956311052995924,
+      "learning_rate": 0.00029868971315930617,
+      "loss": 0.993,
+      "step": 10344
+    },
+    {
+      "epoch": 0.7957080224598108,
+      "learning_rate": 0.00029847269288995444,
+      "loss": 1.3377,
+      "step": 10345
+    },
+    {
+      "epoch": 0.7957849396200293,
+      "learning_rate": 0.0002982557427790341,
+      "loss": 1.064,
+      "step": 10346
+    },
+    {
+      "epoch": 0.7958618567802477,
+      "learning_rate": 0.00029803886283921374,
+      "loss": 0.9866,
+      "step": 10347
+    },
+    {
+      "epoch": 0.7959387739404661,
+      "learning_rate": 0.0002978220530831571,
+      "loss": 1.3031,
+      "step": 10348
+    },
+    {
+      "epoch": 0.7960156911006846,
+      "learning_rate": 0.00029760531352352306,
+      "loss": 1.13,
+      "step": 10349
+    },
+    {
+      "epoch": 0.796092608260903,
+      "learning_rate": 0.00029738864417296816,
+      "loss": 1.2601,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7961695254211214,
+      "learning_rate": 0.00029717204504414377,
+      "loss": 1.0851,
+      "step": 10351
+    },
+    {
+      "epoch": 0.7962464425813399,
+      "learning_rate": 0.0002969555161496971,
+      "loss": 1.4252,
+      "step": 10352
+    },
+    {
+      "epoch": 0.7963233597415583,
+      "learning_rate": 0.0002967390575022717,
+      "loss": 1.2639,
+      "step": 10353
+    },
+    {
+      "epoch": 0.7964002769017768,
+      "learning_rate": 0.0002965226691145064,
+      "loss": 1.2395,
+      "step": 10354
+    },
+    {
+      "epoch": 0.7964771940619952,
+      "learning_rate": 0.00029630635099903683,
+      "loss": 1.0853,
+      "step": 10355
+    },
+    {
+      "epoch": 0.7965541112222136,
+      "learning_rate": 0.0002960901031684942,
+      "loss": 1.3469,
+      "step": 10356
+    },
+    {
+      "epoch": 0.7966310283824322,
+      "learning_rate": 0.00029587392563550487,
+      "loss": 1.2555,
+      "step": 10357
+    },
+    {
+      "epoch": 0.7967079455426506,
+      "learning_rate": 0.00029565781841269217,
+      "loss": 1.2899,
+      "step": 10358
+    },
+    {
+      "epoch": 0.796784862702869,
+      "learning_rate": 0.0002954417815126744,
+      "loss": 1.1721,
+      "step": 10359
+    },
+    {
+      "epoch": 0.7968617798630875,
+      "learning_rate": 0.000295225814948066,
+      "loss": 1.0623,
+      "step": 10360
+    },
+    {
+      "epoch": 0.7969386970233059,
+      "learning_rate": 0.00029500991873147864,
+      "loss": 1.3483,
+      "step": 10361
+    },
+    {
+      "epoch": 0.7970156141835243,
+      "learning_rate": 0.00029479409287551763,
+      "loss": 1.4211,
+      "step": 10362
+    },
+    {
+      "epoch": 0.7970925313437428,
+      "learning_rate": 0.0002945783373927852,
+      "loss": 1.2193,
+      "step": 10363
+    },
+    {
+      "epoch": 0.7971694485039612,
+      "learning_rate": 0.00029436265229588023,
+      "loss": 1.2345,
+      "step": 10364
+    },
+    {
+      "epoch": 0.7972463656641797,
+      "learning_rate": 0.00029414703759739636,
+      "loss": 1.1392,
+      "step": 10365
+    },
+    {
+      "epoch": 0.7973232828243981,
+      "learning_rate": 0.0002939314933099239,
+      "loss": 1.3871,
+      "step": 10366
+    },
+    {
+      "epoch": 0.7974001999846165,
+      "learning_rate": 0.0002937160194460483,
+      "loss": 1.1328,
+      "step": 10367
+    },
+    {
+      "epoch": 0.797477117144835,
+      "learning_rate": 0.00029350061601835116,
+      "loss": 1.1989,
+      "step": 10368
+    },
+    {
+      "epoch": 0.7975540343050534,
+      "learning_rate": 0.0002932852830394108,
+      "loss": 0.9767,
+      "step": 10369
+    },
+    {
+      "epoch": 0.7976309514652719,
+      "learning_rate": 0.00029307002052180093,
+      "loss": 0.736,
+      "step": 10370
+    },
+    {
+      "epoch": 0.7977078686254904,
+      "learning_rate": 0.0002928548284780894,
+      "loss": 1.5841,
+      "step": 10371
+    },
+    {
+      "epoch": 0.7977847857857088,
+      "learning_rate": 0.0002926397069208429,
+      "loss": 1.3683,
+      "step": 10372
+    },
+    {
+      "epoch": 0.7978617029459273,
+      "learning_rate": 0.00029242465586262187,
+      "loss": 1.1142,
+      "step": 10373
+    },
+    {
+      "epoch": 0.7979386201061457,
+      "learning_rate": 0.00029220967531598454,
+      "loss": 1.3105,
+      "step": 10374
+    },
+    {
+      "epoch": 0.7980155372663641,
+      "learning_rate": 0.00029199476529348257,
+      "loss": 1.1201,
+      "step": 10375
+    },
+    {
+      "epoch": 0.7980924544265826,
+      "learning_rate": 0.0002917799258076648,
+      "loss": 1.2185,
+      "step": 10376
+    },
+    {
+      "epoch": 0.798169371586801,
+      "learning_rate": 0.0002915651568710766,
+      "loss": 1.6242,
+      "step": 10377
+    },
+    {
+      "epoch": 0.7982462887470194,
+      "learning_rate": 0.00029135045849625833,
+      "loss": 1.3077,
+      "step": 10378
+    },
+    {
+      "epoch": 0.7983232059072379,
+      "learning_rate": 0.0002911358306957465,
+      "loss": 0.8829,
+      "step": 10379
+    },
+    {
+      "epoch": 0.7984001230674563,
+      "learning_rate": 0.0002909212734820734,
+      "loss": 0.7368,
+      "step": 10380
+    },
+    {
+      "epoch": 0.7984770402276747,
+      "learning_rate": 0.0002907067868677671,
+      "loss": 0.8756,
+      "step": 10381
+    },
+    {
+      "epoch": 0.7985539573878933,
+      "learning_rate": 0.00029049237086535145,
+      "loss": 0.8449,
+      "step": 10382
+    },
+    {
+      "epoch": 0.7986308745481117,
+      "learning_rate": 0.0002902780254873477,
+      "loss": 1.052,
+      "step": 10383
+    },
+    {
+      "epoch": 0.7987077917083302,
+      "learning_rate": 0.00029006375074626986,
+      "loss": 1.1996,
+      "step": 10384
+    },
+    {
+      "epoch": 0.7987847088685486,
+      "learning_rate": 0.00028984954665463097,
+      "loss": 1.2681,
+      "step": 10385
+    },
+    {
+      "epoch": 0.798861626028767,
+      "learning_rate": 0.00028963541322493826,
+      "loss": 1.1129,
+      "step": 10386
+    },
+    {
+      "epoch": 0.7989385431889855,
+      "learning_rate": 0.00028942135046969514,
+      "loss": 1.2128,
+      "step": 10387
+    },
+    {
+      "epoch": 0.7990154603492039,
+      "learning_rate": 0.00028920735840140115,
+      "loss": 0.8635,
+      "step": 10388
+    },
+    {
+      "epoch": 0.7990923775094223,
+      "learning_rate": 0.00028899343703255113,
+      "loss": 1.1278,
+      "step": 10389
+    },
+    {
+      "epoch": 0.7991692946696408,
+      "learning_rate": 0.0002887795863756363,
+      "loss": 1.3238,
+      "step": 10390
+    },
+    {
+      "epoch": 0.7992462118298592,
+      "learning_rate": 0.0002885658064431439,
+      "loss": 1.143,
+      "step": 10391
+    },
+    {
+      "epoch": 0.7993231289900777,
+      "learning_rate": 0.00028835209724755665,
+      "loss": 0.8206,
+      "step": 10392
+    },
+    {
+      "epoch": 0.7994000461502961,
+      "learning_rate": 0.00028813845880135315,
+      "loss": 0.9658,
+      "step": 10393
+    },
+    {
+      "epoch": 0.7994769633105145,
+      "learning_rate": 0.00028792489111700815,
+      "loss": 1.4044,
+      "step": 10394
+    },
+    {
+      "epoch": 0.7995538804707331,
+      "learning_rate": 0.0002877113942069914,
+      "loss": 0.9718,
+      "step": 10395
+    },
+    {
+      "epoch": 0.7996307976309515,
+      "learning_rate": 0.00028749796808377054,
+      "loss": 0.5927,
+      "step": 10396
+    },
+    {
+      "epoch": 0.7997077147911699,
+      "learning_rate": 0.00028728461275980666,
+      "loss": 1.3003,
+      "step": 10397
+    },
+    {
+      "epoch": 0.7997846319513884,
+      "learning_rate": 0.0002870713282475578,
+      "loss": 0.9377,
+      "step": 10398
+    },
+    {
+      "epoch": 0.7998615491116068,
+      "learning_rate": 0.0002868581145594784,
+      "loss": 0.8702,
+      "step": 10399
+    },
+    {
+      "epoch": 0.7999384662718253,
+      "learning_rate": 0.00028664497170801794,
+      "loss": 1.1248,
+      "step": 10400
+    },
+    {
+      "epoch": 0.8000153834320437,
+      "learning_rate": 0.00028643189970562226,
+      "loss": 1.0356,
+      "step": 10401
+    },
+    {
+      "epoch": 0.8000923005922621,
+      "learning_rate": 0.00028621889856473274,
+      "loss": 1.3393,
+      "step": 10402
+    },
+    {
+      "epoch": 0.8001692177524806,
+      "learning_rate": 0.0002860059682977861,
+      "loss": 0.9831,
+      "step": 10403
+    },
+    {
+      "epoch": 0.800246134912699,
+      "learning_rate": 0.0002857931089172168,
+      "loss": 0.8914,
+      "step": 10404
+    },
+    {
+      "epoch": 0.8003230520729174,
+      "learning_rate": 0.00028558032043545325,
+      "loss": 1.3209,
+      "step": 10405
+    },
+    {
+      "epoch": 0.8003999692331359,
+      "learning_rate": 0.0002853676028649202,
+      "loss": 1.1725,
+      "step": 10406
+    },
+    {
+      "epoch": 0.8004768863933543,
+      "learning_rate": 0.00028515495621803886,
+      "loss": 1.2159,
+      "step": 10407
+    },
+    {
+      "epoch": 0.8005538035535728,
+      "learning_rate": 0.0002849423805072257,
+      "loss": 1.2795,
+      "step": 10408
+    },
+    {
+      "epoch": 0.8006307207137913,
+      "learning_rate": 0.00028472987574489295,
+      "loss": 1.0802,
+      "step": 10409
+    },
+    {
+      "epoch": 0.8007076378740097,
+      "learning_rate": 0.00028451744194344985,
+      "loss": 1.1759,
+      "step": 10410
+    },
+    {
+      "epoch": 0.8007845550342282,
+      "learning_rate": 0.0002843050791152993,
+      "loss": 0.8882,
+      "step": 10411
+    },
+    {
+      "epoch": 0.8008614721944466,
+      "learning_rate": 0.0002840927872728427,
+      "loss": 1.1256,
+      "step": 10412
+    },
+    {
+      "epoch": 0.800938389354665,
+      "learning_rate": 0.0002838805664284754,
+      "loss": 1.0765,
+      "step": 10413
+    },
+    {
+      "epoch": 0.8010153065148835,
+      "learning_rate": 0.00028366841659458906,
+      "loss": 1.4075,
+      "step": 10414
+    },
+    {
+      "epoch": 0.8010922236751019,
+      "learning_rate": 0.0002834563377835717,
+      "loss": 1.1554,
+      "step": 10415
+    },
+    {
+      "epoch": 0.8011691408353203,
+      "learning_rate": 0.00028324433000780655,
+      "loss": 1.1122,
+      "step": 10416
+    },
+    {
+      "epoch": 0.8012460579955388,
+      "learning_rate": 0.0002830323932796728,
+      "loss": 0.8693,
+      "step": 10417
+    },
+    {
+      "epoch": 0.8013229751557572,
+      "learning_rate": 0.00028282052761154607,
+      "loss": 1.1575,
+      "step": 10418
+    },
+    {
+      "epoch": 0.8013998923159757,
+      "learning_rate": 0.00028260873301579737,
+      "loss": 1.001,
+      "step": 10419
+    },
+    {
+      "epoch": 0.8014768094761942,
+      "learning_rate": 0.0002823970095047934,
+      "loss": 1.0212,
+      "step": 10420
+    },
+    {
+      "epoch": 0.8015537266364126,
+      "learning_rate": 0.000282185357090897,
+      "loss": 1.3056,
+      "step": 10421
+    },
+    {
+      "epoch": 0.8016306437966311,
+      "learning_rate": 0.00028197377578646657,
+      "loss": 1.321,
+      "step": 10422
+    },
+    {
+      "epoch": 0.8017075609568495,
+      "learning_rate": 0.0002817622656038574,
+      "loss": 1.0864,
+      "step": 10423
+    },
+    {
+      "epoch": 0.8017844781170679,
+      "learning_rate": 0.00028155082655541877,
+      "loss": 0.9638,
+      "step": 10424
+    },
+    {
+      "epoch": 0.8018613952772864,
+      "learning_rate": 0.0002813394586534968,
+      "loss": 1.136,
+      "step": 10425
+    },
+    {
+      "epoch": 0.8019383124375048,
+      "learning_rate": 0.00028112816191043434,
+      "loss": 0.99,
+      "step": 10426
+    },
+    {
+      "epoch": 0.8020152295977232,
+      "learning_rate": 0.0002809169363385686,
+      "loss": 1.1861,
+      "step": 10427
+    },
+    {
+      "epoch": 0.8020921467579417,
+      "learning_rate": 0.00028070578195023364,
+      "loss": 0.9569,
+      "step": 10428
+    },
+    {
+      "epoch": 0.8021690639181601,
+      "learning_rate": 0.0002804946987577589,
+      "loss": 1.2844,
+      "step": 10429
+    },
+    {
+      "epoch": 0.8022459810783786,
+      "learning_rate": 0.00028028368677346897,
+      "loss": 1.193,
+      "step": 10430
+    },
+    {
+      "epoch": 0.802322898238597,
+      "learning_rate": 0.0002800727460096864,
+      "loss": 1.2253,
+      "step": 10431
+    },
+    {
+      "epoch": 0.8023998153988154,
+      "learning_rate": 0.0002798618764787278,
+      "loss": 1.1087,
+      "step": 10432
+    },
+    {
+      "epoch": 0.802476732559034,
+      "learning_rate": 0.00027965107819290496,
+      "loss": 1.1771,
+      "step": 10433
+    },
+    {
+      "epoch": 0.8025536497192524,
+      "learning_rate": 0.00027944035116452795,
+      "loss": 1.2884,
+      "step": 10434
+    },
+    {
+      "epoch": 0.8026305668794708,
+      "learning_rate": 0.000279229695405901,
+      "loss": 1.1639,
+      "step": 10435
+    },
+    {
+      "epoch": 0.8027074840396893,
+      "learning_rate": 0.0002790191109293243,
+      "loss": 1.49,
+      "step": 10436
+    },
+    {
+      "epoch": 0.8027844011999077,
+      "learning_rate": 0.00027880859774709403,
+      "loss": 1.5186,
+      "step": 10437
+    },
+    {
+      "epoch": 0.8028613183601262,
+      "learning_rate": 0.00027859815587150225,
+      "loss": 1.1989,
+      "step": 10438
+    },
+    {
+      "epoch": 0.8029382355203446,
+      "learning_rate": 0.00027838778531483747,
+      "loss": 0.9743,
+      "step": 10439
+    },
+    {
+      "epoch": 0.803015152680563,
+      "learning_rate": 0.0002781774860893829,
+      "loss": 0.9675,
+      "step": 10440
+    },
+    {
+      "epoch": 0.8030920698407815,
+      "learning_rate": 0.00027796725820741834,
+      "loss": 1.2377,
+      "step": 10441
+    },
+    {
+      "epoch": 0.8031689870009999,
+      "learning_rate": 0.00027775710168121917,
+      "loss": 1.145,
+      "step": 10442
+    },
+    {
+      "epoch": 0.8032459041612183,
+      "learning_rate": 0.0002775470165230566,
+      "loss": 1.266,
+      "step": 10443
+    },
+    {
+      "epoch": 0.8033228213214368,
+      "learning_rate": 0.00027733700274519736,
+      "loss": 1.2017,
+      "step": 10444
+    },
+    {
+      "epoch": 0.8033997384816552,
+      "learning_rate": 0.0002771270603599055,
+      "loss": 1.1181,
+      "step": 10445
+    },
+    {
+      "epoch": 0.8034766556418736,
+      "learning_rate": 0.0002769171893794387,
+      "loss": 0.6137,
+      "step": 10446
+    },
+    {
+      "epoch": 0.8035535728020922,
+      "learning_rate": 0.0002767073898160514,
+      "loss": 1.1542,
+      "step": 10447
+    },
+    {
+      "epoch": 0.8036304899623106,
+      "learning_rate": 0.00027649766168199495,
+      "loss": 1.2338,
+      "step": 10448
+    },
+    {
+      "epoch": 0.8037074071225291,
+      "learning_rate": 0.00027628800498951517,
+      "loss": 0.9504,
+      "step": 10449
+    },
+    {
+      "epoch": 0.8037843242827475,
+      "learning_rate": 0.0002760784197508539,
+      "loss": 1.4597,
+      "step": 10450
+    },
+    {
+      "epoch": 0.8038612414429659,
+      "learning_rate": 0.00027586890597824953,
+      "loss": 1.0946,
+      "step": 10451
+    },
+    {
+      "epoch": 0.8039381586031844,
+      "learning_rate": 0.000275659463683935,
+      "loss": 1.1617,
+      "step": 10452
+    },
+    {
+      "epoch": 0.8040150757634028,
+      "learning_rate": 0.0002754500928801407,
+      "loss": 1.0196,
+      "step": 10453
+    },
+    {
+      "epoch": 0.8040919929236212,
+      "learning_rate": 0.00027524079357909203,
+      "loss": 1.1086,
+      "step": 10454
+    },
+    {
+      "epoch": 0.8041689100838397,
+      "learning_rate": 0.00027503156579300914,
+      "loss": 1.1331,
+      "step": 10455
+    },
+    {
+      "epoch": 0.8042458272440581,
+      "learning_rate": 0.0002748224095341101,
+      "loss": 1.089,
+      "step": 10456
+    },
+    {
+      "epoch": 0.8043227444042766,
+      "learning_rate": 0.0002746133248146071,
+      "loss": 0.9148,
+      "step": 10457
+    },
+    {
+      "epoch": 0.804399661564495,
+      "learning_rate": 0.00027440431164670975,
+      "loss": 1.2087,
+      "step": 10458
+    },
+    {
+      "epoch": 0.8044765787247135,
+      "learning_rate": 0.0002741953700426218,
+      "loss": 1.1022,
+      "step": 10459
+    },
+    {
+      "epoch": 0.804553495884932,
+      "learning_rate": 0.0002739865000145431,
+      "loss": 1.1847,
+      "step": 10460
+    },
+    {
+      "epoch": 0.8046304130451504,
+      "learning_rate": 0.0002737777015746708,
+      "loss": 1.2637,
+      "step": 10461
+    },
+    {
+      "epoch": 0.8047073302053688,
+      "learning_rate": 0.00027356897473519644,
+      "loss": 1.1909,
+      "step": 10462
+    },
+    {
+      "epoch": 0.8047842473655873,
+      "learning_rate": 0.00027336031950830794,
+      "loss": 1.1146,
+      "step": 10463
+    },
+    {
+      "epoch": 0.8048611645258057,
+      "learning_rate": 0.00027315173590618854,
+      "loss": 1.2136,
+      "step": 10464
+    },
+    {
+      "epoch": 0.8049380816860241,
+      "learning_rate": 0.000272943223941018,
+      "loss": 1.2451,
+      "step": 10465
+    },
+    {
+      "epoch": 0.8050149988462426,
+      "learning_rate": 0.0002727347836249711,
+      "loss": 1.2267,
+      "step": 10466
+    },
+    {
+      "epoch": 0.805091916006461,
+      "learning_rate": 0.0002725264149702198,
+      "loss": 1.2632,
+      "step": 10467
+    },
+    {
+      "epoch": 0.8051688331666795,
+      "learning_rate": 0.0002723181179889298,
+      "loss": 1.211,
+      "step": 10468
+    },
+    {
+      "epoch": 0.8052457503268979,
+      "learning_rate": 0.00027210989269326456,
+      "loss": 1.2043,
+      "step": 10469
+    },
+    {
+      "epoch": 0.8053226674871163,
+      "learning_rate": 0.0002719017390953824,
+      "loss": 1.116,
+      "step": 10470
+    },
+    {
+      "epoch": 0.8053995846473349,
+      "learning_rate": 0.0002716936572074377,
+      "loss": 1.2985,
+      "step": 10471
+    },
+    {
+      "epoch": 0.8054765018075533,
+      "learning_rate": 0.00027148564704158044,
+      "loss": 0.9905,
+      "step": 10472
+    },
+    {
+      "epoch": 0.8055534189677717,
+      "learning_rate": 0.0002712777086099564,
+      "loss": 1.1847,
+      "step": 10473
+    },
+    {
+      "epoch": 0.8056303361279902,
+      "learning_rate": 0.0002710698419247074,
+      "loss": 1.1079,
+      "step": 10474
+    },
+    {
+      "epoch": 0.8057072532882086,
+      "learning_rate": 0.0002708620469979712,
+      "loss": 1.5432,
+      "step": 10475
+    },
+    {
+      "epoch": 0.8057841704484271,
+      "learning_rate": 0.0002706543238418812,
+      "loss": 1.3157,
+      "step": 10476
+    },
+    {
+      "epoch": 0.8058610876086455,
+      "learning_rate": 0.0002704466724685665,
+      "loss": 0.9149,
+      "step": 10477
+    },
+    {
+      "epoch": 0.8059380047688639,
+      "learning_rate": 0.00027023909289015176,
+      "loss": 1.751,
+      "step": 10478
+    },
+    {
+      "epoch": 0.8060149219290824,
+      "learning_rate": 0.00027003158511875787,
+      "loss": 1.4602,
+      "step": 10479
+    },
+    {
+      "epoch": 0.8060918390893008,
+      "learning_rate": 0.0002698241491665021,
+      "loss": 1.0065,
+      "step": 10480
+    },
+    {
+      "epoch": 0.8061687562495192,
+      "learning_rate": 0.00026961678504549614,
+      "loss": 0.8948,
+      "step": 10481
+    },
+    {
+      "epoch": 0.8062456734097377,
+      "learning_rate": 0.0002694094927678479,
+      "loss": 1.0458,
+      "step": 10482
+    },
+    {
+      "epoch": 0.8063225905699561,
+      "learning_rate": 0.0002692022723456623,
+      "loss": 1.2696,
+      "step": 10483
+    },
+    {
+      "epoch": 0.8063995077301745,
+      "learning_rate": 0.00026899512379103873,
+      "loss": 1.0358,
+      "step": 10484
+    },
+    {
+      "epoch": 0.8064764248903931,
+      "learning_rate": 0.00026878804711607276,
+      "loss": 1.123,
+      "step": 10485
+    },
+    {
+      "epoch": 0.8065533420506115,
+      "learning_rate": 0.0002685810423328558,
+      "loss": 1.144,
+      "step": 10486
+    },
+    {
+      "epoch": 0.80663025921083,
+      "learning_rate": 0.0002683741094534749,
+      "loss": 1.4369,
+      "step": 10487
+    },
+    {
+      "epoch": 0.8067071763710484,
+      "learning_rate": 0.00026816724849001354,
+      "loss": 1.2492,
+      "step": 10488
+    },
+    {
+      "epoch": 0.8067840935312668,
+      "learning_rate": 0.00026796045945455077,
+      "loss": 1.2483,
+      "step": 10489
+    },
+    {
+      "epoch": 0.8068610106914853,
+      "learning_rate": 0.00026775374235916006,
+      "loss": 0.9917,
+      "step": 10490
+    },
+    {
+      "epoch": 0.8069379278517037,
+      "learning_rate": 0.0002675470972159129,
+      "loss": 1.4104,
+      "step": 10491
+    },
+    {
+      "epoch": 0.8070148450119221,
+      "learning_rate": 0.0002673405240368752,
+      "loss": 1.2326,
+      "step": 10492
+    },
+    {
+      "epoch": 0.8070917621721406,
+      "learning_rate": 0.00026713402283410877,
+      "loss": 0.9989,
+      "step": 10493
+    },
+    {
+      "epoch": 0.807168679332359,
+      "learning_rate": 0.0002669275936196718,
+      "loss": 1.417,
+      "step": 10494
+    },
+    {
+      "epoch": 0.8072455964925775,
+      "learning_rate": 0.0002667212364056173,
+      "loss": 1.1293,
+      "step": 10495
+    },
+    {
+      "epoch": 0.807322513652796,
+      "learning_rate": 0.0002665149512039955,
+      "loss": 0.891,
+      "step": 10496
+    },
+    {
+      "epoch": 0.8073994308130144,
+      "learning_rate": 0.0002663087380268511,
+      "loss": 1.0769,
+      "step": 10497
+    },
+    {
+      "epoch": 0.8074763479732329,
+      "learning_rate": 0.00026610259688622523,
+      "loss": 0.8323,
+      "step": 10498
+    },
+    {
+      "epoch": 0.8075532651334513,
+      "learning_rate": 0.00026589652779415485,
+      "loss": 1.1398,
+      "step": 10499
+    },
+    {
+      "epoch": 0.8076301822936697,
+      "learning_rate": 0.0002656905307626721,
+      "loss": 0.9009,
+      "step": 10500
+    },
+    {
+      "epoch": 0.8077070994538882,
+      "learning_rate": 0.00026548460580380547,
+      "loss": 1.4085,
+      "step": 10501
+    },
+    {
+      "epoch": 0.8077840166141066,
+      "learning_rate": 0.0002652787529295799,
+      "loss": 1.3816,
+      "step": 10502
+    },
+    {
+      "epoch": 0.8078609337743251,
+      "learning_rate": 0.0002650729721520141,
+      "loss": 1.3336,
+      "step": 10503
+    },
+    {
+      "epoch": 0.8079378509345435,
+      "learning_rate": 0.00026486726348312465,
+      "loss": 0.9549,
+      "step": 10504
+    },
+    {
+      "epoch": 0.8080147680947619,
+      "learning_rate": 0.000264661626934923,
+      "loss": 1.2421,
+      "step": 10505
+    },
+    {
+      "epoch": 0.8080916852549804,
+      "learning_rate": 0.0002644560625194165,
+      "loss": 0.9791,
+      "step": 10506
+    },
+    {
+      "epoch": 0.8081686024151988,
+      "learning_rate": 0.00026425057024860804,
+      "loss": 1.113,
+      "step": 10507
+    },
+    {
+      "epoch": 0.8082455195754172,
+      "learning_rate": 0.0002640451501344967,
+      "loss": 1.2074,
+      "step": 10508
+    },
+    {
+      "epoch": 0.8083224367356358,
+      "learning_rate": 0.00026383980218907686,
+      "loss": 0.786,
+      "step": 10509
+    },
+    {
+      "epoch": 0.8083993538958542,
+      "learning_rate": 0.0002636345264243395,
+      "loss": 1.3403,
+      "step": 10510
+    },
+    {
+      "epoch": 0.8084762710560726,
+      "learning_rate": 0.0002634293228522711,
+      "loss": 1.1428,
+      "step": 10511
+    },
+    {
+      "epoch": 0.8085531882162911,
+      "learning_rate": 0.0002632241914848525,
+      "loss": 1.1962,
+      "step": 10512
+    },
+    {
+      "epoch": 0.8086301053765095,
+      "learning_rate": 0.00026301913233406257,
+      "loss": 1.6917,
+      "step": 10513
+    },
+    {
+      "epoch": 0.808707022536728,
+      "learning_rate": 0.0002628141454118744,
+      "loss": 1.0791,
+      "step": 10514
+    },
+    {
+      "epoch": 0.8087839396969464,
+      "learning_rate": 0.0002626092307302582,
+      "loss": 0.9986,
+      "step": 10515
+    },
+    {
+      "epoch": 0.8088608568571648,
+      "learning_rate": 0.00026240438830117853,
+      "loss": 1.1085,
+      "step": 10516
+    },
+    {
+      "epoch": 0.8089377740173833,
+      "learning_rate": 0.00026219961813659585,
+      "loss": 1.2558,
+      "step": 10517
+    },
+    {
+      "epoch": 0.8090146911776017,
+      "learning_rate": 0.0002619949202484679,
+      "loss": 1.0907,
+      "step": 10518
+    },
+    {
+      "epoch": 0.8090916083378201,
+      "learning_rate": 0.00026179029464874667,
+      "loss": 0.9909,
+      "step": 10519
+    },
+    {
+      "epoch": 0.8091685254980386,
+      "learning_rate": 0.00026158574134938064,
+      "loss": 1.5166,
+      "step": 10520
+    },
+    {
+      "epoch": 0.809245442658257,
+      "learning_rate": 0.0002613812603623137,
+      "loss": 1.1873,
+      "step": 10521
+    },
+    {
+      "epoch": 0.8093223598184756,
+      "learning_rate": 0.00026117685169948563,
+      "loss": 1.0856,
+      "step": 10522
+    },
+    {
+      "epoch": 0.809399276978694,
+      "learning_rate": 0.00026097251537283257,
+      "loss": 1.3302,
+      "step": 10523
+    },
+    {
+      "epoch": 0.8094761941389124,
+      "learning_rate": 0.000260768251394286,
+      "loss": 1.1469,
+      "step": 10524
+    },
+    {
+      "epoch": 0.8095531112991309,
+      "learning_rate": 0.0002605640597757721,
+      "loss": 0.9616,
+      "step": 10525
+    },
+    {
+      "epoch": 0.8096300284593493,
+      "learning_rate": 0.00026035994052921456,
+      "loss": 0.9726,
+      "step": 10526
+    },
+    {
+      "epoch": 0.8097069456195677,
+      "learning_rate": 0.0002601558936665323,
+      "loss": 1.0679,
+      "step": 10527
+    },
+    {
+      "epoch": 0.8097838627797862,
+      "learning_rate": 0.0002599519191996394,
+      "loss": 0.9663,
+      "step": 10528
+    },
+    {
+      "epoch": 0.8098607799400046,
+      "learning_rate": 0.0002597480171404463,
+      "loss": 1.2546,
+      "step": 10529
+    },
+    {
+      "epoch": 0.809937697100223,
+      "learning_rate": 0.0002595441875008591,
+      "loss": 1.072,
+      "step": 10530
+    },
+    {
+      "epoch": 0.8100146142604415,
+      "learning_rate": 0.0002593404302927791,
+      "loss": 1.074,
+      "step": 10531
+    },
+    {
+      "epoch": 0.8100915314206599,
+      "learning_rate": 0.0002591367455281047,
+      "loss": 1.1389,
+      "step": 10532
+    },
+    {
+      "epoch": 0.8101684485808784,
+      "learning_rate": 0.00025893313321872894,
+      "loss": 1.0699,
+      "step": 10533
+    },
+    {
+      "epoch": 0.8102453657410968,
+      "learning_rate": 0.00025872959337654096,
+      "loss": 1.2549,
+      "step": 10534
+    },
+    {
+      "epoch": 0.8103222829013152,
+      "learning_rate": 0.0002585261260134257,
+      "loss": 1.3641,
+      "step": 10535
+    },
+    {
+      "epoch": 0.8103992000615338,
+      "learning_rate": 0.0002583227311412633,
+      "loss": 0.9831,
+      "step": 10536
+    },
+    {
+      "epoch": 0.8104761172217522,
+      "learning_rate": 0.0002581194087719315,
+      "loss": 1.3057,
+      "step": 10537
+    },
+    {
+      "epoch": 0.8105530343819706,
+      "learning_rate": 0.0002579161589173013,
+      "loss": 1.1308,
+      "step": 10538
+    },
+    {
+      "epoch": 0.8106299515421891,
+      "learning_rate": 0.00025771298158924066,
+      "loss": 1.0699,
+      "step": 10539
+    },
+    {
+      "epoch": 0.8107068687024075,
+      "learning_rate": 0.00025750987679961415,
+      "loss": 1.2459,
+      "step": 10540
+    },
+    {
+      "epoch": 0.810783785862626,
+      "learning_rate": 0.0002573068445602809,
+      "loss": 1.464,
+      "step": 10541
+    },
+    {
+      "epoch": 0.8108607030228444,
+      "learning_rate": 0.0002571038848830959,
+      "loss": 1.5286,
+      "step": 10542
+    },
+    {
+      "epoch": 0.8109376201830628,
+      "learning_rate": 0.0002569009977799104,
+      "loss": 1.2353,
+      "step": 10543
+    },
+    {
+      "epoch": 0.8110145373432813,
+      "learning_rate": 0.0002566981832625709,
+      "loss": 1.4405,
+      "step": 10544
+    },
+    {
+      "epoch": 0.8110914545034997,
+      "learning_rate": 0.0002564954413429204,
+      "loss": 1.166,
+      "step": 10545
+    },
+    {
+      "epoch": 0.8111683716637181,
+      "learning_rate": 0.00025629277203279715,
+      "loss": 0.9393,
+      "step": 10546
+    },
+    {
+      "epoch": 0.8112452888239366,
+      "learning_rate": 0.0002560901753440352,
+      "loss": 0.853,
+      "step": 10547
+    },
+    {
+      "epoch": 0.811322205984155,
+      "learning_rate": 0.00025588765128846413,
+      "loss": 1.2401,
+      "step": 10548
+    },
+    {
+      "epoch": 0.8113991231443735,
+      "learning_rate": 0.00025568519987790976,
+      "loss": 1.0539,
+      "step": 10549
+    },
+    {
+      "epoch": 0.811476040304592,
+      "learning_rate": 0.0002554828211241933,
+      "loss": 1.314,
+      "step": 10550
+    },
+    {
+      "epoch": 0.8115529574648104,
+      "learning_rate": 0.0002552805150391321,
+      "loss": 1.0729,
+      "step": 10551
+    },
+    {
+      "epoch": 0.8116298746250289,
+      "learning_rate": 0.00025507828163453834,
+      "loss": 1.1186,
+      "step": 10552
+    },
+    {
+      "epoch": 0.8117067917852473,
+      "learning_rate": 0.00025487612092222147,
+      "loss": 1.188,
+      "step": 10553
+    },
+    {
+      "epoch": 0.8117837089454657,
+      "learning_rate": 0.0002546740329139855,
+      "loss": 1.4008,
+      "step": 10554
+    },
+    {
+      "epoch": 0.8118606261056842,
+      "learning_rate": 0.0002544720176216307,
+      "loss": 1.1238,
+      "step": 10555
+    },
+    {
+      "epoch": 0.8119375432659026,
+      "learning_rate": 0.00025427007505695275,
+      "loss": 1.4683,
+      "step": 10556
+    },
+    {
+      "epoch": 0.812014460426121,
+      "learning_rate": 0.00025406820523174364,
+      "loss": 1.5652,
+      "step": 10557
+    },
+    {
+      "epoch": 0.8120913775863395,
+      "learning_rate": 0.00025386640815778985,
+      "loss": 1.0449,
+      "step": 10558
+    },
+    {
+      "epoch": 0.8121682947465579,
+      "learning_rate": 0.0002536646838468756,
+      "loss": 1.2106,
+      "step": 10559
+    },
+    {
+      "epoch": 0.8122452119067765,
+      "learning_rate": 0.00025346303231077943,
+      "loss": 1.4836,
+      "step": 10560
+    },
+    {
+      "epoch": 0.8123221290669949,
+      "learning_rate": 0.000253261453561276,
+      "loss": 0.9808,
+      "step": 10561
+    },
+    {
+      "epoch": 0.8123990462272133,
+      "learning_rate": 0.0002530599476101354,
+      "loss": 0.9315,
+      "step": 10562
+    },
+    {
+      "epoch": 0.8124759633874318,
+      "learning_rate": 0.00025285851446912405,
+      "loss": 0.9325,
+      "step": 10563
+    },
+    {
+      "epoch": 0.8125528805476502,
+      "learning_rate": 0.0002526571541500039,
+      "loss": 0.9049,
+      "step": 10564
+    },
+    {
+      "epoch": 0.8126297977078686,
+      "learning_rate": 0.00025245586666453236,
+      "loss": 1.0214,
+      "step": 10565
+    },
+    {
+      "epoch": 0.8127067148680871,
+      "learning_rate": 0.0002522546520244627,
+      "loss": 1.0878,
+      "step": 10566
+    },
+    {
+      "epoch": 0.8127836320283055,
+      "learning_rate": 0.0002520535102415445,
+      "loss": 1.2121,
+      "step": 10567
+    },
+    {
+      "epoch": 0.8128605491885239,
+      "learning_rate": 0.0002518524413275224,
+      "loss": 1.1159,
+      "step": 10568
+    },
+    {
+      "epoch": 0.8129374663487424,
+      "learning_rate": 0.00025165144529413716,
+      "loss": 0.9496,
+      "step": 10569
+    },
+    {
+      "epoch": 0.8130143835089608,
+      "learning_rate": 0.00025145052215312484,
+      "loss": 1.2616,
+      "step": 10570
+    },
+    {
+      "epoch": 0.8130913006691793,
+      "learning_rate": 0.0002512496719162175,
+      "loss": 1.0543,
+      "step": 10571
+    },
+    {
+      "epoch": 0.8131682178293977,
+      "learning_rate": 0.0002510488945951437,
+      "loss": 1.1819,
+      "step": 10572
+    },
+    {
+      "epoch": 0.8132451349896161,
+      "learning_rate": 0.0002508481902016269,
+      "loss": 1.0739,
+      "step": 10573
+    },
+    {
+      "epoch": 0.8133220521498347,
+      "learning_rate": 0.00025064755874738517,
+      "loss": 0.873,
+      "step": 10574
+    },
+    {
+      "epoch": 0.8133989693100531,
+      "learning_rate": 0.000250447000244135,
+      "loss": 1.2985,
+      "step": 10575
+    },
+    {
+      "epoch": 0.8134758864702715,
+      "learning_rate": 0.0002502465147035868,
+      "loss": 1.3009,
+      "step": 10576
+    },
+    {
+      "epoch": 0.81355280363049,
+      "learning_rate": 0.00025004610213744707,
+      "loss": 1.106,
+      "step": 10577
+    },
+    {
+      "epoch": 0.8136297207907084,
+      "learning_rate": 0.00024984576255741815,
+      "loss": 0.9423,
+      "step": 10578
+    },
+    {
+      "epoch": 0.8137066379509269,
+      "learning_rate": 0.00024964549597519754,
+      "loss": 1.2203,
+      "step": 10579
+    },
+    {
+      "epoch": 0.8137835551111453,
+      "learning_rate": 0.00024944530240247984,
+      "loss": 0.921,
+      "step": 10580
+    },
+    {
+      "epoch": 0.8138604722713637,
+      "learning_rate": 0.0002492451818509543,
+      "loss": 0.9101,
+      "step": 10581
+    },
+    {
+      "epoch": 0.8139373894315822,
+      "learning_rate": 0.0002490451343323061,
+      "loss": 1.046,
+      "step": 10582
+    },
+    {
+      "epoch": 0.8140143065918006,
+      "learning_rate": 0.0002488451598582164,
+      "loss": 0.9688,
+      "step": 10583
+    },
+    {
+      "epoch": 0.814091223752019,
+      "learning_rate": 0.00024864525844036153,
+      "loss": 0.7727,
+      "step": 10584
+    },
+    {
+      "epoch": 0.8141681409122375,
+      "learning_rate": 0.0002484454300904137,
+      "loss": 1.0728,
+      "step": 10585
+    },
+    {
+      "epoch": 0.814245058072456,
+      "learning_rate": 0.0002482456748200425,
+      "loss": 0.9359,
+      "step": 10586
+    },
+    {
+      "epoch": 0.8143219752326744,
+      "learning_rate": 0.00024804599264091046,
+      "loss": 1.4848,
+      "step": 10587
+    },
+    {
+      "epoch": 0.8143988923928929,
+      "learning_rate": 0.00024784638356467717,
+      "loss": 0.8717,
+      "step": 10588
+    },
+    {
+      "epoch": 0.8144758095531113,
+      "learning_rate": 0.00024764684760299893,
+      "loss": 1.0713,
+      "step": 10589
+    },
+    {
+      "epoch": 0.8145527267133298,
+      "learning_rate": 0.00024744738476752647,
+      "loss": 1.663,
+      "step": 10590
+    },
+    {
+      "epoch": 0.8146296438735482,
+      "learning_rate": 0.00024724799506990665,
+      "loss": 0.8926,
+      "step": 10591
+    },
+    {
+      "epoch": 0.8147065610337666,
+      "learning_rate": 0.00024704867852178193,
+      "loss": 0.9111,
+      "step": 10592
+    },
+    {
+      "epoch": 0.8147834781939851,
+      "learning_rate": 0.00024684943513479034,
+      "loss": 1.673,
+      "step": 10593
+    },
+    {
+      "epoch": 0.8148603953542035,
+      "learning_rate": 0.00024665026492056666,
+      "loss": 0.9266,
+      "step": 10594
+    },
+    {
+      "epoch": 0.8149373125144219,
+      "learning_rate": 0.00024645116789074044,
+      "loss": 1.0953,
+      "step": 10595
+    },
+    {
+      "epoch": 0.8150142296746404,
+      "learning_rate": 0.00024625214405693617,
+      "loss": 1.0139,
+      "step": 10596
+    },
+    {
+      "epoch": 0.8150911468348588,
+      "learning_rate": 0.00024605319343077637,
+      "loss": 0.9901,
+      "step": 10597
+    },
+    {
+      "epoch": 0.8151680639950774,
+      "learning_rate": 0.0002458543160238769,
+      "loss": 1.2695,
+      "step": 10598
+    },
+    {
+      "epoch": 0.8152449811552958,
+      "learning_rate": 0.0002456555118478519,
+      "loss": 1.116,
+      "step": 10599
+    },
+    {
+      "epoch": 0.8153218983155142,
+      "learning_rate": 0.00024545678091430817,
+      "loss": 1.2579,
+      "step": 10600
+    },
+    {
+      "epoch": 0.8153988154757327,
+      "learning_rate": 0.00024525812323485026,
+      "loss": 1.2549,
+      "step": 10601
+    },
+    {
+      "epoch": 0.8154757326359511,
+      "learning_rate": 0.0002450595388210787,
+      "loss": 1.6772,
+      "step": 10602
+    },
+    {
+      "epoch": 0.8155526497961695,
+      "learning_rate": 0.0002448610276845883,
+      "loss": 1.288,
+      "step": 10603
+    },
+    {
+      "epoch": 0.815629566956388,
+      "learning_rate": 0.00024466258983697064,
+      "loss": 1.0606,
+      "step": 10604
+    },
+    {
+      "epoch": 0.8157064841166064,
+      "learning_rate": 0.00024446422528981264,
+      "loss": 1.4465,
+      "step": 10605
+    },
+    {
+      "epoch": 0.8157834012768248,
+      "learning_rate": 0.00024426593405469694,
+      "loss": 1.2004,
+      "step": 10606
+    },
+    {
+      "epoch": 0.8158603184370433,
+      "learning_rate": 0.0002440677161432019,
+      "loss": 1.357,
+      "step": 10607
+    },
+    {
+      "epoch": 0.8159372355972617,
+      "learning_rate": 0.00024386957156690225,
+      "loss": 1.0785,
+      "step": 10608
+    },
+    {
+      "epoch": 0.8160141527574802,
+      "learning_rate": 0.00024367150033736696,
+      "loss": 1.1325,
+      "step": 10609
+    },
+    {
+      "epoch": 0.8160910699176986,
+      "learning_rate": 0.00024347350246616223,
+      "loss": 1.4211,
+      "step": 10610
+    },
+    {
+      "epoch": 0.816167987077917,
+      "learning_rate": 0.00024327557796484934,
+      "loss": 1.2836,
+      "step": 10611
+    },
+    {
+      "epoch": 0.8162449042381356,
+      "learning_rate": 0.0002430777268449852,
+      "loss": 1.1152,
+      "step": 10612
+    },
+    {
+      "epoch": 0.816321821398354,
+      "learning_rate": 0.00024287994911812255,
+      "loss": 1.3357,
+      "step": 10613
+    },
+    {
+      "epoch": 0.8163987385585724,
+      "learning_rate": 0.00024268224479580969,
+      "loss": 1.209,
+      "step": 10614
+    },
+    {
+      "epoch": 0.8164756557187909,
+      "learning_rate": 0.00024248461388959053,
+      "loss": 1.2637,
+      "step": 10615
+    },
+    {
+      "epoch": 0.8165525728790093,
+      "learning_rate": 0.00024228705641100585,
+      "loss": 0.9578,
+      "step": 10616
+    },
+    {
+      "epoch": 0.8166294900392278,
+      "learning_rate": 0.00024208957237159057,
+      "loss": 1.1937,
+      "step": 10617
+    },
+    {
+      "epoch": 0.8167064071994462,
+      "learning_rate": 0.00024189216178287614,
+      "loss": 1.1317,
+      "step": 10618
+    },
+    {
+      "epoch": 0.8167833243596646,
+      "learning_rate": 0.00024169482465638954,
+      "loss": 1.4476,
+      "step": 10619
+    },
+    {
+      "epoch": 0.8168602415198831,
+      "learning_rate": 0.00024149756100365317,
+      "loss": 1.1497,
+      "step": 10620
+    },
+    {
+      "epoch": 0.8169371586801015,
+      "learning_rate": 0.0002413003708361865,
+      "loss": 1.1194,
+      "step": 10621
+    },
+    {
+      "epoch": 0.8170140758403199,
+      "learning_rate": 0.0002411032541655025,
+      "loss": 0.917,
+      "step": 10622
+    },
+    {
+      "epoch": 0.8170909930005384,
+      "learning_rate": 0.00024090621100311126,
+      "loss": 1.1918,
+      "step": 10623
+    },
+    {
+      "epoch": 0.8171679101607568,
+      "learning_rate": 0.0002407092413605188,
+      "loss": 1.0003,
+      "step": 10624
+    },
+    {
+      "epoch": 0.8172448273209754,
+      "learning_rate": 0.00024051234524922623,
+      "loss": 1.2944,
+      "step": 10625
+    },
+    {
+      "epoch": 0.8173217444811938,
+      "learning_rate": 0.0002403155226807305,
+      "loss": 1.279,
+      "step": 10626
+    },
+    {
+      "epoch": 0.8173986616414122,
+      "learning_rate": 0.00024011877366652407,
+      "loss": 1.1764,
+      "step": 10627
+    },
+    {
+      "epoch": 0.8174755788016307,
+      "learning_rate": 0.00023992209821809514,
+      "loss": 1.0114,
+      "step": 10628
+    },
+    {
+      "epoch": 0.8175524959618491,
+      "learning_rate": 0.00023972549634692847,
+      "loss": 0.8522,
+      "step": 10629
+    },
+    {
+      "epoch": 0.8176294131220675,
+      "learning_rate": 0.0002395289680645039,
+      "loss": 1.3193,
+      "step": 10630
+    },
+    {
+      "epoch": 0.817706330282286,
+      "learning_rate": 0.00023933251338229577,
+      "loss": 1.3139,
+      "step": 10631
+    },
+    {
+      "epoch": 0.8177832474425044,
+      "learning_rate": 0.0002391361323117764,
+      "loss": 1.1277,
+      "step": 10632
+    },
+    {
+      "epoch": 0.8178601646027228,
+      "learning_rate": 0.0002389398248644124,
+      "loss": 1.0314,
+      "step": 10633
+    },
+    {
+      "epoch": 0.8179370817629413,
+      "learning_rate": 0.00023874359105166616,
+      "loss": 0.8421,
+      "step": 10634
+    },
+    {
+      "epoch": 0.8180139989231597,
+      "learning_rate": 0.00023854743088499598,
+      "loss": 1.0115,
+      "step": 10635
+    },
+    {
+      "epoch": 0.8180909160833783,
+      "learning_rate": 0.0002383513443758557,
+      "loss": 1.1188,
+      "step": 10636
+    },
+    {
+      "epoch": 0.8181678332435967,
+      "learning_rate": 0.0002381553315356957,
+      "loss": 0.9807,
+      "step": 10637
+    },
+    {
+      "epoch": 0.8182447504038151,
+      "learning_rate": 0.00023795939237596092,
+      "loss": 1.0553,
+      "step": 10638
+    },
+    {
+      "epoch": 0.8183216675640336,
+      "learning_rate": 0.00023776352690809245,
+      "loss": 1.022,
+      "step": 10639
+    },
+    {
+      "epoch": 0.818398584724252,
+      "learning_rate": 0.00023756773514352708,
+      "loss": 1.2374,
+      "step": 10640
+    },
+    {
+      "epoch": 0.8184755018844704,
+      "learning_rate": 0.00023737201709369754,
+      "loss": 1.1,
+      "step": 10641
+    },
+    {
+      "epoch": 0.8185524190446889,
+      "learning_rate": 0.00023717637277003117,
+      "loss": 0.9657,
+      "step": 10642
+    },
+    {
+      "epoch": 0.8186293362049073,
+      "learning_rate": 0.00023698080218395335,
+      "loss": 1.0946,
+      "step": 10643
+    },
+    {
+      "epoch": 0.8187062533651258,
+      "learning_rate": 0.00023678530534688207,
+      "loss": 1.2417,
+      "step": 10644
+    },
+    {
+      "epoch": 0.8187831705253442,
+      "learning_rate": 0.0002365898822702336,
+      "loss": 0.8935,
+      "step": 10645
+    },
+    {
+      "epoch": 0.8188600876855626,
+      "learning_rate": 0.00023639453296541858,
+      "loss": 1.094,
+      "step": 10646
+    },
+    {
+      "epoch": 0.8189370048457811,
+      "learning_rate": 0.00023619925744384373,
+      "loss": 1.0916,
+      "step": 10647
+    },
+    {
+      "epoch": 0.8190139220059995,
+      "learning_rate": 0.0002360040557169113,
+      "loss": 1.0222,
+      "step": 10648
+    },
+    {
+      "epoch": 0.8190908391662179,
+      "learning_rate": 0.00023580892779601949,
+      "loss": 1.042,
+      "step": 10649
+    },
+    {
+      "epoch": 0.8191677563264365,
+      "learning_rate": 0.00023561387369256133,
+      "loss": 1.0733,
+      "step": 10650
+    },
+    {
+      "epoch": 0.8192446734866549,
+      "learning_rate": 0.00023541889341792722,
+      "loss": 1.036,
+      "step": 10651
+    },
+    {
+      "epoch": 0.8193215906468733,
+      "learning_rate": 0.0002352239869835024,
+      "loss": 1.2857,
+      "step": 10652
+    },
+    {
+      "epoch": 0.8193985078070918,
+      "learning_rate": 0.00023502915440066607,
+      "loss": 0.8074,
+      "step": 10653
+    },
+    {
+      "epoch": 0.8194754249673102,
+      "learning_rate": 0.00023483439568079635,
+      "loss": 1.1465,
+      "step": 10654
+    },
+    {
+      "epoch": 0.8195523421275287,
+      "learning_rate": 0.00023463971083526424,
+      "loss": 1.1189,
+      "step": 10655
+    },
+    {
+      "epoch": 0.8196292592877471,
+      "learning_rate": 0.00023444509987543887,
+      "loss": 1.1119,
+      "step": 10656
+    },
+    {
+      "epoch": 0.8197061764479655,
+      "learning_rate": 0.0002342505628126828,
+      "loss": 1.2435,
+      "step": 10657
+    },
+    {
+      "epoch": 0.819783093608184,
+      "learning_rate": 0.00023405609965835512,
+      "loss": 0.6836,
+      "step": 10658
+    },
+    {
+      "epoch": 0.8198600107684024,
+      "learning_rate": 0.00023386171042381155,
+      "loss": 1.1743,
+      "step": 10659
+    },
+    {
+      "epoch": 0.8199369279286208,
+      "learning_rate": 0.00023366739512040204,
+      "loss": 1.7128,
+      "step": 10660
+    },
+    {
+      "epoch": 0.8200138450888393,
+      "learning_rate": 0.00023347315375947314,
+      "loss": 1.4693,
+      "step": 10661
+    },
+    {
+      "epoch": 0.8200907622490577,
+      "learning_rate": 0.00023327898635236682,
+      "loss": 1.5164,
+      "step": 10662
+    },
+    {
+      "epoch": 0.8201676794092763,
+      "learning_rate": 0.00023308489291042035,
+      "loss": 1.0043,
+      "step": 10663
+    },
+    {
+      "epoch": 0.8202445965694947,
+      "learning_rate": 0.00023289087344496745,
+      "loss": 1.3352,
+      "step": 10664
+    },
+    {
+      "epoch": 0.8203215137297131,
+      "learning_rate": 0.00023269692796733742,
+      "loss": 0.928,
+      "step": 10665
+    },
+    {
+      "epoch": 0.8203984308899316,
+      "learning_rate": 0.00023250305648885388,
+      "loss": 1.1806,
+      "step": 10666
+    },
+    {
+      "epoch": 0.82047534805015,
+      "learning_rate": 0.00023230925902083827,
+      "loss": 1.5782,
+      "step": 10667
+    },
+    {
+      "epoch": 0.8205522652103684,
+      "learning_rate": 0.000232115535574606,
+      "loss": 1.1112,
+      "step": 10668
+    },
+    {
+      "epoch": 0.8206291823705869,
+      "learning_rate": 0.00023192188616146908,
+      "loss": 1.3983,
+      "step": 10669
+    },
+    {
+      "epoch": 0.8207060995308053,
+      "learning_rate": 0.00023172831079273476,
+      "loss": 0.8936,
+      "step": 10670
+    },
+    {
+      "epoch": 0.8207830166910237,
+      "learning_rate": 0.000231534809479706,
+      "loss": 1.1173,
+      "step": 10671
+    },
+    {
+      "epoch": 0.8208599338512422,
+      "learning_rate": 0.00023134138223368145,
+      "loss": 1.4427,
+      "step": 10672
+    },
+    {
+      "epoch": 0.8209368510114606,
+      "learning_rate": 0.00023114802906595622,
+      "loss": 0.9714,
+      "step": 10673
+    },
+    {
+      "epoch": 0.8210137681716791,
+      "learning_rate": 0.0002309547499878198,
+      "loss": 1.3299,
+      "step": 10674
+    },
+    {
+      "epoch": 0.8210906853318976,
+      "learning_rate": 0.00023076154501055812,
+      "loss": 1.1207,
+      "step": 10675
+    },
+    {
+      "epoch": 0.821167602492116,
+      "learning_rate": 0.00023056841414545253,
+      "loss": 1.7848,
+      "step": 10676
+    },
+    {
+      "epoch": 0.8212445196523345,
+      "learning_rate": 0.00023037535740377986,
+      "loss": 1.3192,
+      "step": 10677
+    },
+    {
+      "epoch": 0.8213214368125529,
+      "learning_rate": 0.00023018237479681403,
+      "loss": 1.3084,
+      "step": 10678
+    },
+    {
+      "epoch": 0.8213983539727713,
+      "learning_rate": 0.0002299894663358224,
+      "loss": 1.2936,
+      "step": 10679
+    },
+    {
+      "epoch": 0.8214752711329898,
+      "learning_rate": 0.00022979663203206895,
+      "loss": 1.2973,
+      "step": 10680
+    },
+    {
+      "epoch": 0.8215521882932082,
+      "learning_rate": 0.0002296038718968142,
+      "loss": 0.8722,
+      "step": 10681
+    },
+    {
+      "epoch": 0.8216291054534267,
+      "learning_rate": 0.0002294111859413136,
+      "loss": 1.0495,
+      "step": 10682
+    },
+    {
+      "epoch": 0.8217060226136451,
+      "learning_rate": 0.00022921857417681785,
+      "loss": 1.1018,
+      "step": 10683
+    },
+    {
+      "epoch": 0.8217829397738635,
+      "learning_rate": 0.00022902603661457393,
+      "loss": 1.0434,
+      "step": 10684
+    },
+    {
+      "epoch": 0.821859856934082,
+      "learning_rate": 0.00022883357326582403,
+      "loss": 1.5451,
+      "step": 10685
+    },
+    {
+      "epoch": 0.8219367740943004,
+      "learning_rate": 0.00022864118414180686,
+      "loss": 1.0836,
+      "step": 10686
+    },
+    {
+      "epoch": 0.8220136912545188,
+      "learning_rate": 0.00022844886925375618,
+      "loss": 0.9152,
+      "step": 10687
+    },
+    {
+      "epoch": 0.8220906084147374,
+      "learning_rate": 0.00022825662861290075,
+      "loss": 1.0145,
+      "step": 10688
+    },
+    {
+      "epoch": 0.8221675255749558,
+      "learning_rate": 0.0002280644622304664,
+      "loss": 1.1943,
+      "step": 10689
+    },
+    {
+      "epoch": 0.8222444427351742,
+      "learning_rate": 0.00022787237011767365,
+      "loss": 1.2422,
+      "step": 10690
+    },
+    {
+      "epoch": 0.8223213598953927,
+      "learning_rate": 0.00022768035228573884,
+      "loss": 1.1322,
+      "step": 10691
+    },
+    {
+      "epoch": 0.8223982770556111,
+      "learning_rate": 0.0002274884087458744,
+      "loss": 1.1814,
+      "step": 10692
+    },
+    {
+      "epoch": 0.8224751942158296,
+      "learning_rate": 0.00022729653950928763,
+      "loss": 0.9778,
+      "step": 10693
+    },
+    {
+      "epoch": 0.822552111376048,
+      "learning_rate": 0.0002271047445871826,
+      "loss": 0.9629,
+      "step": 10694
+    },
+    {
+      "epoch": 0.8226290285362664,
+      "learning_rate": 0.0002269130239907582,
+      "loss": 0.7977,
+      "step": 10695
+    },
+    {
+      "epoch": 0.8227059456964849,
+      "learning_rate": 0.00022672137773120893,
+      "loss": 1.2588,
+      "step": 10696
+    },
+    {
+      "epoch": 0.8227828628567033,
+      "learning_rate": 0.00022652980581972553,
+      "loss": 0.876,
+      "step": 10697
+    },
+    {
+      "epoch": 0.8228597800169217,
+      "learning_rate": 0.00022633830826749392,
+      "loss": 0.8721,
+      "step": 10698
+    },
+    {
+      "epoch": 0.8229366971771402,
+      "learning_rate": 0.00022614688508569558,
+      "loss": 1.2949,
+      "step": 10699
+    },
+    {
+      "epoch": 0.8230136143373586,
+      "learning_rate": 0.00022595553628550847,
+      "loss": 1.1846,
+      "step": 10700
+    },
+    {
+      "epoch": 0.8230905314975772,
+      "learning_rate": 0.0002257642618781053,
+      "loss": 0.7855,
+      "step": 10701
+    },
+    {
+      "epoch": 0.8231674486577956,
+      "learning_rate": 0.00022557306187465494,
+      "loss": 1.0839,
+      "step": 10702
+    },
+    {
+      "epoch": 0.823244365818014,
+      "learning_rate": 0.00022538193628632164,
+      "loss": 0.9837,
+      "step": 10703
+    },
+    {
+      "epoch": 0.8233212829782325,
+      "learning_rate": 0.0002251908851242655,
+      "loss": 1.0355,
+      "step": 10704
+    },
+    {
+      "epoch": 0.8233982001384509,
+      "learning_rate": 0.00022499990839964224,
+      "loss": 0.9371,
+      "step": 10705
+    },
+    {
+      "epoch": 0.8234751172986693,
+      "learning_rate": 0.00022480900612360295,
+      "loss": 1.0933,
+      "step": 10706
+    },
+    {
+      "epoch": 0.8235520344588878,
+      "learning_rate": 0.00022461817830729435,
+      "loss": 1.2113,
+      "step": 10707
+    },
+    {
+      "epoch": 0.8236289516191062,
+      "learning_rate": 0.00022442742496186008,
+      "loss": 1.0762,
+      "step": 10708
+    },
+    {
+      "epoch": 0.8237058687793246,
+      "learning_rate": 0.0002242367460984377,
+      "loss": 1.0536,
+      "step": 10709
+    },
+    {
+      "epoch": 0.8237827859395431,
+      "learning_rate": 0.0002240461417281613,
+      "loss": 1.0907,
+      "step": 10710
+    },
+    {
+      "epoch": 0.8238597030997615,
+      "learning_rate": 0.00022385561186216046,
+      "loss": 1.6937,
+      "step": 10711
+    },
+    {
+      "epoch": 0.82393662025998,
+      "learning_rate": 0.00022366515651155999,
+      "loss": 1.5094,
+      "step": 10712
+    },
+    {
+      "epoch": 0.8240135374201984,
+      "learning_rate": 0.00022347477568748165,
+      "loss": 1.0817,
+      "step": 10713
+    },
+    {
+      "epoch": 0.8240904545804169,
+      "learning_rate": 0.00022328446940104185,
+      "loss": 1.0013,
+      "step": 10714
+    },
+    {
+      "epoch": 0.8241673717406354,
+      "learning_rate": 0.00022309423766335158,
+      "loss": 0.9153,
+      "step": 10715
+    },
+    {
+      "epoch": 0.8242442889008538,
+      "learning_rate": 0.00022290408048552008,
+      "loss": 1.051,
+      "step": 10716
+    },
+    {
+      "epoch": 0.8243212060610722,
+      "learning_rate": 0.00022271399787865016,
+      "loss": 1.0477,
+      "step": 10717
+    },
+    {
+      "epoch": 0.8243981232212907,
+      "learning_rate": 0.0002225239898538411,
+      "loss": 1.0299,
+      "step": 10718
+    },
+    {
+      "epoch": 0.8244750403815091,
+      "learning_rate": 0.00022233405642218767,
+      "loss": 1.0845,
+      "step": 10719
+    },
+    {
+      "epoch": 0.8245519575417276,
+      "learning_rate": 0.00022214419759477989,
+      "loss": 1.2588,
+      "step": 10720
+    },
+    {
+      "epoch": 0.824628874701946,
+      "learning_rate": 0.0002219544133827045,
+      "loss": 1.2467,
+      "step": 10721
+    },
+    {
+      "epoch": 0.8247057918621644,
+      "learning_rate": 0.00022176470379704283,
+      "loss": 1.2005,
+      "step": 10722
+    },
+    {
+      "epoch": 0.8247827090223829,
+      "learning_rate": 0.0002215750688488722,
+      "loss": 0.8573,
+      "step": 10723
+    },
+    {
+      "epoch": 0.8248596261826013,
+      "learning_rate": 0.0002213855085492657,
+      "loss": 1.2205,
+      "step": 10724
+    },
+    {
+      "epoch": 0.8249365433428197,
+      "learning_rate": 0.00022119602290929187,
+      "loss": 1.3672,
+      "step": 10725
+    },
+    {
+      "epoch": 0.8250134605030383,
+      "learning_rate": 0.00022100661194001482,
+      "loss": 1.318,
+      "step": 10726
+    },
+    {
+      "epoch": 0.8250903776632567,
+      "learning_rate": 0.00022081727565249505,
+      "loss": 1.3921,
+      "step": 10727
+    },
+    {
+      "epoch": 0.8251672948234752,
+      "learning_rate": 0.00022062801405778722,
+      "loss": 1.3497,
+      "step": 10728
+    },
+    {
+      "epoch": 0.8252442119836936,
+      "learning_rate": 0.00022043882716694314,
+      "loss": 0.9178,
+      "step": 10729
+    },
+    {
+      "epoch": 0.825321129143912,
+      "learning_rate": 0.00022024971499100965,
+      "loss": 0.969,
+      "step": 10730
+    },
+    {
+      "epoch": 0.8253980463041305,
+      "learning_rate": 0.00022006067754102893,
+      "loss": 0.9037,
+      "step": 10731
+    },
+    {
+      "epoch": 0.8254749634643489,
+      "learning_rate": 0.00021987171482803914,
+      "loss": 1.1075,
+      "step": 10732
+    },
+    {
+      "epoch": 0.8255518806245673,
+      "learning_rate": 0.00021968282686307412,
+      "loss": 1.393,
+      "step": 10733
+    },
+    {
+      "epoch": 0.8256287977847858,
+      "learning_rate": 0.0002194940136571627,
+      "loss": 1.3812,
+      "step": 10734
+    },
+    {
+      "epoch": 0.8257057149450042,
+      "learning_rate": 0.00021930527522133077,
+      "loss": 0.9854,
+      "step": 10735
+    },
+    {
+      "epoch": 0.8257826321052226,
+      "learning_rate": 0.0002191166115665988,
+      "loss": 0.9693,
+      "step": 10736
+    },
+    {
+      "epoch": 0.8258595492654411,
+      "learning_rate": 0.00021892802270398216,
+      "loss": 1.1318,
+      "step": 10737
+    },
+    {
+      "epoch": 0.8259364664256595,
+      "learning_rate": 0.0002187395086444937,
+      "loss": 1.0286,
+      "step": 10738
+    },
+    {
+      "epoch": 0.8260133835858781,
+      "learning_rate": 0.0002185510693991403,
+      "loss": 1.1151,
+      "step": 10739
+    },
+    {
+      "epoch": 0.8260903007460965,
+      "learning_rate": 0.0002183627049789262,
+      "loss": 1.4246,
+      "step": 10740
+    },
+    {
+      "epoch": 0.8261672179063149,
+      "learning_rate": 0.000218174415394849,
+      "loss": 1.0629,
+      "step": 10741
+    },
+    {
+      "epoch": 0.8262441350665334,
+      "learning_rate": 0.0002179862006579033,
+      "loss": 1.3737,
+      "step": 10742
+    },
+    {
+      "epoch": 0.8263210522267518,
+      "learning_rate": 0.00021779806077907966,
+      "loss": 1.0107,
+      "step": 10743
+    },
+    {
+      "epoch": 0.8263979693869702,
+      "learning_rate": 0.00021760999576936357,
+      "loss": 1.1733,
+      "step": 10744
+    },
+    {
+      "epoch": 0.8264748865471887,
+      "learning_rate": 0.0002174220056397363,
+      "loss": 1.6141,
+      "step": 10745
+    },
+    {
+      "epoch": 0.8265518037074071,
+      "learning_rate": 0.00021723409040117487,
+      "loss": 1.1024,
+      "step": 10746
+    },
+    {
+      "epoch": 0.8266287208676256,
+      "learning_rate": 0.00021704625006465134,
+      "loss": 1.4375,
+      "step": 10747
+    },
+    {
+      "epoch": 0.826705638027844,
+      "learning_rate": 0.0002168584846411348,
+      "loss": 1.1996,
+      "step": 10748
+    },
+    {
+      "epoch": 0.8267825551880624,
+      "learning_rate": 0.00021667079414158875,
+      "loss": 1.2223,
+      "step": 10749
+    },
+    {
+      "epoch": 0.8268594723482809,
+      "learning_rate": 0.00021648317857697198,
+      "loss": 1.2452,
+      "step": 10750
+    },
+    {
+      "epoch": 0.8269363895084993,
+      "learning_rate": 0.00021629563795824035,
+      "loss": 1.4031,
+      "step": 10751
+    },
+    {
+      "epoch": 0.8270133066687178,
+      "learning_rate": 0.00021610817229634417,
+      "loss": 0.9523,
+      "step": 10752
+    },
+    {
+      "epoch": 0.8270902238289363,
+      "learning_rate": 0.00021592078160222994,
+      "loss": 0.6703,
+      "step": 10753
+    },
+    {
+      "epoch": 0.8271671409891547,
+      "learning_rate": 0.00021573346588683946,
+      "loss": 1.1687,
+      "step": 10754
+    },
+    {
+      "epoch": 0.8272440581493731,
+      "learning_rate": 0.00021554622516111043,
+      "loss": 1.5381,
+      "step": 10755
+    },
+    {
+      "epoch": 0.8273209753095916,
+      "learning_rate": 0.0002153590594359755,
+      "loss": 1.1097,
+      "step": 10756
+    },
+    {
+      "epoch": 0.82739789246981,
+      "learning_rate": 0.0002151719687223642,
+      "loss": 0.9915,
+      "step": 10757
+    },
+    {
+      "epoch": 0.8274748096300285,
+      "learning_rate": 0.00021498495303120085,
+      "loss": 1.5455,
+      "step": 10758
+    },
+    {
+      "epoch": 0.8275517267902469,
+      "learning_rate": 0.00021479801237340518,
+      "loss": 1.4729,
+      "step": 10759
+    },
+    {
+      "epoch": 0.8276286439504653,
+      "learning_rate": 0.00021461114675989297,
+      "loss": 1.0881,
+      "step": 10760
+    },
+    {
+      "epoch": 0.8277055611106838,
+      "learning_rate": 0.00021442435620157513,
+      "loss": 1.2372,
+      "step": 10761
+    },
+    {
+      "epoch": 0.8277824782709022,
+      "learning_rate": 0.00021423764070935963,
+      "loss": 1.3053,
+      "step": 10762
+    },
+    {
+      "epoch": 0.8278593954311206,
+      "learning_rate": 0.0002140510002941481,
+      "loss": 1.3,
+      "step": 10763
+    },
+    {
+      "epoch": 0.8279363125913392,
+      "learning_rate": 0.00021386443496683826,
+      "loss": 0.9578,
+      "step": 10764
+    },
+    {
+      "epoch": 0.8280132297515576,
+      "learning_rate": 0.00021367794473832496,
+      "loss": 1.2897,
+      "step": 10765
+    },
+    {
+      "epoch": 0.8280901469117761,
+      "learning_rate": 0.00021349152961949697,
+      "loss": 1.1766,
+      "step": 10766
+    },
+    {
+      "epoch": 0.8281670640719945,
+      "learning_rate": 0.0002133051896212394,
+      "loss": 1.1793,
+      "step": 10767
+    },
+    {
+      "epoch": 0.8282439812322129,
+      "learning_rate": 0.00021311892475443258,
+      "loss": 0.8118,
+      "step": 10768
+    },
+    {
+      "epoch": 0.8283208983924314,
+      "learning_rate": 0.00021293273502995274,
+      "loss": 1.0039,
+      "step": 10769
+    },
+    {
+      "epoch": 0.8283978155526498,
+      "learning_rate": 0.0002127466204586721,
+      "loss": 0.9122,
+      "step": 10770
+    },
+    {
+      "epoch": 0.8284747327128682,
+      "learning_rate": 0.00021256058105145826,
+      "loss": 1.2355,
+      "step": 10771
+    },
+    {
+      "epoch": 0.8285516498730867,
+      "learning_rate": 0.00021237461681917303,
+      "loss": 0.9744,
+      "step": 10772
+    },
+    {
+      "epoch": 0.8286285670333051,
+      "learning_rate": 0.00021218872777267623,
+      "loss": 0.9132,
+      "step": 10773
+    },
+    {
+      "epoch": 0.8287054841935235,
+      "learning_rate": 0.00021200291392282184,
+      "loss": 1.4135,
+      "step": 10774
+    },
+    {
+      "epoch": 0.828782401353742,
+      "learning_rate": 0.00021181717528045952,
+      "loss": 1.5483,
+      "step": 10775
+    },
+    {
+      "epoch": 0.8288593185139604,
+      "learning_rate": 0.00021163151185643492,
+      "loss": 1.1269,
+      "step": 10776
+    },
+    {
+      "epoch": 0.828936235674179,
+      "learning_rate": 0.00021144592366158872,
+      "loss": 1.0945,
+      "step": 10777
+    },
+    {
+      "epoch": 0.8290131528343974,
+      "learning_rate": 0.00021126041070675821,
+      "loss": 1.1136,
+      "step": 10778
+    },
+    {
+      "epoch": 0.8290900699946158,
+      "learning_rate": 0.00021107497300277544,
+      "loss": 0.9133,
+      "step": 10779
+    },
+    {
+      "epoch": 0.8291669871548343,
+      "learning_rate": 0.00021088961056046822,
+      "loss": 0.9289,
+      "step": 10780
+    },
+    {
+      "epoch": 0.8292439043150527,
+      "learning_rate": 0.00021070432339066026,
+      "loss": 1.1636,
+      "step": 10781
+    },
+    {
+      "epoch": 0.8293208214752711,
+      "learning_rate": 0.00021051911150417037,
+      "loss": 1.0531,
+      "step": 10782
+    },
+    {
+      "epoch": 0.8293977386354896,
+      "learning_rate": 0.00021033397491181322,
+      "loss": 1.2099,
+      "step": 10783
+    },
+    {
+      "epoch": 0.829474655795708,
+      "learning_rate": 0.00021014891362439986,
+      "loss": 1.0567,
+      "step": 10784
+    },
+    {
+      "epoch": 0.8295515729559265,
+      "learning_rate": 0.00020996392765273514,
+      "loss": 0.7669,
+      "step": 10785
+    },
+    {
+      "epoch": 0.8296284901161449,
+      "learning_rate": 0.0002097790170076214,
+      "loss": 1.1889,
+      "step": 10786
+    },
+    {
+      "epoch": 0.8297054072763633,
+      "learning_rate": 0.00020959418169985556,
+      "loss": 0.9922,
+      "step": 10787
+    },
+    {
+      "epoch": 0.8297823244365818,
+      "learning_rate": 0.0002094094217402301,
+      "loss": 1.3012,
+      "step": 10788
+    },
+    {
+      "epoch": 0.8298592415968002,
+      "learning_rate": 0.00020922473713953373,
+      "loss": 1.0292,
+      "step": 10789
+    },
+    {
+      "epoch": 0.8299361587570186,
+      "learning_rate": 0.00020904012790855014,
+      "loss": 1.3112,
+      "step": 10790
+    },
+    {
+      "epoch": 0.8300130759172372,
+      "learning_rate": 0.00020885559405805842,
+      "loss": 1.3005,
+      "step": 10791
+    },
+    {
+      "epoch": 0.8300899930774556,
+      "learning_rate": 0.0002086711355988346,
+      "loss": 1.5351,
+      "step": 10792
+    },
+    {
+      "epoch": 0.830166910237674,
+      "learning_rate": 0.00020848675254164922,
+      "loss": 1.3236,
+      "step": 10793
+    },
+    {
+      "epoch": 0.8302438273978925,
+      "learning_rate": 0.00020830244489726784,
+      "loss": 1.219,
+      "step": 10794
+    },
+    {
+      "epoch": 0.8303207445581109,
+      "learning_rate": 0.00020811821267645303,
+      "loss": 1.4288,
+      "step": 10795
+    },
+    {
+      "epoch": 0.8303976617183294,
+      "learning_rate": 0.00020793405588996202,
+      "loss": 1.5462,
+      "step": 10796
+    },
+    {
+      "epoch": 0.8304745788785478,
+      "learning_rate": 0.0002077499745485485,
+      "loss": 1.122,
+      "step": 10797
+    },
+    {
+      "epoch": 0.8305514960387662,
+      "learning_rate": 0.0002075659686629604,
+      "loss": 1.2538,
+      "step": 10798
+    },
+    {
+      "epoch": 0.8306284131989847,
+      "learning_rate": 0.00020738203824394213,
+      "loss": 1.1927,
+      "step": 10799
+    },
+    {
+      "epoch": 0.8307053303592031,
+      "learning_rate": 0.0002071981833022341,
+      "loss": 1.1776,
+      "step": 10800
+    },
+    {
+      "epoch": 0.8307822475194215,
+      "learning_rate": 0.00020701440384857139,
+      "loss": 1.1312,
+      "step": 10801
+    },
+    {
+      "epoch": 0.83085916467964,
+      "learning_rate": 0.00020683069989368524,
+      "loss": 1.0845,
+      "step": 10802
+    },
+    {
+      "epoch": 0.8309360818398585,
+      "learning_rate": 0.00020664707144830207,
+      "loss": 1.2887,
+      "step": 10803
+    },
+    {
+      "epoch": 0.831012999000077,
+      "learning_rate": 0.00020646351852314415,
+      "loss": 1.193,
+      "step": 10804
+    },
+    {
+      "epoch": 0.8310899161602954,
+      "learning_rate": 0.00020628004112892973,
+      "loss": 1.2198,
+      "step": 10805
+    },
+    {
+      "epoch": 0.8311668333205138,
+      "learning_rate": 0.0002060966392763724,
+      "loss": 1.1254,
+      "step": 10806
+    },
+    {
+      "epoch": 0.8312437504807323,
+      "learning_rate": 0.00020591331297617994,
+      "loss": 1.6282,
+      "step": 10807
+    },
+    {
+      "epoch": 0.8313206676409507,
+      "learning_rate": 0.00020573006223905828,
+      "loss": 1.0798,
+      "step": 10808
+    },
+    {
+      "epoch": 0.8313975848011691,
+      "learning_rate": 0.00020554688707570703,
+      "loss": 1.1077,
+      "step": 10809
+    },
+    {
+      "epoch": 0.8314745019613876,
+      "learning_rate": 0.00020536378749682227,
+      "loss": 1.4114,
+      "step": 10810
+    },
+    {
+      "epoch": 0.831551419121606,
+      "learning_rate": 0.00020518076351309502,
+      "loss": 1.2684,
+      "step": 10811
+    },
+    {
+      "epoch": 0.8316283362818244,
+      "learning_rate": 0.00020499781513521248,
+      "loss": 1.2452,
+      "step": 10812
+    },
+    {
+      "epoch": 0.8317052534420429,
+      "learning_rate": 0.00020481494237385684,
+      "loss": 1.1362,
+      "step": 10813
+    },
+    {
+      "epoch": 0.8317821706022613,
+      "learning_rate": 0.00020463214523970668,
+      "loss": 1.2861,
+      "step": 10814
+    },
+    {
+      "epoch": 0.8318590877624799,
+      "learning_rate": 0.00020444942374343584,
+      "loss": 0.899,
+      "step": 10815
+    },
+    {
+      "epoch": 0.8319360049226983,
+      "learning_rate": 0.00020426677789571308,
+      "loss": 1.0705,
+      "step": 10816
+    },
+    {
+      "epoch": 0.8320129220829167,
+      "learning_rate": 0.0002040842077072037,
+      "loss": 0.8464,
+      "step": 10817
+    },
+    {
+      "epoch": 0.8320898392431352,
+      "learning_rate": 0.0002039017131885677,
+      "loss": 1.2695,
+      "step": 10818
+    },
+    {
+      "epoch": 0.8321667564033536,
+      "learning_rate": 0.00020371929435046188,
+      "loss": 1.0746,
+      "step": 10819
+    },
+    {
+      "epoch": 0.832243673563572,
+      "learning_rate": 0.00020353695120353723,
+      "loss": 0.9288,
+      "step": 10820
+    },
+    {
+      "epoch": 0.8323205907237905,
+      "learning_rate": 0.00020335468375844086,
+      "loss": 1.1749,
+      "step": 10821
+    },
+    {
+      "epoch": 0.8323975078840089,
+      "learning_rate": 0.0002031724920258161,
+      "loss": 1.0873,
+      "step": 10822
+    },
+    {
+      "epoch": 0.8324744250442274,
+      "learning_rate": 0.000202990376016301,
+      "loss": 1.2259,
+      "step": 10823
+    },
+    {
+      "epoch": 0.8325513422044458,
+      "learning_rate": 0.00020280833574052966,
+      "loss": 1.2233,
+      "step": 10824
+    },
+    {
+      "epoch": 0.8326282593646642,
+      "learning_rate": 0.0002026263712091314,
+      "loss": 1.1299,
+      "step": 10825
+    },
+    {
+      "epoch": 0.8327051765248827,
+      "learning_rate": 0.0002024444824327311,
+      "loss": 1.2207,
+      "step": 10826
+    },
+    {
+      "epoch": 0.8327820936851011,
+      "learning_rate": 0.00020226266942195025,
+      "loss": 1.6364,
+      "step": 10827
+    },
+    {
+      "epoch": 0.8328590108453195,
+      "learning_rate": 0.0002020809321874047,
+      "loss": 0.9727,
+      "step": 10828
+    },
+    {
+      "epoch": 0.8329359280055381,
+      "learning_rate": 0.0002018992707397056,
+      "loss": 1.3102,
+      "step": 10829
+    },
+    {
+      "epoch": 0.8330128451657565,
+      "learning_rate": 0.0002017176850894613,
+      "loss": 1.2125,
+      "step": 10830
+    },
+    {
+      "epoch": 0.8330897623259749,
+      "learning_rate": 0.00020153617524727436,
+      "loss": 1.2022,
+      "step": 10831
+    },
+    {
+      "epoch": 0.8331666794861934,
+      "learning_rate": 0.00020135474122374343,
+      "loss": 0.9467,
+      "step": 10832
+    },
+    {
+      "epoch": 0.8332435966464118,
+      "learning_rate": 0.00020117338302946252,
+      "loss": 0.5734,
+      "step": 10833
+    },
+    {
+      "epoch": 0.8333205138066303,
+      "learning_rate": 0.00020099210067502104,
+      "loss": 1.1923,
+      "step": 10834
+    },
+    {
+      "epoch": 0.8333974309668487,
+      "learning_rate": 0.000200810894171005,
+      "loss": 0.9654,
+      "step": 10835
+    },
+    {
+      "epoch": 0.8334743481270671,
+      "learning_rate": 0.00020062976352799478,
+      "loss": 1.4249,
+      "step": 10836
+    },
+    {
+      "epoch": 0.8335512652872856,
+      "learning_rate": 0.0002004487087565669,
+      "loss": 1.0348,
+      "step": 10837
+    },
+    {
+      "epoch": 0.833628182447504,
+      "learning_rate": 0.00020026772986729324,
+      "loss": 0.9627,
+      "step": 10838
+    },
+    {
+      "epoch": 0.8337050996077224,
+      "learning_rate": 0.00020008682687074137,
+      "loss": 1.2621,
+      "step": 10839
+    },
+    {
+      "epoch": 0.833782016767941,
+      "learning_rate": 0.00019990599977747398,
+      "loss": 1.4122,
+      "step": 10840
+    },
+    {
+      "epoch": 0.8338589339281594,
+      "learning_rate": 0.00019972524859805096,
+      "loss": 0.7352,
+      "step": 10841
+    },
+    {
+      "epoch": 0.8339358510883779,
+      "learning_rate": 0.00019954457334302522,
+      "loss": 1.0593,
+      "step": 10842
+    },
+    {
+      "epoch": 0.8340127682485963,
+      "learning_rate": 0.00019936397402294732,
+      "loss": 0.8924,
+      "step": 10843
+    },
+    {
+      "epoch": 0.8340896854088147,
+      "learning_rate": 0.00019918345064836245,
+      "loss": 1.2799,
+      "step": 10844
+    },
+    {
+      "epoch": 0.8341666025690332,
+      "learning_rate": 0.00019900300322981174,
+      "loss": 1.2259,
+      "step": 10845
+    },
+    {
+      "epoch": 0.8342435197292516,
+      "learning_rate": 0.00019882263177783139,
+      "loss": 0.8012,
+      "step": 10846
+    },
+    {
+      "epoch": 0.83432043688947,
+      "learning_rate": 0.00019864233630295365,
+      "loss": 1.1015,
+      "step": 10847
+    },
+    {
+      "epoch": 0.8343973540496885,
+      "learning_rate": 0.00019846211681570592,
+      "loss": 1.0674,
+      "step": 10848
+    },
+    {
+      "epoch": 0.8344742712099069,
+      "learning_rate": 0.0001982819733266118,
+      "loss": 1.1149,
+      "step": 10849
+    },
+    {
+      "epoch": 0.8345511883701254,
+      "learning_rate": 0.00019810190584618987,
+      "loss": 1.0664,
+      "step": 10850
+    },
+    {
+      "epoch": 0.8346281055303438,
+      "learning_rate": 0.00019792191438495454,
+      "loss": 1.2631,
+      "step": 10851
+    },
+    {
+      "epoch": 0.8347050226905622,
+      "learning_rate": 0.0001977419989534156,
+      "loss": 0.9738,
+      "step": 10852
+    },
+    {
+      "epoch": 0.8347819398507808,
+      "learning_rate": 0.0001975621595620783,
+      "loss": 1.1769,
+      "step": 10853
+    },
+    {
+      "epoch": 0.8348588570109992,
+      "learning_rate": 0.0001973823962214444,
+      "loss": 1.1159,
+      "step": 10854
+    },
+    {
+      "epoch": 0.8349357741712176,
+      "learning_rate": 0.00019720270894200964,
+      "loss": 1.1199,
+      "step": 10855
+    },
+    {
+      "epoch": 0.8350126913314361,
+      "learning_rate": 0.00019702309773426618,
+      "loss": 1.0585,
+      "step": 10856
+    },
+    {
+      "epoch": 0.8350896084916545,
+      "learning_rate": 0.00019684356260870212,
+      "loss": 0.7886,
+      "step": 10857
+    },
+    {
+      "epoch": 0.8351665256518729,
+      "learning_rate": 0.00019666410357580073,
+      "loss": 1.6249,
+      "step": 10858
+    },
+    {
+      "epoch": 0.8352434428120914,
+      "learning_rate": 0.00019648472064604062,
+      "loss": 1.015,
+      "step": 10859
+    },
+    {
+      "epoch": 0.8353203599723098,
+      "learning_rate": 0.00019630541382989608,
+      "loss": 1.0349,
+      "step": 10860
+    },
+    {
+      "epoch": 0.8353972771325283,
+      "learning_rate": 0.00019612618313783693,
+      "loss": 1.2132,
+      "step": 10861
+    },
+    {
+      "epoch": 0.8354741942927467,
+      "learning_rate": 0.0001959470285803289,
+      "loss": 1.3673,
+      "step": 10862
+    },
+    {
+      "epoch": 0.8355511114529651,
+      "learning_rate": 0.00019576795016783304,
+      "loss": 0.7714,
+      "step": 10863
+    },
+    {
+      "epoch": 0.8356280286131836,
+      "learning_rate": 0.00019558894791080596,
+      "loss": 0.9195,
+      "step": 10864
+    },
+    {
+      "epoch": 0.835704945773402,
+      "learning_rate": 0.00019541002181969942,
+      "loss": 1.4968,
+      "step": 10865
+    },
+    {
+      "epoch": 0.8357818629336204,
+      "learning_rate": 0.00019523117190496143,
+      "loss": 0.7924,
+      "step": 10866
+    },
+    {
+      "epoch": 0.835858780093839,
+      "learning_rate": 0.00019505239817703513,
+      "loss": 1.1613,
+      "step": 10867
+    },
+    {
+      "epoch": 0.8359356972540574,
+      "learning_rate": 0.00019487370064635918,
+      "loss": 1.0735,
+      "step": 10868
+    },
+    {
+      "epoch": 0.8360126144142759,
+      "learning_rate": 0.00019469507932336772,
+      "loss": 1.4261,
+      "step": 10869
+    },
+    {
+      "epoch": 0.8360895315744943,
+      "learning_rate": 0.00019451653421849124,
+      "loss": 1.076,
+      "step": 10870
+    },
+    {
+      "epoch": 0.8361664487347127,
+      "learning_rate": 0.00019433806534215493,
+      "loss": 1.3461,
+      "step": 10871
+    },
+    {
+      "epoch": 0.8362433658949312,
+      "learning_rate": 0.00019415967270477975,
+      "loss": 0.9975,
+      "step": 10872
+    },
+    {
+      "epoch": 0.8363202830551496,
+      "learning_rate": 0.0001939813563167822,
+      "loss": 1.06,
+      "step": 10873
+    },
+    {
+      "epoch": 0.836397200215368,
+      "learning_rate": 0.00019380311618857432,
+      "loss": 1.4747,
+      "step": 10874
+    },
+    {
+      "epoch": 0.8364741173755865,
+      "learning_rate": 0.00019362495233056344,
+      "loss": 1.2087,
+      "step": 10875
+    },
+    {
+      "epoch": 0.8365510345358049,
+      "learning_rate": 0.0001934468647531532,
+      "loss": 1.3013,
+      "step": 10876
+    },
+    {
+      "epoch": 0.8366279516960233,
+      "learning_rate": 0.00019326885346674283,
+      "loss": 0.9965,
+      "step": 10877
+    },
+    {
+      "epoch": 0.8367048688562418,
+      "learning_rate": 0.00019309091848172517,
+      "loss": 1.1463,
+      "step": 10878
+    },
+    {
+      "epoch": 0.8367817860164602,
+      "learning_rate": 0.00019291305980849105,
+      "loss": 1.2645,
+      "step": 10879
+    },
+    {
+      "epoch": 0.8368587031766788,
+      "learning_rate": 0.00019273527745742537,
+      "loss": 1.1815,
+      "step": 10880
+    },
+    {
+      "epoch": 0.8369356203368972,
+      "learning_rate": 0.00019255757143890977,
+      "loss": 1.062,
+      "step": 10881
+    },
+    {
+      "epoch": 0.8370125374971156,
+      "learning_rate": 0.00019237994176331984,
+      "loss": 1.0678,
+      "step": 10882
+    },
+    {
+      "epoch": 0.8370894546573341,
+      "learning_rate": 0.00019220238844102762,
+      "loss": 1.1956,
+      "step": 10883
+    },
+    {
+      "epoch": 0.8371663718175525,
+      "learning_rate": 0.00019202491148240125,
+      "loss": 1.1834,
+      "step": 10884
+    },
+    {
+      "epoch": 0.8372432889777709,
+      "learning_rate": 0.0001918475108978035,
+      "loss": 1.1271,
+      "step": 10885
+    },
+    {
+      "epoch": 0.8373202061379894,
+      "learning_rate": 0.00019167018669759273,
+      "loss": 1.2681,
+      "step": 10886
+    },
+    {
+      "epoch": 0.8373971232982078,
+      "learning_rate": 0.00019149293889212334,
+      "loss": 1.3992,
+      "step": 10887
+    },
+    {
+      "epoch": 0.8374740404584263,
+      "learning_rate": 0.00019131576749174467,
+      "loss": 1.0591,
+      "step": 10888
+    },
+    {
+      "epoch": 0.8375509576186447,
+      "learning_rate": 0.0001911386725068025,
+      "loss": 1.1505,
+      "step": 10889
+    },
+    {
+      "epoch": 0.8376278747788631,
+      "learning_rate": 0.00019096165394763754,
+      "loss": 0.7873,
+      "step": 10890
+    },
+    {
+      "epoch": 0.8377047919390816,
+      "learning_rate": 0.00019078471182458535,
+      "loss": 1.3096,
+      "step": 10891
+    },
+    {
+      "epoch": 0.8377817090993,
+      "learning_rate": 0.00019060784614797848,
+      "loss": 1.1801,
+      "step": 10892
+    },
+    {
+      "epoch": 0.8378586262595185,
+      "learning_rate": 0.0001904310569281442,
+      "loss": 1.2087,
+      "step": 10893
+    },
+    {
+      "epoch": 0.837935543419737,
+      "learning_rate": 0.00019025434417540527,
+      "loss": 1.1831,
+      "step": 10894
+    },
+    {
+      "epoch": 0.8380124605799554,
+      "learning_rate": 0.00019007770790008006,
+      "loss": 1.3378,
+      "step": 10895
+    },
+    {
+      "epoch": 0.8380893777401738,
+      "learning_rate": 0.00018990114811248283,
+      "loss": 1.2176,
+      "step": 10896
+    },
+    {
+      "epoch": 0.8381662949003923,
+      "learning_rate": 0.0001897246648229225,
+      "loss": 0.8967,
+      "step": 10897
+    },
+    {
+      "epoch": 0.8382432120606107,
+      "learning_rate": 0.00018954825804170483,
+      "loss": 1.0882,
+      "step": 10898
+    },
+    {
+      "epoch": 0.8383201292208292,
+      "learning_rate": 0.0001893719277791301,
+      "loss": 1.1616,
+      "step": 10899
+    },
+    {
+      "epoch": 0.8383970463810476,
+      "learning_rate": 0.00018919567404549438,
+      "loss": 1.2519,
+      "step": 10900
+    },
+    {
+      "epoch": 0.838473963541266,
+      "learning_rate": 0.00018901949685108945,
+      "loss": 1.2597,
+      "step": 10901
+    },
+    {
+      "epoch": 0.8385508807014845,
+      "learning_rate": 0.00018884339620620205,
+      "loss": 1.0918,
+      "step": 10902
+    },
+    {
+      "epoch": 0.8386277978617029,
+      "learning_rate": 0.0001886673721211157,
+      "loss": 1.1742,
+      "step": 10903
+    },
+    {
+      "epoch": 0.8387047150219213,
+      "learning_rate": 0.00018849142460610792,
+      "loss": 1.1534,
+      "step": 10904
+    },
+    {
+      "epoch": 0.8387816321821399,
+      "learning_rate": 0.00018831555367145237,
+      "loss": 1.1128,
+      "step": 10905
+    },
+    {
+      "epoch": 0.8388585493423583,
+      "learning_rate": 0.00018813975932741882,
+      "loss": 1.1676,
+      "step": 10906
+    },
+    {
+      "epoch": 0.8389354665025768,
+      "learning_rate": 0.0001879640415842721,
+      "loss": 0.8937,
+      "step": 10907
+    },
+    {
+      "epoch": 0.8390123836627952,
+      "learning_rate": 0.00018778840045227213,
+      "loss": 1.0896,
+      "step": 10908
+    },
+    {
+      "epoch": 0.8390893008230136,
+      "learning_rate": 0.0001876128359416752,
+      "loss": 1.0042,
+      "step": 10909
+    },
+    {
+      "epoch": 0.8391662179832321,
+      "learning_rate": 0.00018743734806273216,
+      "loss": 1.146,
+      "step": 10910
+    },
+    {
+      "epoch": 0.8392431351434505,
+      "learning_rate": 0.0001872619368256906,
+      "loss": 1.3506,
+      "step": 10911
+    },
+    {
+      "epoch": 0.8393200523036689,
+      "learning_rate": 0.00018708660224079298,
+      "loss": 1.035,
+      "step": 10912
+    },
+    {
+      "epoch": 0.8393969694638874,
+      "learning_rate": 0.00018691134431827634,
+      "loss": 0.9361,
+      "step": 10913
+    },
+    {
+      "epoch": 0.8394738866241058,
+      "learning_rate": 0.00018673616306837492,
+      "loss": 1.183,
+      "step": 10914
+    },
+    {
+      "epoch": 0.8395508037843242,
+      "learning_rate": 0.00018656105850131777,
+      "loss": 0.7885,
+      "step": 10915
+    },
+    {
+      "epoch": 0.8396277209445427,
+      "learning_rate": 0.00018638603062732918,
+      "loss": 1.3637,
+      "step": 10916
+    },
+    {
+      "epoch": 0.8397046381047611,
+      "learning_rate": 0.0001862110794566293,
+      "loss": 1.2342,
+      "step": 10917
+    },
+    {
+      "epoch": 0.8397815552649797,
+      "learning_rate": 0.0001860362049994333,
+      "loss": 1.1524,
+      "step": 10918
+    },
+    {
+      "epoch": 0.8398584724251981,
+      "learning_rate": 0.0001858614072659532,
+      "loss": 1.0019,
+      "step": 10919
+    },
+    {
+      "epoch": 0.8399353895854165,
+      "learning_rate": 0.0001856866862663948,
+      "loss": 1.1074,
+      "step": 10920
+    },
+    {
+      "epoch": 0.840012306745635,
+      "learning_rate": 0.00018551204201096077,
+      "loss": 1.1043,
+      "step": 10921
+    },
+    {
+      "epoch": 0.8400892239058534,
+      "learning_rate": 0.00018533747450984834,
+      "loss": 1.2372,
+      "step": 10922
+    },
+    {
+      "epoch": 0.8401661410660718,
+      "learning_rate": 0.00018516298377325096,
+      "loss": 0.9992,
+      "step": 10923
+    },
+    {
+      "epoch": 0.8402430582262903,
+      "learning_rate": 0.00018498856981135703,
+      "loss": 1.3434,
+      "step": 10924
+    },
+    {
+      "epoch": 0.8403199753865087,
+      "learning_rate": 0.00018481423263435153,
+      "loss": 1.2279,
+      "step": 10925
+    },
+    {
+      "epoch": 0.8403968925467272,
+      "learning_rate": 0.00018463997225241302,
+      "loss": 0.9389,
+      "step": 10926
+    },
+    {
+      "epoch": 0.8404738097069456,
+      "learning_rate": 0.00018446578867571767,
+      "loss": 1.5451,
+      "step": 10927
+    },
+    {
+      "epoch": 0.840550726867164,
+      "learning_rate": 0.00018429168191443586,
+      "loss": 0.9528,
+      "step": 10928
+    },
+    {
+      "epoch": 0.8406276440273825,
+      "learning_rate": 0.0001841176519787341,
+      "loss": 1.0809,
+      "step": 10929
+    },
+    {
+      "epoch": 0.840704561187601,
+      "learning_rate": 0.00018394369887877394,
+      "loss": 1.2432,
+      "step": 10930
+    },
+    {
+      "epoch": 0.8407814783478194,
+      "learning_rate": 0.00018376982262471276,
+      "loss": 1.2846,
+      "step": 10931
+    },
+    {
+      "epoch": 0.8408583955080379,
+      "learning_rate": 0.0001835960232267031,
+      "loss": 1.5822,
+      "step": 10932
+    },
+    {
+      "epoch": 0.8409353126682563,
+      "learning_rate": 0.00018342230069489385,
+      "loss": 1.1378,
+      "step": 10933
+    },
+    {
+      "epoch": 0.8410122298284747,
+      "learning_rate": 0.0001832486550394289,
+      "loss": 1.1332,
+      "step": 10934
+    },
+    {
+      "epoch": 0.8410891469886932,
+      "learning_rate": 0.00018307508627044662,
+      "loss": 1.1435,
+      "step": 10935
+    },
+    {
+      "epoch": 0.8411660641489116,
+      "learning_rate": 0.00018290159439808274,
+      "loss": 1.3661,
+      "step": 10936
+    },
+    {
+      "epoch": 0.8412429813091301,
+      "learning_rate": 0.00018272817943246718,
+      "loss": 0.9954,
+      "step": 10937
+    },
+    {
+      "epoch": 0.8413198984693485,
+      "learning_rate": 0.00018255484138372663,
+      "loss": 1.2505,
+      "step": 10938
+    },
+    {
+      "epoch": 0.8413968156295669,
+      "learning_rate": 0.0001823815802619817,
+      "loss": 1.0661,
+      "step": 10939
+    },
+    {
+      "epoch": 0.8414737327897854,
+      "learning_rate": 0.00018220839607734912,
+      "loss": 0.9761,
+      "step": 10940
+    },
+    {
+      "epoch": 0.8415506499500038,
+      "learning_rate": 0.00018203528883994197,
+      "loss": 1.3057,
+      "step": 10941
+    },
+    {
+      "epoch": 0.8416275671102222,
+      "learning_rate": 0.00018186225855986783,
+      "loss": 1.1566,
+      "step": 10942
+    },
+    {
+      "epoch": 0.8417044842704408,
+      "learning_rate": 0.00018168930524723026,
+      "loss": 1.2366,
+      "step": 10943
+    },
+    {
+      "epoch": 0.8417814014306592,
+      "learning_rate": 0.00018151642891212788,
+      "loss": 1.225,
+      "step": 10944
+    },
+    {
+      "epoch": 0.8418583185908777,
+      "learning_rate": 0.00018134362956465512,
+      "loss": 1.2227,
+      "step": 10945
+    },
+    {
+      "epoch": 0.8419352357510961,
+      "learning_rate": 0.00018117090721490238,
+      "loss": 0.9353,
+      "step": 10946
+    },
+    {
+      "epoch": 0.8420121529113145,
+      "learning_rate": 0.0001809982618729551,
+      "loss": 1.1659,
+      "step": 10947
+    },
+    {
+      "epoch": 0.842089070071533,
+      "learning_rate": 0.00018082569354889338,
+      "loss": 1.3555,
+      "step": 10948
+    },
+    {
+      "epoch": 0.8421659872317514,
+      "learning_rate": 0.00018065320225279453,
+      "loss": 0.8292,
+      "step": 10949
+    },
+    {
+      "epoch": 0.8422429043919698,
+      "learning_rate": 0.0001804807879947301,
+      "loss": 1.7349,
+      "step": 10950
+    },
+    {
+      "epoch": 0.8423198215521883,
+      "learning_rate": 0.00018030845078476776,
+      "loss": 1.0104,
+      "step": 10951
+    },
+    {
+      "epoch": 0.8423967387124067,
+      "learning_rate": 0.00018013619063297028,
+      "loss": 1.321,
+      "step": 10952
+    },
+    {
+      "epoch": 0.8424736558726252,
+      "learning_rate": 0.00017996400754939575,
+      "loss": 1.0762,
+      "step": 10953
+    },
+    {
+      "epoch": 0.8425505730328436,
+      "learning_rate": 0.00017979190154409898,
+      "loss": 0.8755,
+      "step": 10954
+    },
+    {
+      "epoch": 0.842627490193062,
+      "learning_rate": 0.00017961987262712892,
+      "loss": 1.2364,
+      "step": 10955
+    },
+    {
+      "epoch": 0.8427044073532806,
+      "learning_rate": 0.00017944792080853056,
+      "loss": 0.9473,
+      "step": 10956
+    },
+    {
+      "epoch": 0.842781324513499,
+      "learning_rate": 0.00017927604609834435,
+      "loss": 1.2731,
+      "step": 10957
+    },
+    {
+      "epoch": 0.8428582416737174,
+      "learning_rate": 0.0001791042485066061,
+      "loss": 0.9911,
+      "step": 10958
+    },
+    {
+      "epoch": 0.8429351588339359,
+      "learning_rate": 0.0001789325280433473,
+      "loss": 1.261,
+      "step": 10959
+    },
+    {
+      "epoch": 0.8430120759941543,
+      "learning_rate": 0.00017876088471859542,
+      "loss": 1.2491,
+      "step": 10960
+    },
+    {
+      "epoch": 0.8430889931543727,
+      "learning_rate": 0.00017858931854237227,
+      "loss": 1.2727,
+      "step": 10961
+    },
+    {
+      "epoch": 0.8431659103145912,
+      "learning_rate": 0.00017841782952469565,
+      "loss": 1.2604,
+      "step": 10962
+    },
+    {
+      "epoch": 0.8432428274748096,
+      "learning_rate": 0.0001782464176755794,
+      "loss": 1.2202,
+      "step": 10963
+    },
+    {
+      "epoch": 0.8433197446350281,
+      "learning_rate": 0.00017807508300503257,
+      "loss": 0.8169,
+      "step": 10964
+    },
+    {
+      "epoch": 0.8433966617952465,
+      "learning_rate": 0.00017790382552305907,
+      "loss": 1.2243,
+      "step": 10965
+    },
+    {
+      "epoch": 0.8434735789554649,
+      "learning_rate": 0.00017773264523965925,
+      "loss": 1.3926,
+      "step": 10966
+    },
+    {
+      "epoch": 0.8435504961156834,
+      "learning_rate": 0.00017756154216482783,
+      "loss": 1.0529,
+      "step": 10967
+    },
+    {
+      "epoch": 0.8436274132759018,
+      "learning_rate": 0.00017739051630855663,
+      "loss": 1.1103,
+      "step": 10968
+    },
+    {
+      "epoch": 0.8437043304361203,
+      "learning_rate": 0.0001772195676808318,
+      "loss": 1.0959,
+      "step": 10969
+    },
+    {
+      "epoch": 0.8437812475963388,
+      "learning_rate": 0.0001770486962916344,
+      "loss": 0.7226,
+      "step": 10970
+    },
+    {
+      "epoch": 0.8438581647565572,
+      "learning_rate": 0.00017687790215094274,
+      "loss": 0.8492,
+      "step": 10971
+    },
+    {
+      "epoch": 0.8439350819167757,
+      "learning_rate": 0.00017670718526872887,
+      "loss": 1.2645,
+      "step": 10972
+    },
+    {
+      "epoch": 0.8440119990769941,
+      "learning_rate": 0.0001765365456549623,
+      "loss": 1.0794,
+      "step": 10973
+    },
+    {
+      "epoch": 0.8440889162372125,
+      "learning_rate": 0.00017636598331960556,
+      "loss": 1.0627,
+      "step": 10974
+    },
+    {
+      "epoch": 0.844165833397431,
+      "learning_rate": 0.00017619549827261837,
+      "loss": 1.168,
+      "step": 10975
+    },
+    {
+      "epoch": 0.8442427505576494,
+      "learning_rate": 0.0001760250905239561,
+      "loss": 0.9741,
+      "step": 10976
+    },
+    {
+      "epoch": 0.8443196677178678,
+      "learning_rate": 0.00017585476008356843,
+      "loss": 1.2224,
+      "step": 10977
+    },
+    {
+      "epoch": 0.8443965848780863,
+      "learning_rate": 0.00017568450696140141,
+      "loss": 1.1392,
+      "step": 10978
+    },
+    {
+      "epoch": 0.8444735020383047,
+      "learning_rate": 0.00017551433116739612,
+      "loss": 1.1975,
+      "step": 10979
+    },
+    {
+      "epoch": 0.8445504191985231,
+      "learning_rate": 0.00017534423271148942,
+      "loss": 0.9705,
+      "step": 10980
+    },
+    {
+      "epoch": 0.8446273363587417,
+      "learning_rate": 0.00017517421160361319,
+      "loss": 0.9703,
+      "step": 10981
+    },
+    {
+      "epoch": 0.8447042535189601,
+      "learning_rate": 0.000175004267853696,
+      "loss": 1.2962,
+      "step": 10982
+    },
+    {
+      "epoch": 0.8447811706791786,
+      "learning_rate": 0.0001748344014716599,
+      "loss": 1.2797,
+      "step": 10983
+    },
+    {
+      "epoch": 0.844858087839397,
+      "learning_rate": 0.00017466461246742448,
+      "loss": 1.0698,
+      "step": 10984
+    },
+    {
+      "epoch": 0.8449350049996154,
+      "learning_rate": 0.00017449490085090364,
+      "loss": 1.005,
+      "step": 10985
+    },
+    {
+      "epoch": 0.8450119221598339,
+      "learning_rate": 0.00017432526663200694,
+      "loss": 0.8879,
+      "step": 10986
+    },
+    {
+      "epoch": 0.8450888393200523,
+      "learning_rate": 0.00017415570982063944,
+      "loss": 1.1531,
+      "step": 10987
+    },
+    {
+      "epoch": 0.8451657564802707,
+      "learning_rate": 0.00017398623042670175,
+      "loss": 1.3468,
+      "step": 10988
+    },
+    {
+      "epoch": 0.8452426736404892,
+      "learning_rate": 0.0001738168284600898,
+      "loss": 1.2146,
+      "step": 10989
+    },
+    {
+      "epoch": 0.8453195908007076,
+      "learning_rate": 0.00017364750393069562,
+      "loss": 1.4544,
+      "step": 10990
+    },
+    {
+      "epoch": 0.8453965079609261,
+      "learning_rate": 0.000173478256848406,
+      "loss": 0.9959,
+      "step": 10991
+    },
+    {
+      "epoch": 0.8454734251211445,
+      "learning_rate": 0.0001733090872231034,
+      "loss": 1.2112,
+      "step": 10992
+    },
+    {
+      "epoch": 0.8455503422813629,
+      "learning_rate": 0.0001731399950646657,
+      "loss": 0.7243,
+      "step": 10993
+    },
+    {
+      "epoch": 0.8456272594415815,
+      "learning_rate": 0.00017297098038296637,
+      "loss": 1.3572,
+      "step": 10994
+    },
+    {
+      "epoch": 0.8457041766017999,
+      "learning_rate": 0.00017280204318787518,
+      "loss": 1.1198,
+      "step": 10995
+    },
+    {
+      "epoch": 0.8457810937620183,
+      "learning_rate": 0.0001726331834892554,
+      "loss": 1.2802,
+      "step": 10996
+    },
+    {
+      "epoch": 0.8458580109222368,
+      "learning_rate": 0.0001724644012969671,
+      "loss": 0.6415,
+      "step": 10997
+    },
+    {
+      "epoch": 0.8459349280824552,
+      "learning_rate": 0.00017229569662086631,
+      "loss": 1.3411,
+      "step": 10998
+    },
+    {
+      "epoch": 0.8460118452426736,
+      "learning_rate": 0.00017212706947080343,
+      "loss": 1.438,
+      "step": 10999
+    },
+    {
+      "epoch": 0.8460887624028921,
+      "learning_rate": 0.00017195851985662492,
+      "loss": 1.42,
+      "step": 11000
+    },
+    {
+      "epoch": 0.8461656795631105,
+      "learning_rate": 0.00017179004778817243,
+      "loss": 1.189,
+      "step": 11001
+    },
+    {
+      "epoch": 0.846242596723329,
+      "learning_rate": 0.00017162165327528307,
+      "loss": 1.1512,
+      "step": 11002
+    },
+    {
+      "epoch": 0.8463195138835474,
+      "learning_rate": 0.00017145333632778999,
+      "loss": 1.4032,
+      "step": 11003
+    },
+    {
+      "epoch": 0.8463964310437658,
+      "learning_rate": 0.00017128509695552114,
+      "loss": 1.7623,
+      "step": 11004
+    },
+    {
+      "epoch": 0.8464733482039843,
+      "learning_rate": 0.00017111693516830019,
+      "loss": 0.8291,
+      "step": 11005
+    },
+    {
+      "epoch": 0.8465502653642027,
+      "learning_rate": 0.00017094885097594642,
+      "loss": 1.1812,
+      "step": 11006
+    },
+    {
+      "epoch": 0.8466271825244212,
+      "learning_rate": 0.00017078084438827435,
+      "loss": 1.0123,
+      "step": 11007
+    },
+    {
+      "epoch": 0.8467040996846397,
+      "learning_rate": 0.0001706129154150941,
+      "loss": 1.2354,
+      "step": 11008
+    },
+    {
+      "epoch": 0.8467810168448581,
+      "learning_rate": 0.00017044506406621102,
+      "loss": 1.1638,
+      "step": 11009
+    },
+    {
+      "epoch": 0.8468579340050766,
+      "learning_rate": 0.00017027729035142607,
+      "loss": 1.0892,
+      "step": 11010
+    },
+    {
+      "epoch": 0.846934851165295,
+      "learning_rate": 0.0001701095942805363,
+      "loss": 1.315,
+      "step": 11011
+    },
+    {
+      "epoch": 0.8470117683255134,
+      "learning_rate": 0.00016994197586333315,
+      "loss": 1.2207,
+      "step": 11012
+    },
+    {
+      "epoch": 0.8470886854857319,
+      "learning_rate": 0.00016977443510960433,
+      "loss": 1.2968,
+      "step": 11013
+    },
+    {
+      "epoch": 0.8471656026459503,
+      "learning_rate": 0.0001696069720291325,
+      "loss": 0.9474,
+      "step": 11014
+    },
+    {
+      "epoch": 0.8472425198061687,
+      "learning_rate": 0.000169439586631696,
+      "loss": 1.1572,
+      "step": 11015
+    },
+    {
+      "epoch": 0.8473194369663872,
+      "learning_rate": 0.00016927227892706852,
+      "loss": 1.0062,
+      "step": 11016
+    },
+    {
+      "epoch": 0.8473963541266056,
+      "learning_rate": 0.00016910504892501992,
+      "loss": 1.2504,
+      "step": 11017
+    },
+    {
+      "epoch": 0.847473271286824,
+      "learning_rate": 0.00016893789663531471,
+      "loss": 1.1055,
+      "step": 11018
+    },
+    {
+      "epoch": 0.8475501884470426,
+      "learning_rate": 0.00016877082206771227,
+      "loss": 1.0222,
+      "step": 11019
+    },
+    {
+      "epoch": 0.847627105607261,
+      "learning_rate": 0.00016860382523196926,
+      "loss": 1.2512,
+      "step": 11020
+    },
+    {
+      "epoch": 0.8477040227674795,
+      "learning_rate": 0.00016843690613783642,
+      "loss": 0.9854,
+      "step": 11021
+    },
+    {
+      "epoch": 0.8477809399276979,
+      "learning_rate": 0.00016827006479506042,
+      "loss": 1.1565,
+      "step": 11022
+    },
+    {
+      "epoch": 0.8478578570879163,
+      "learning_rate": 0.00016810330121338313,
+      "loss": 1.1712,
+      "step": 11023
+    },
+    {
+      "epoch": 0.8479347742481348,
+      "learning_rate": 0.00016793661540254195,
+      "loss": 1.1295,
+      "step": 11024
+    },
+    {
+      "epoch": 0.8480116914083532,
+      "learning_rate": 0.00016777000737227028,
+      "loss": 0.8593,
+      "step": 11025
+    },
+    {
+      "epoch": 0.8480886085685716,
+      "learning_rate": 0.0001676034771322963,
+      "loss": 1.0704,
+      "step": 11026
+    },
+    {
+      "epoch": 0.8481655257287901,
+      "learning_rate": 0.0001674370246923439,
+      "loss": 1.2174,
+      "step": 11027
+    },
+    {
+      "epoch": 0.8482424428890085,
+      "learning_rate": 0.0001672706500621325,
+      "loss": 1.0962,
+      "step": 11028
+    },
+    {
+      "epoch": 0.848319360049227,
+      "learning_rate": 0.0001671043532513763,
+      "loss": 0.9685,
+      "step": 11029
+    },
+    {
+      "epoch": 0.8483962772094454,
+      "learning_rate": 0.0001669381342697864,
+      "loss": 1.2179,
+      "step": 11030
+    },
+    {
+      "epoch": 0.8484731943696638,
+      "learning_rate": 0.00016677199312706848,
+      "loss": 1.2541,
+      "step": 11031
+    },
+    {
+      "epoch": 0.8485501115298824,
+      "learning_rate": 0.00016660592983292266,
+      "loss": 1.2204,
+      "step": 11032
+    },
+    {
+      "epoch": 0.8486270286901008,
+      "learning_rate": 0.0001664399443970465,
+      "loss": 1.0257,
+      "step": 11033
+    },
+    {
+      "epoch": 0.8487039458503192,
+      "learning_rate": 0.00016627403682913179,
+      "loss": 1.3559,
+      "step": 11034
+    },
+    {
+      "epoch": 0.8487808630105377,
+      "learning_rate": 0.00016610820713886605,
+      "loss": 0.806,
+      "step": 11035
+    },
+    {
+      "epoch": 0.8488577801707561,
+      "learning_rate": 0.00016594245533593238,
+      "loss": 0.9572,
+      "step": 11036
+    },
+    {
+      "epoch": 0.8489346973309745,
+      "learning_rate": 0.0001657767814300089,
+      "loss": 0.7416,
+      "step": 11037
+    },
+    {
+      "epoch": 0.849011614491193,
+      "learning_rate": 0.00016561118543076937,
+      "loss": 1.2735,
+      "step": 11038
+    },
+    {
+      "epoch": 0.8490885316514114,
+      "learning_rate": 0.00016544566734788386,
+      "loss": 0.7056,
+      "step": 11039
+    },
+    {
+      "epoch": 0.8491654488116299,
+      "learning_rate": 0.00016528022719101648,
+      "loss": 1.0106,
+      "step": 11040
+    },
+    {
+      "epoch": 0.8492423659718483,
+      "learning_rate": 0.0001651148649698277,
+      "loss": 0.8831,
+      "step": 11041
+    },
+    {
+      "epoch": 0.8493192831320667,
+      "learning_rate": 0.00016494958069397309,
+      "loss": 1.3064,
+      "step": 11042
+    },
+    {
+      "epoch": 0.8493962002922852,
+      "learning_rate": 0.00016478437437310363,
+      "loss": 1.1316,
+      "step": 11043
+    },
+    {
+      "epoch": 0.8494731174525036,
+      "learning_rate": 0.00016461924601686656,
+      "loss": 1.179,
+      "step": 11044
+    },
+    {
+      "epoch": 0.849550034612722,
+      "learning_rate": 0.0001644541956349032,
+      "loss": 1.3716,
+      "step": 11045
+    },
+    {
+      "epoch": 0.8496269517729406,
+      "learning_rate": 0.00016428922323685101,
+      "loss": 0.8783,
+      "step": 11046
+    },
+    {
+      "epoch": 0.849703868933159,
+      "learning_rate": 0.0001641243288323433,
+      "loss": 1.3848,
+      "step": 11047
+    },
+    {
+      "epoch": 0.8497807860933775,
+      "learning_rate": 0.0001639595124310083,
+      "loss": 0.7817,
+      "step": 11048
+    },
+    {
+      "epoch": 0.8498577032535959,
+      "learning_rate": 0.00016379477404246973,
+      "loss": 0.8854,
+      "step": 11049
+    },
+    {
+      "epoch": 0.8499346204138143,
+      "learning_rate": 0.00016363011367634685,
+      "loss": 1.3142,
+      "step": 11050
+    },
+    {
+      "epoch": 0.8500115375740328,
+      "learning_rate": 0.00016346553134225395,
+      "loss": 1.1499,
+      "step": 11051
+    },
+    {
+      "epoch": 0.8500884547342512,
+      "learning_rate": 0.00016330102704980204,
+      "loss": 0.986,
+      "step": 11052
+    },
+    {
+      "epoch": 0.8501653718944696,
+      "learning_rate": 0.00016313660080859643,
+      "loss": 0.9656,
+      "step": 11053
+    },
+    {
+      "epoch": 0.8502422890546881,
+      "learning_rate": 0.00016297225262823745,
+      "loss": 1.2177,
+      "step": 11054
+    },
+    {
+      "epoch": 0.8503192062149065,
+      "learning_rate": 0.0001628079825183224,
+      "loss": 0.8681,
+      "step": 11055
+    },
+    {
+      "epoch": 0.8503961233751249,
+      "learning_rate": 0.00016264379048844263,
+      "loss": 1.1863,
+      "step": 11056
+    },
+    {
+      "epoch": 0.8504730405353434,
+      "learning_rate": 0.00016247967654818595,
+      "loss": 1.0493,
+      "step": 11057
+    },
+    {
+      "epoch": 0.8505499576955619,
+      "learning_rate": 0.00016231564070713473,
+      "loss": 1.2471,
+      "step": 11058
+    },
+    {
+      "epoch": 0.8506268748557804,
+      "learning_rate": 0.0001621516829748671,
+      "loss": 1.0571,
+      "step": 11059
+    },
+    {
+      "epoch": 0.8507037920159988,
+      "learning_rate": 0.00016198780336095726,
+      "loss": 1.2245,
+      "step": 11060
+    },
+    {
+      "epoch": 0.8507807091762172,
+      "learning_rate": 0.00016182400187497386,
+      "loss": 1.179,
+      "step": 11061
+    },
+    {
+      "epoch": 0.8508576263364357,
+      "learning_rate": 0.0001616602785264818,
+      "loss": 1.7001,
+      "step": 11062
+    },
+    {
+      "epoch": 0.8509345434966541,
+      "learning_rate": 0.0001614966333250409,
+      "loss": 0.9951,
+      "step": 11063
+    },
+    {
+      "epoch": 0.8510114606568725,
+      "learning_rate": 0.00016133306628020633,
+      "loss": 1.1903,
+      "step": 11064
+    },
+    {
+      "epoch": 0.851088377817091,
+      "learning_rate": 0.00016116957740152898,
+      "loss": 1.2345,
+      "step": 11065
+    },
+    {
+      "epoch": 0.8511652949773094,
+      "learning_rate": 0.0001610061666985559,
+      "loss": 1.3622,
+      "step": 11066
+    },
+    {
+      "epoch": 0.8512422121375279,
+      "learning_rate": 0.00016084283418082757,
+      "loss": 0.9611,
+      "step": 11067
+    },
+    {
+      "epoch": 0.8513191292977463,
+      "learning_rate": 0.00016067957985788222,
+      "loss": 1.3876,
+      "step": 11068
+    },
+    {
+      "epoch": 0.8513960464579647,
+      "learning_rate": 0.00016051640373925192,
+      "loss": 1.1933,
+      "step": 11069
+    },
+    {
+      "epoch": 0.8514729636181833,
+      "learning_rate": 0.00016035330583446467,
+      "loss": 0.8288,
+      "step": 11070
+    },
+    {
+      "epoch": 0.8515498807784017,
+      "learning_rate": 0.0001601902861530442,
+      "loss": 1.235,
+      "step": 11071
+    },
+    {
+      "epoch": 0.8516267979386201,
+      "learning_rate": 0.00016002734470450908,
+      "loss": 0.9444,
+      "step": 11072
+    },
+    {
+      "epoch": 0.8517037150988386,
+      "learning_rate": 0.00015986448149837368,
+      "loss": 1.292,
+      "step": 11073
+    },
+    {
+      "epoch": 0.851780632259057,
+      "learning_rate": 0.00015970169654414807,
+      "loss": 1.448,
+      "step": 11074
+    },
+    {
+      "epoch": 0.8518575494192755,
+      "learning_rate": 0.00015953898985133713,
+      "loss": 1.6343,
+      "step": 11075
+    },
+    {
+      "epoch": 0.8519344665794939,
+      "learning_rate": 0.0001593763614294416,
+      "loss": 0.9717,
+      "step": 11076
+    },
+    {
+      "epoch": 0.8520113837397123,
+      "learning_rate": 0.00015921381128795759,
+      "loss": 1.1789,
+      "step": 11077
+    },
+    {
+      "epoch": 0.8520883008999308,
+      "learning_rate": 0.00015905133943637612,
+      "loss": 1.2542,
+      "step": 11078
+    },
+    {
+      "epoch": 0.8521652180601492,
+      "learning_rate": 0.00015888894588418512,
+      "loss": 1.2433,
+      "step": 11079
+    },
+    {
+      "epoch": 0.8522421352203676,
+      "learning_rate": 0.00015872663064086585,
+      "loss": 1.2715,
+      "step": 11080
+    },
+    {
+      "epoch": 0.8523190523805861,
+      "learning_rate": 0.0001585643937158962,
+      "loss": 1.5103,
+      "step": 11081
+    },
+    {
+      "epoch": 0.8523959695408045,
+      "learning_rate": 0.00015840223511874997,
+      "loss": 1.3577,
+      "step": 11082
+    },
+    {
+      "epoch": 0.852472886701023,
+      "learning_rate": 0.00015824015485889521,
+      "loss": 1.3692,
+      "step": 11083
+    },
+    {
+      "epoch": 0.8525498038612415,
+      "learning_rate": 0.00015807815294579635,
+      "loss": 1.246,
+      "step": 11084
+    },
+    {
+      "epoch": 0.8526267210214599,
+      "learning_rate": 0.00015791622938891254,
+      "loss": 1.1344,
+      "step": 11085
+    },
+    {
+      "epoch": 0.8527036381816784,
+      "learning_rate": 0.0001577543841976985,
+      "loss": 1.0745,
+      "step": 11086
+    },
+    {
+      "epoch": 0.8527805553418968,
+      "learning_rate": 0.0001575926173816052,
+      "loss": 1.0883,
+      "step": 11087
+    },
+    {
+      "epoch": 0.8528574725021152,
+      "learning_rate": 0.0001574309289500782,
+      "loss": 0.9141,
+      "step": 11088
+    },
+    {
+      "epoch": 0.8529343896623337,
+      "learning_rate": 0.000157269318912558,
+      "loss": 1.255,
+      "step": 11089
+    },
+    {
+      "epoch": 0.8530113068225521,
+      "learning_rate": 0.00015710778727848174,
+      "loss": 1.2988,
+      "step": 11090
+    },
+    {
+      "epoch": 0.8530882239827705,
+      "learning_rate": 0.00015694633405728147,
+      "loss": 1.0386,
+      "step": 11091
+    },
+    {
+      "epoch": 0.853165141142989,
+      "learning_rate": 0.00015678495925838438,
+      "loss": 1.1427,
+      "step": 11092
+    },
+    {
+      "epoch": 0.8532420583032074,
+      "learning_rate": 0.00015662366289121338,
+      "loss": 1.1574,
+      "step": 11093
+    },
+    {
+      "epoch": 0.8533189754634259,
+      "learning_rate": 0.00015646244496518648,
+      "loss": 1.198,
+      "step": 11094
+    },
+    {
+      "epoch": 0.8533958926236443,
+      "learning_rate": 0.00015630130548971777,
+      "loss": 1.2668,
+      "step": 11095
+    },
+    {
+      "epoch": 0.8534728097838628,
+      "learning_rate": 0.00015614024447421627,
+      "loss": 1.2075,
+      "step": 11096
+    },
+    {
+      "epoch": 0.8535497269440813,
+      "learning_rate": 0.00015597926192808658,
+      "loss": 1.3358,
+      "step": 11097
+    },
+    {
+      "epoch": 0.8536266441042997,
+      "learning_rate": 0.0001558183578607284,
+      "loss": 1.3366,
+      "step": 11098
+    },
+    {
+      "epoch": 0.8537035612645181,
+      "learning_rate": 0.00015565753228153717,
+      "loss": 1.1686,
+      "step": 11099
+    },
+    {
+      "epoch": 0.8537804784247366,
+      "learning_rate": 0.00015549678519990323,
+      "loss": 1.1919,
+      "step": 11100
+    },
+    {
+      "epoch": 0.853857395584955,
+      "learning_rate": 0.00015533611662521408,
+      "loss": 0.9732,
+      "step": 11101
+    },
+    {
+      "epoch": 0.8539343127451734,
+      "learning_rate": 0.00015517552656684985,
+      "loss": 1.2336,
+      "step": 11102
+    },
+    {
+      "epoch": 0.8540112299053919,
+      "learning_rate": 0.00015501501503418808,
+      "loss": 1.2513,
+      "step": 11103
+    },
+    {
+      "epoch": 0.8540881470656103,
+      "learning_rate": 0.00015485458203660156,
+      "loss": 1.1273,
+      "step": 11104
+    },
+    {
+      "epoch": 0.8541650642258288,
+      "learning_rate": 0.00015469422758345786,
+      "loss": 1.333,
+      "step": 11105
+    },
+    {
+      "epoch": 0.8542419813860472,
+      "learning_rate": 0.00015453395168412026,
+      "loss": 0.917,
+      "step": 11106
+    },
+    {
+      "epoch": 0.8543188985462656,
+      "learning_rate": 0.0001543737543479475,
+      "loss": 1.2608,
+      "step": 11107
+    },
+    {
+      "epoch": 0.8543958157064842,
+      "learning_rate": 0.0001542136355842932,
+      "loss": 1.4238,
+      "step": 11108
+    },
+    {
+      "epoch": 0.8544727328667026,
+      "learning_rate": 0.00015405359540250757,
+      "loss": 0.9659,
+      "step": 11109
+    },
+    {
+      "epoch": 0.854549650026921,
+      "learning_rate": 0.0001538936338119355,
+      "loss": 1.1839,
+      "step": 11110
+    },
+    {
+      "epoch": 0.8546265671871395,
+      "learning_rate": 0.0001537337508219165,
+      "loss": 1.2577,
+      "step": 11111
+    },
+    {
+      "epoch": 0.8547034843473579,
+      "learning_rate": 0.00015357394644178706,
+      "loss": 0.7599,
+      "step": 11112
+    },
+    {
+      "epoch": 0.8547804015075764,
+      "learning_rate": 0.00015341422068087778,
+      "loss": 1.1779,
+      "step": 11113
+    },
+    {
+      "epoch": 0.8548573186677948,
+      "learning_rate": 0.00015325457354851618,
+      "loss": 1.4574,
+      "step": 11114
+    },
+    {
+      "epoch": 0.8549342358280132,
+      "learning_rate": 0.0001530950050540233,
+      "loss": 0.805,
+      "step": 11115
+    },
+    {
+      "epoch": 0.8550111529882317,
+      "learning_rate": 0.0001529355152067165,
+      "loss": 0.7373,
+      "step": 11116
+    },
+    {
+      "epoch": 0.8550880701484501,
+      "learning_rate": 0.0001527761040159092,
+      "loss": 1.3628,
+      "step": 11117
+    },
+    {
+      "epoch": 0.8551649873086685,
+      "learning_rate": 0.00015261677149090908,
+      "loss": 1.4228,
+      "step": 11118
+    },
+    {
+      "epoch": 0.855241904468887,
+      "learning_rate": 0.00015245751764101989,
+      "loss": 1.4497,
+      "step": 11119
+    },
+    {
+      "epoch": 0.8553188216291054,
+      "learning_rate": 0.0001522983424755407,
+      "loss": 1.2204,
+      "step": 11120
+    },
+    {
+      "epoch": 0.8553957387893238,
+      "learning_rate": 0.0001521392460037659,
+      "loss": 1.2439,
+      "step": 11121
+    },
+    {
+      "epoch": 0.8554726559495424,
+      "learning_rate": 0.00015198022823498486,
+      "loss": 1.0578,
+      "step": 11122
+    },
+    {
+      "epoch": 0.8555495731097608,
+      "learning_rate": 0.0001518212891784837,
+      "loss": 1.174,
+      "step": 11123
+    },
+    {
+      "epoch": 0.8556264902699793,
+      "learning_rate": 0.00015166242884354198,
+      "loss": 1.1494,
+      "step": 11124
+    },
+    {
+      "epoch": 0.8557034074301977,
+      "learning_rate": 0.00015150364723943643,
+      "loss": 1.0044,
+      "step": 11125
+    },
+    {
+      "epoch": 0.8557803245904161,
+      "learning_rate": 0.00015134494437543834,
+      "loss": 1.2293,
+      "step": 11126
+    },
+    {
+      "epoch": 0.8558572417506346,
+      "learning_rate": 0.00015118632026081457,
+      "loss": 1.3714,
+      "step": 11127
+    },
+    {
+      "epoch": 0.855934158910853,
+      "learning_rate": 0.00015102777490482727,
+      "loss": 1.324,
+      "step": 11128
+    },
+    {
+      "epoch": 0.8560110760710714,
+      "learning_rate": 0.00015086930831673402,
+      "loss": 1.3615,
+      "step": 11129
+    },
+    {
+      "epoch": 0.8560879932312899,
+      "learning_rate": 0.00015071092050578756,
+      "loss": 1.0112,
+      "step": 11130
+    },
+    {
+      "epoch": 0.8561649103915083,
+      "learning_rate": 0.00015055261148123682,
+      "loss": 1.5117,
+      "step": 11131
+    },
+    {
+      "epoch": 0.8562418275517268,
+      "learning_rate": 0.00015039438125232562,
+      "loss": 0.9133,
+      "step": 11132
+    },
+    {
+      "epoch": 0.8563187447119452,
+      "learning_rate": 0.00015023622982829305,
+      "loss": 0.9773,
+      "step": 11133
+    },
+    {
+      "epoch": 0.8563956618721636,
+      "learning_rate": 0.00015007815721837354,
+      "loss": 1.2601,
+      "step": 11134
+    },
+    {
+      "epoch": 0.8564725790323822,
+      "learning_rate": 0.0001499201634317972,
+      "loss": 1.1649,
+      "step": 11135
+    },
+    {
+      "epoch": 0.8565494961926006,
+      "learning_rate": 0.00014976224847779003,
+      "loss": 1.318,
+      "step": 11136
+    },
+    {
+      "epoch": 0.856626413352819,
+      "learning_rate": 0.00014960441236557193,
+      "loss": 1.1077,
+      "step": 11137
+    },
+    {
+      "epoch": 0.8567033305130375,
+      "learning_rate": 0.0001494466551043594,
+      "loss": 1.1102,
+      "step": 11138
+    },
+    {
+      "epoch": 0.8567802476732559,
+      "learning_rate": 0.00014928897670336456,
+      "loss": 1.279,
+      "step": 11139
+    },
+    {
+      "epoch": 0.8568571648334743,
+      "learning_rate": 0.00014913137717179387,
+      "loss": 1.0541,
+      "step": 11140
+    },
+    {
+      "epoch": 0.8569340819936928,
+      "learning_rate": 0.00014897385651884998,
+      "loss": 1.0891,
+      "step": 11141
+    },
+    {
+      "epoch": 0.8570109991539112,
+      "learning_rate": 0.00014881641475373054,
+      "loss": 1.2631,
+      "step": 11142
+    },
+    {
+      "epoch": 0.8570879163141297,
+      "learning_rate": 0.00014865905188562867,
+      "loss": 1.088,
+      "step": 11143
+    },
+    {
+      "epoch": 0.8571648334743481,
+      "learning_rate": 0.00014850176792373338,
+      "loss": 1.1013,
+      "step": 11144
+    },
+    {
+      "epoch": 0.8572417506345665,
+      "learning_rate": 0.00014834456287722848,
+      "loss": 1.132,
+      "step": 11145
+    },
+    {
+      "epoch": 0.857318667794785,
+      "learning_rate": 0.00014818743675529278,
+      "loss": 1.3541,
+      "step": 11146
+    },
+    {
+      "epoch": 0.8573955849550035,
+      "learning_rate": 0.00014803038956710162,
+      "loss": 1.2979,
+      "step": 11147
+    },
+    {
+      "epoch": 0.8574725021152219,
+      "learning_rate": 0.0001478734213218252,
+      "loss": 0.6047,
+      "step": 11148
+    },
+    {
+      "epoch": 0.8575494192754404,
+      "learning_rate": 0.00014771653202862877,
+      "loss": 1.199,
+      "step": 11149
+    },
+    {
+      "epoch": 0.8576263364356588,
+      "learning_rate": 0.00014755972169667325,
+      "loss": 1.1863,
+      "step": 11150
+    },
+    {
+      "epoch": 0.8577032535958773,
+      "learning_rate": 0.00014740299033511495,
+      "loss": 1.0097,
+      "step": 11151
+    },
+    {
+      "epoch": 0.8577801707560957,
+      "learning_rate": 0.00014724633795310587,
+      "loss": 1.2456,
+      "step": 11152
+    },
+    {
+      "epoch": 0.8578570879163141,
+      "learning_rate": 0.0001470897645597929,
+      "loss": 0.9478,
+      "step": 11153
+    },
+    {
+      "epoch": 0.8579340050765326,
+      "learning_rate": 0.00014693327016431868,
+      "loss": 1.3032,
+      "step": 11154
+    },
+    {
+      "epoch": 0.858010922236751,
+      "learning_rate": 0.00014677685477582075,
+      "loss": 1.2877,
+      "step": 11155
+    },
+    {
+      "epoch": 0.8580878393969694,
+      "learning_rate": 0.00014662051840343265,
+      "loss": 1.2861,
+      "step": 11156
+    },
+    {
+      "epoch": 0.8581647565571879,
+      "learning_rate": 0.00014646426105628275,
+      "loss": 1.3966,
+      "step": 11157
+    },
+    {
+      "epoch": 0.8582416737174063,
+      "learning_rate": 0.00014630808274349573,
+      "loss": 1.1975,
+      "step": 11158
+    },
+    {
+      "epoch": 0.8583185908776247,
+      "learning_rate": 0.0001461519834741903,
+      "loss": 1.2797,
+      "step": 11159
+    },
+    {
+      "epoch": 0.8583955080378433,
+      "learning_rate": 0.00014599596325748116,
+      "loss": 0.9818,
+      "step": 11160
+    },
+    {
+      "epoch": 0.8584724251980617,
+      "learning_rate": 0.00014584002210247922,
+      "loss": 1.2435,
+      "step": 11161
+    },
+    {
+      "epoch": 0.8585493423582802,
+      "learning_rate": 0.00014568416001828964,
+      "loss": 1.1772,
+      "step": 11162
+    },
+    {
+      "epoch": 0.8586262595184986,
+      "learning_rate": 0.0001455283770140134,
+      "loss": 0.9049,
+      "step": 11163
+    },
+    {
+      "epoch": 0.858703176678717,
+      "learning_rate": 0.00014537267309874664,
+      "loss": 1.3119,
+      "step": 11164
+    },
+    {
+      "epoch": 0.8587800938389355,
+      "learning_rate": 0.0001452170482815811,
+      "loss": 0.9825,
+      "step": 11165
+    },
+    {
+      "epoch": 0.8588570109991539,
+      "learning_rate": 0.00014506150257160438,
+      "loss": 1.3915,
+      "step": 11166
+    },
+    {
+      "epoch": 0.8589339281593723,
+      "learning_rate": 0.00014490603597789854,
+      "loss": 1.8288,
+      "step": 11167
+    },
+    {
+      "epoch": 0.8590108453195908,
+      "learning_rate": 0.0001447506485095416,
+      "loss": 1.139,
+      "step": 11168
+    },
+    {
+      "epoch": 0.8590877624798092,
+      "learning_rate": 0.00014459534017560654,
+      "loss": 1.1336,
+      "step": 11169
+    },
+    {
+      "epoch": 0.8591646796400277,
+      "learning_rate": 0.00014444011098516192,
+      "loss": 1.4344,
+      "step": 11170
+    },
+    {
+      "epoch": 0.8592415968002461,
+      "learning_rate": 0.00014428496094727228,
+      "loss": 1.0494,
+      "step": 11171
+    },
+    {
+      "epoch": 0.8593185139604645,
+      "learning_rate": 0.00014412989007099692,
+      "loss": 1.2346,
+      "step": 11172
+    },
+    {
+      "epoch": 0.8593954311206831,
+      "learning_rate": 0.00014397489836538973,
+      "loss": 1.1324,
+      "step": 11173
+    },
+    {
+      "epoch": 0.8594723482809015,
+      "learning_rate": 0.00014381998583950186,
+      "loss": 1.3367,
+      "step": 11174
+    },
+    {
+      "epoch": 0.8595492654411199,
+      "learning_rate": 0.00014366515250237817,
+      "loss": 1.2786,
+      "step": 11175
+    },
+    {
+      "epoch": 0.8596261826013384,
+      "learning_rate": 0.00014351039836305995,
+      "loss": 0.9845,
+      "step": 11176
+    },
+    {
+      "epoch": 0.8597030997615568,
+      "learning_rate": 0.00014335572343058316,
+      "loss": 1.1422,
+      "step": 11177
+    },
+    {
+      "epoch": 0.8597800169217753,
+      "learning_rate": 0.00014320112771397926,
+      "loss": 1.3816,
+      "step": 11178
+    },
+    {
+      "epoch": 0.8598569340819937,
+      "learning_rate": 0.00014304661122227597,
+      "loss": 1.5254,
+      "step": 11179
+    },
+    {
+      "epoch": 0.8599338512422121,
+      "learning_rate": 0.0001428921739644951,
+      "loss": 1.329,
+      "step": 11180
+    },
+    {
+      "epoch": 0.8600107684024306,
+      "learning_rate": 0.00014273781594965445,
+      "loss": 0.9533,
+      "step": 11181
+    },
+    {
+      "epoch": 0.860087685562649,
+      "learning_rate": 0.00014258353718676748,
+      "loss": 0.9062,
+      "step": 11182
+    },
+    {
+      "epoch": 0.8601646027228674,
+      "learning_rate": 0.0001424293376848423,
+      "loss": 0.987,
+      "step": 11183
+    },
+    {
+      "epoch": 0.860241519883086,
+      "learning_rate": 0.00014227521745288253,
+      "loss": 1.0842,
+      "step": 11184
+    },
+    {
+      "epoch": 0.8603184370433044,
+      "learning_rate": 0.00014212117649988848,
+      "loss": 1.2322,
+      "step": 11185
+    },
+    {
+      "epoch": 0.8603953542035228,
+      "learning_rate": 0.0001419672148348538,
+      "loss": 0.941,
+      "step": 11186
+    },
+    {
+      "epoch": 0.8604722713637413,
+      "learning_rate": 0.00014181333246676846,
+      "loss": 1.1537,
+      "step": 11187
+    },
+    {
+      "epoch": 0.8605491885239597,
+      "learning_rate": 0.0001416595294046183,
+      "loss": 1.3088,
+      "step": 11188
+    },
+    {
+      "epoch": 0.8606261056841782,
+      "learning_rate": 0.0001415058056573839,
+      "loss": 1.2471,
+      "step": 11189
+    },
+    {
+      "epoch": 0.8607030228443966,
+      "learning_rate": 0.0001413521612340415,
+      "loss": 1.3289,
+      "step": 11190
+    },
+    {
+      "epoch": 0.860779940004615,
+      "learning_rate": 0.00014119859614356218,
+      "loss": 1.1891,
+      "step": 11191
+    },
+    {
+      "epoch": 0.8608568571648335,
+      "learning_rate": 0.00014104511039491263,
+      "loss": 1.2171,
+      "step": 11192
+    },
+    {
+      "epoch": 0.8609337743250519,
+      "learning_rate": 0.00014089170399705553,
+      "loss": 0.9272,
+      "step": 11193
+    },
+    {
+      "epoch": 0.8610106914852703,
+      "learning_rate": 0.0001407383769589487,
+      "loss": 1.0157,
+      "step": 11194
+    },
+    {
+      "epoch": 0.8610876086454888,
+      "learning_rate": 0.00014058512928954414,
+      "loss": 1.2028,
+      "step": 11195
+    },
+    {
+      "epoch": 0.8611645258057072,
+      "learning_rate": 0.0001404319609977907,
+      "loss": 1.1606,
+      "step": 11196
+    },
+    {
+      "epoch": 0.8612414429659258,
+      "learning_rate": 0.00014027887209263192,
+      "loss": 0.9443,
+      "step": 11197
+    },
+    {
+      "epoch": 0.8613183601261442,
+      "learning_rate": 0.00014012586258300725,
+      "loss": 0.9861,
+      "step": 11198
+    },
+    {
+      "epoch": 0.8613952772863626,
+      "learning_rate": 0.00013997293247785042,
+      "loss": 0.9838,
+      "step": 11199
+    },
+    {
+      "epoch": 0.8614721944465811,
+      "learning_rate": 0.0001398200817860913,
+      "loss": 1.4646,
+      "step": 11200
+    },
+    {
+      "epoch": 0.8615491116067995,
+      "learning_rate": 0.0001396673105166552,
+      "loss": 0.9187,
+      "step": 11201
+    },
+    {
+      "epoch": 0.8616260287670179,
+      "learning_rate": 0.00013951461867846271,
+      "loss": 1.0571,
+      "step": 11202
+    },
+    {
+      "epoch": 0.8617029459272364,
+      "learning_rate": 0.0001393620062804295,
+      "loss": 0.979,
+      "step": 11203
+    },
+    {
+      "epoch": 0.8617798630874548,
+      "learning_rate": 0.00013920947333146677,
+      "loss": 1.1068,
+      "step": 11204
+    },
+    {
+      "epoch": 0.8618567802476732,
+      "learning_rate": 0.00013905701984048109,
+      "loss": 1.1729,
+      "step": 11205
+    },
+    {
+      "epoch": 0.8619336974078917,
+      "learning_rate": 0.00013890464581637395,
+      "loss": 1.2125,
+      "step": 11206
+    },
+    {
+      "epoch": 0.8620106145681101,
+      "learning_rate": 0.0001387523512680438,
+      "loss": 1.3013,
+      "step": 11207
+    },
+    {
+      "epoch": 0.8620875317283286,
+      "learning_rate": 0.00013860013620438182,
+      "loss": 1.1944,
+      "step": 11208
+    },
+    {
+      "epoch": 0.862164448888547,
+      "learning_rate": 0.00013844800063427692,
+      "loss": 1.2554,
+      "step": 11209
+    },
+    {
+      "epoch": 0.8622413660487654,
+      "learning_rate": 0.00013829594456661232,
+      "loss": 1.2009,
+      "step": 11210
+    },
+    {
+      "epoch": 0.862318283208984,
+      "learning_rate": 0.00013814396801026675,
+      "loss": 1.027,
+      "step": 11211
+    },
+    {
+      "epoch": 0.8623952003692024,
+      "learning_rate": 0.0001379920709741141,
+      "loss": 0.7939,
+      "step": 11212
+    },
+    {
+      "epoch": 0.8624721175294208,
+      "learning_rate": 0.00013784025346702379,
+      "loss": 1.0146,
+      "step": 11213
+    },
+    {
+      "epoch": 0.8625490346896393,
+      "learning_rate": 0.00013768851549786056,
+      "loss": 1.5782,
+      "step": 11214
+    },
+    {
+      "epoch": 0.8626259518498577,
+      "learning_rate": 0.0001375368570754848,
+      "loss": 1.0979,
+      "step": 11215
+    },
+    {
+      "epoch": 0.8627028690100762,
+      "learning_rate": 0.00013738527820875198,
+      "loss": 1.163,
+      "step": 11216
+    },
+    {
+      "epoch": 0.8627797861702946,
+      "learning_rate": 0.00013723377890651278,
+      "loss": 0.9345,
+      "step": 11217
+    },
+    {
+      "epoch": 0.862856703330513,
+      "learning_rate": 0.00013708235917761353,
+      "loss": 1.1542,
+      "step": 11218
+    },
+    {
+      "epoch": 0.8629336204907315,
+      "learning_rate": 0.00013693101903089545,
+      "loss": 1.1283,
+      "step": 11219
+    },
+    {
+      "epoch": 0.8630105376509499,
+      "learning_rate": 0.0001367797584751963,
+      "loss": 0.9504,
+      "step": 11220
+    },
+    {
+      "epoch": 0.8630874548111683,
+      "learning_rate": 0.00013662857751934755,
+      "loss": 0.9247,
+      "step": 11221
+    },
+    {
+      "epoch": 0.8631643719713868,
+      "learning_rate": 0.0001364774761721766,
+      "loss": 1.3241,
+      "step": 11222
+    },
+    {
+      "epoch": 0.8632412891316052,
+      "learning_rate": 0.00013632645444250723,
+      "loss": 1.2632,
+      "step": 11223
+    },
+    {
+      "epoch": 0.8633182062918237,
+      "learning_rate": 0.00013617551233915725,
+      "loss": 0.9872,
+      "step": 11224
+    },
+    {
+      "epoch": 0.8633951234520422,
+      "learning_rate": 0.0001360246498709406,
+      "loss": 1.1072,
+      "step": 11225
+    },
+    {
+      "epoch": 0.8634720406122606,
+      "learning_rate": 0.00013587386704666605,
+      "loss": 1.4621,
+      "step": 11226
+    },
+    {
+      "epoch": 0.8635489577724791,
+      "learning_rate": 0.00013572316387513772,
+      "loss": 1.1754,
+      "step": 11227
+    },
+    {
+      "epoch": 0.8636258749326975,
+      "learning_rate": 0.00013557254036515609,
+      "loss": 1.1917,
+      "step": 11228
+    },
+    {
+      "epoch": 0.8637027920929159,
+      "learning_rate": 0.00013542199652551596,
+      "loss": 1.0406,
+      "step": 11229
+    },
+    {
+      "epoch": 0.8637797092531344,
+      "learning_rate": 0.0001352715323650071,
+      "loss": 1.1595,
+      "step": 11230
+    },
+    {
+      "epoch": 0.8638566264133528,
+      "learning_rate": 0.00013512114789241604,
+      "loss": 1.4831,
+      "step": 11231
+    },
+    {
+      "epoch": 0.8639335435735712,
+      "learning_rate": 0.00013497084311652353,
+      "loss": 1.4832,
+      "step": 11232
+    },
+    {
+      "epoch": 0.8640104607337897,
+      "learning_rate": 0.00013482061804610622,
+      "loss": 1.3306,
+      "step": 11233
+    },
+    {
+      "epoch": 0.8640873778940081,
+      "learning_rate": 0.00013467047268993565,
+      "loss": 1.2282,
+      "step": 11234
+    },
+    {
+      "epoch": 0.8641642950542266,
+      "learning_rate": 0.00013452040705677875,
+      "loss": 1.1446,
+      "step": 11235
+    },
+    {
+      "epoch": 0.864241212214445,
+      "learning_rate": 0.0001343704211553989,
+      "loss": 1.041,
+      "step": 11236
+    },
+    {
+      "epoch": 0.8643181293746635,
+      "learning_rate": 0.00013422051499455319,
+      "loss": 1.1237,
+      "step": 11237
+    },
+    {
+      "epoch": 0.864395046534882,
+      "learning_rate": 0.00013407068858299498,
+      "loss": 0.8111,
+      "step": 11238
+    },
+    {
+      "epoch": 0.8644719636951004,
+      "learning_rate": 0.0001339209419294728,
+      "loss": 1.0908,
+      "step": 11239
+    },
+    {
+      "epoch": 0.8645488808553188,
+      "learning_rate": 0.00013377127504273062,
+      "loss": 1.1345,
+      "step": 11240
+    },
+    {
+      "epoch": 0.8646257980155373,
+      "learning_rate": 0.00013362168793150714,
+      "loss": 1.269,
+      "step": 11241
+    },
+    {
+      "epoch": 0.8647027151757557,
+      "learning_rate": 0.00013347218060453787,
+      "loss": 1.4384,
+      "step": 11242
+    },
+    {
+      "epoch": 0.8647796323359741,
+      "learning_rate": 0.00013332275307055197,
+      "loss": 1.1863,
+      "step": 11243
+    },
+    {
+      "epoch": 0.8648565494961926,
+      "learning_rate": 0.0001331734053382745,
+      "loss": 1.0937,
+      "step": 11244
+    },
+    {
+      "epoch": 0.864933466656411,
+      "learning_rate": 0.0001330241374164266,
+      "loss": 1.4534,
+      "step": 11245
+    },
+    {
+      "epoch": 0.8650103838166295,
+      "learning_rate": 0.00013287494931372401,
+      "loss": 1.105,
+      "step": 11246
+    },
+    {
+      "epoch": 0.8650873009768479,
+      "learning_rate": 0.00013272584103887774,
+      "loss": 1.1691,
+      "step": 11247
+    },
+    {
+      "epoch": 0.8651642181370663,
+      "learning_rate": 0.00013257681260059483,
+      "loss": 1.1537,
+      "step": 11248
+    },
+    {
+      "epoch": 0.8652411352972849,
+      "learning_rate": 0.00013242786400757651,
+      "loss": 1.097,
+      "step": 11249
+    },
+    {
+      "epoch": 0.8653180524575033,
+      "learning_rate": 0.00013227899526852062,
+      "loss": 0.9412,
+      "step": 11250
+    },
+    {
+      "epoch": 0.8653949696177217,
+      "learning_rate": 0.00013213020639212003,
+      "loss": 1.1851,
+      "step": 11251
+    },
+    {
+      "epoch": 0.8654718867779402,
+      "learning_rate": 0.00013198149738706167,
+      "loss": 1.1797,
+      "step": 11252
+    },
+    {
+      "epoch": 0.8655488039381586,
+      "learning_rate": 0.00013183286826202973,
+      "loss": 1.1292,
+      "step": 11253
+    },
+    {
+      "epoch": 0.8656257210983771,
+      "learning_rate": 0.0001316843190257021,
+      "loss": 1.3159,
+      "step": 11254
+    },
+    {
+      "epoch": 0.8657026382585955,
+      "learning_rate": 0.0001315358496867537,
+      "loss": 0.8295,
+      "step": 11255
+    },
+    {
+      "epoch": 0.8657795554188139,
+      "learning_rate": 0.0001313874602538529,
+      "loss": 1.0835,
+      "step": 11256
+    },
+    {
+      "epoch": 0.8658564725790324,
+      "learning_rate": 0.00013123915073566427,
+      "loss": 1.1864,
+      "step": 11257
+    },
+    {
+      "epoch": 0.8659333897392508,
+      "learning_rate": 0.0001310909211408486,
+      "loss": 1.0999,
+      "step": 11258
+    },
+    {
+      "epoch": 0.8660103068994692,
+      "learning_rate": 0.00013094277147806055,
+      "loss": 1.0201,
+      "step": 11259
+    },
+    {
+      "epoch": 0.8660872240596877,
+      "learning_rate": 0.00013079470175595092,
+      "loss": 1.3484,
+      "step": 11260
+    },
+    {
+      "epoch": 0.8661641412199061,
+      "learning_rate": 0.00013064671198316564,
+      "loss": 1.1714,
+      "step": 11261
+    },
+    {
+      "epoch": 0.8662410583801246,
+      "learning_rate": 0.00013049880216834593,
+      "loss": 1.1316,
+      "step": 11262
+    },
+    {
+      "epoch": 0.8663179755403431,
+      "learning_rate": 0.00013035097232012805,
+      "loss": 1.0454,
+      "step": 11263
+    },
+    {
+      "epoch": 0.8663948927005615,
+      "learning_rate": 0.00013020322244714495,
+      "loss": 1.5866,
+      "step": 11264
+    },
+    {
+      "epoch": 0.86647180986078,
+      "learning_rate": 0.0001300555525580227,
+      "loss": 0.8978,
+      "step": 11265
+    },
+    {
+      "epoch": 0.8665487270209984,
+      "learning_rate": 0.00012990796266138476,
+      "loss": 1.2143,
+      "step": 11266
+    },
+    {
+      "epoch": 0.8666256441812168,
+      "learning_rate": 0.00012976045276584874,
+      "loss": 1.3761,
+      "step": 11267
+    },
+    {
+      "epoch": 0.8667025613414353,
+      "learning_rate": 0.00012961302288002792,
+      "loss": 0.7814,
+      "step": 11268
+    },
+    {
+      "epoch": 0.8667794785016537,
+      "learning_rate": 0.00012946567301253087,
+      "loss": 1.6697,
+      "step": 11269
+    },
+    {
+      "epoch": 0.8668563956618721,
+      "learning_rate": 0.00012931840317196143,
+      "loss": 0.7685,
+      "step": 11270
+    },
+    {
+      "epoch": 0.8669333128220906,
+      "learning_rate": 0.00012917121336691883,
+      "loss": 0.9561,
+      "step": 11271
+    },
+    {
+      "epoch": 0.867010229982309,
+      "learning_rate": 0.00012902410360599777,
+      "loss": 1.2734,
+      "step": 11272
+    },
+    {
+      "epoch": 0.8670871471425275,
+      "learning_rate": 0.0001288770738977883,
+      "loss": 1.241,
+      "step": 11273
+    },
+    {
+      "epoch": 0.867164064302746,
+      "learning_rate": 0.00012873012425087526,
+      "loss": 0.8868,
+      "step": 11274
+    },
+    {
+      "epoch": 0.8672409814629644,
+      "learning_rate": 0.00012858325467383946,
+      "loss": 1.0719,
+      "step": 11275
+    },
+    {
+      "epoch": 0.8673178986231829,
+      "learning_rate": 0.00012843646517525636,
+      "loss": 1.1167,
+      "step": 11276
+    },
+    {
+      "epoch": 0.8673948157834013,
+      "learning_rate": 0.00012828975576369812,
+      "loss": 0.6641,
+      "step": 11277
+    },
+    {
+      "epoch": 0.8674717329436197,
+      "learning_rate": 0.00012814312644773019,
+      "loss": 0.9545,
+      "step": 11278
+    },
+    {
+      "epoch": 0.8675486501038382,
+      "learning_rate": 0.00012799657723591473,
+      "loss": 1.1492,
+      "step": 11279
+    },
+    {
+      "epoch": 0.8676255672640566,
+      "learning_rate": 0.0001278501081368091,
+      "loss": 1.3582,
+      "step": 11280
+    },
+    {
+      "epoch": 0.8677024844242751,
+      "learning_rate": 0.00012770371915896573,
+      "loss": 1.2481,
+      "step": 11281
+    },
+    {
+      "epoch": 0.8677794015844935,
+      "learning_rate": 0.00012755741031093232,
+      "loss": 1.2525,
+      "step": 11282
+    },
+    {
+      "epoch": 0.8678563187447119,
+      "learning_rate": 0.0001274111816012522,
+      "loss": 1.2596,
+      "step": 11283
+    },
+    {
+      "epoch": 0.8679332359049304,
+      "learning_rate": 0.0001272650330384632,
+      "loss": 0.8427,
+      "step": 11284
+    },
+    {
+      "epoch": 0.8680101530651488,
+      "learning_rate": 0.0001271189646311,
+      "loss": 1.0808,
+      "step": 11285
+    },
+    {
+      "epoch": 0.8680870702253672,
+      "learning_rate": 0.00012697297638769145,
+      "loss": 1.0008,
+      "step": 11286
+    },
+    {
+      "epoch": 0.8681639873855858,
+      "learning_rate": 0.00012682706831676106,
+      "loss": 1.4559,
+      "step": 11287
+    },
+    {
+      "epoch": 0.8682409045458042,
+      "learning_rate": 0.00012668124042682967,
+      "loss": 1.5305,
+      "step": 11288
+    },
+    {
+      "epoch": 0.8683178217060226,
+      "learning_rate": 0.0001265354927264118,
+      "loss": 1.0053,
+      "step": 11289
+    },
+    {
+      "epoch": 0.8683947388662411,
+      "learning_rate": 0.000126389825224018,
+      "loss": 1.183,
+      "step": 11290
+    },
+    {
+      "epoch": 0.8684716560264595,
+      "learning_rate": 0.00012624423792815376,
+      "loss": 1.0886,
+      "step": 11291
+    },
+    {
+      "epoch": 0.868548573186678,
+      "learning_rate": 0.00012609873084732,
+      "loss": 1.2597,
+      "step": 11292
+    },
+    {
+      "epoch": 0.8686254903468964,
+      "learning_rate": 0.00012595330399001336,
+      "loss": 0.8537,
+      "step": 11293
+    },
+    {
+      "epoch": 0.8687024075071148,
+      "learning_rate": 0.00012580795736472528,
+      "loss": 0.8841,
+      "step": 11294
+    },
+    {
+      "epoch": 0.8687793246673333,
+      "learning_rate": 0.00012566269097994277,
+      "loss": 1.1812,
+      "step": 11295
+    },
+    {
+      "epoch": 0.8688562418275517,
+      "learning_rate": 0.00012551750484414802,
+      "loss": 0.7635,
+      "step": 11296
+    },
+    {
+      "epoch": 0.8689331589877701,
+      "learning_rate": 0.00012537239896581864,
+      "loss": 0.9661,
+      "step": 11297
+    },
+    {
+      "epoch": 0.8690100761479886,
+      "learning_rate": 0.00012522737335342717,
+      "loss": 1.2354,
+      "step": 11298
+    },
+    {
+      "epoch": 0.869086993308207,
+      "learning_rate": 0.0001250824280154426,
+      "loss": 1.0257,
+      "step": 11299
+    },
+    {
+      "epoch": 0.8691639104684256,
+      "learning_rate": 0.0001249375629603276,
+      "loss": 1.1796,
+      "step": 11300
+    },
+    {
+      "epoch": 0.869240827628644,
+      "learning_rate": 0.0001247927781965415,
+      "loss": 1.1895,
+      "step": 11301
+    },
+    {
+      "epoch": 0.8693177447888624,
+      "learning_rate": 0.0001246480737325382,
+      "loss": 1.1967,
+      "step": 11302
+    },
+    {
+      "epoch": 0.8693946619490809,
+      "learning_rate": 0.0001245034495767675,
+      "loss": 1.1796,
+      "step": 11303
+    },
+    {
+      "epoch": 0.8694715791092993,
+      "learning_rate": 0.00012435890573767361,
+      "loss": 1.1002,
+      "step": 11304
+    },
+    {
+      "epoch": 0.8695484962695177,
+      "learning_rate": 0.0001242144422236971,
+      "loss": 1.2278,
+      "step": 11305
+    },
+    {
+      "epoch": 0.8696254134297362,
+      "learning_rate": 0.00012407005904327274,
+      "loss": 0.9246,
+      "step": 11306
+    },
+    {
+      "epoch": 0.8697023305899546,
+      "learning_rate": 0.00012392575620483192,
+      "loss": 0.9358,
+      "step": 11307
+    },
+    {
+      "epoch": 0.869779247750173,
+      "learning_rate": 0.0001237815337168001,
+      "loss": 1.2992,
+      "step": 11308
+    },
+    {
+      "epoch": 0.8698561649103915,
+      "learning_rate": 0.00012363739158759908,
+      "loss": 1.2913,
+      "step": 11309
+    },
+    {
+      "epoch": 0.8699330820706099,
+      "learning_rate": 0.00012349332982564503,
+      "loss": 1.1436,
+      "step": 11310
+    },
+    {
+      "epoch": 0.8700099992308284,
+      "learning_rate": 0.0001233493484393498,
+      "loss": 0.8496,
+      "step": 11311
+    },
+    {
+      "epoch": 0.8700869163910468,
+      "learning_rate": 0.0001232054474371213,
+      "loss": 1.3898,
+      "step": 11312
+    },
+    {
+      "epoch": 0.8701638335512653,
+      "learning_rate": 0.00012306162682736133,
+      "loss": 1.2717,
+      "step": 11313
+    },
+    {
+      "epoch": 0.8702407507114838,
+      "learning_rate": 0.00012291788661846764,
+      "loss": 1.4172,
+      "step": 11314
+    },
+    {
+      "epoch": 0.8703176678717022,
+      "learning_rate": 0.00012277422681883403,
+      "loss": 0.8232,
+      "step": 11315
+    },
+    {
+      "epoch": 0.8703945850319206,
+      "learning_rate": 0.00012263064743684866,
+      "loss": 1.4226,
+      "step": 11316
+    },
+    {
+      "epoch": 0.8704715021921391,
+      "learning_rate": 0.00012248714848089526,
+      "loss": 1.2452,
+      "step": 11317
+    },
+    {
+      "epoch": 0.8705484193523575,
+      "learning_rate": 0.00012234372995935283,
+      "loss": 1.2566,
+      "step": 11318
+    },
+    {
+      "epoch": 0.870625336512576,
+      "learning_rate": 0.0001222003918805955,
+      "loss": 0.9356,
+      "step": 11319
+    },
+    {
+      "epoch": 0.8707022536727944,
+      "learning_rate": 0.00012205713425299358,
+      "loss": 1.2432,
+      "step": 11320
+    },
+    {
+      "epoch": 0.8707791708330128,
+      "learning_rate": 0.00012191395708491154,
+      "loss": 1.3988,
+      "step": 11321
+    },
+    {
+      "epoch": 0.8708560879932313,
+      "learning_rate": 0.0001217708603847097,
+      "loss": 1.4929,
+      "step": 11322
+    },
+    {
+      "epoch": 0.8709330051534497,
+      "learning_rate": 0.00012162784416074369,
+      "loss": 1.0432,
+      "step": 11323
+    },
+    {
+      "epoch": 0.8710099223136681,
+      "learning_rate": 0.0001214849084213645,
+      "loss": 1.2808,
+      "step": 11324
+    },
+    {
+      "epoch": 0.8710868394738867,
+      "learning_rate": 0.00012134205317491798,
+      "loss": 1.1407,
+      "step": 11325
+    },
+    {
+      "epoch": 0.8711637566341051,
+      "learning_rate": 0.00012119927842974592,
+      "loss": 1.1389,
+      "step": 11326
+    },
+    {
+      "epoch": 0.8712406737943235,
+      "learning_rate": 0.00012105658419418486,
+      "loss": 1.197,
+      "step": 11327
+    },
+    {
+      "epoch": 0.871317590954542,
+      "learning_rate": 0.00012091397047656677,
+      "loss": 1.1079,
+      "step": 11328
+    },
+    {
+      "epoch": 0.8713945081147604,
+      "learning_rate": 0.00012077143728521933,
+      "loss": 1.0327,
+      "step": 11329
+    },
+    {
+      "epoch": 0.8714714252749789,
+      "learning_rate": 0.00012062898462846523,
+      "loss": 1.0879,
+      "step": 11330
+    },
+    {
+      "epoch": 0.8715483424351973,
+      "learning_rate": 0.00012048661251462212,
+      "loss": 0.8193,
+      "step": 11331
+    },
+    {
+      "epoch": 0.8716252595954157,
+      "learning_rate": 0.00012034432095200354,
+      "loss": 1.2071,
+      "step": 11332
+    },
+    {
+      "epoch": 0.8717021767556342,
+      "learning_rate": 0.00012020210994891751,
+      "loss": 1.6704,
+      "step": 11333
+    },
+    {
+      "epoch": 0.8717790939158526,
+      "learning_rate": 0.00012005997951366854,
+      "loss": 1.1841,
+      "step": 11334
+    },
+    {
+      "epoch": 0.871856011076071,
+      "learning_rate": 0.00011991792965455584,
+      "loss": 1.469,
+      "step": 11335
+    },
+    {
+      "epoch": 0.8719329282362895,
+      "learning_rate": 0.00011977596037987293,
+      "loss": 1.0647,
+      "step": 11336
+    },
+    {
+      "epoch": 0.8720098453965079,
+      "learning_rate": 0.00011963407169791052,
+      "loss": 1.1584,
+      "step": 11337
+    },
+    {
+      "epoch": 0.8720867625567265,
+      "learning_rate": 0.00011949226361695297,
+      "loss": 1.1649,
+      "step": 11338
+    },
+    {
+      "epoch": 0.8721636797169449,
+      "learning_rate": 0.00011935053614528118,
+      "loss": 1.0068,
+      "step": 11339
+    },
+    {
+      "epoch": 0.8722405968771633,
+      "learning_rate": 0.00011920888929117035,
+      "loss": 1.155,
+      "step": 11340
+    },
+    {
+      "epoch": 0.8723175140373818,
+      "learning_rate": 0.0001190673230628912,
+      "loss": 0.931,
+      "step": 11341
+    },
+    {
+      "epoch": 0.8723944311976002,
+      "learning_rate": 0.00011892583746871045,
+      "loss": 0.879,
+      "step": 11342
+    },
+    {
+      "epoch": 0.8724713483578186,
+      "learning_rate": 0.00011878443251688947,
+      "loss": 0.7196,
+      "step": 11343
+    },
+    {
+      "epoch": 0.8725482655180371,
+      "learning_rate": 0.00011864310821568486,
+      "loss": 1.265,
+      "step": 11344
+    },
+    {
+      "epoch": 0.8726251826782555,
+      "learning_rate": 0.00011850186457334882,
+      "loss": 1.2083,
+      "step": 11345
+    },
+    {
+      "epoch": 0.8727020998384739,
+      "learning_rate": 0.00011836070159812878,
+      "loss": 1.1793,
+      "step": 11346
+    },
+    {
+      "epoch": 0.8727790169986924,
+      "learning_rate": 0.0001182196192982668,
+      "loss": 1.3855,
+      "step": 11347
+    },
+    {
+      "epoch": 0.8728559341589108,
+      "learning_rate": 0.00011807861768200196,
+      "loss": 1.7267,
+      "step": 11348
+    },
+    {
+      "epoch": 0.8729328513191293,
+      "learning_rate": 0.00011793769675756632,
+      "loss": 1.0257,
+      "step": 11349
+    },
+    {
+      "epoch": 0.8730097684793477,
+      "learning_rate": 0.00011779685653318916,
+      "loss": 1.1233,
+      "step": 11350
+    },
+    {
+      "epoch": 0.8730866856395662,
+      "learning_rate": 0.00011765609701709407,
+      "loss": 0.7911,
+      "step": 11351
+    },
+    {
+      "epoch": 0.8731636027997847,
+      "learning_rate": 0.00011751541821750012,
+      "loss": 1.1124,
+      "step": 11352
+    },
+    {
+      "epoch": 0.8732405199600031,
+      "learning_rate": 0.00011737482014262157,
+      "loss": 0.952,
+      "step": 11353
+    },
+    {
+      "epoch": 0.8733174371202215,
+      "learning_rate": 0.00011723430280066838,
+      "loss": 1.2934,
+      "step": 11354
+    },
+    {
+      "epoch": 0.87339435428044,
+      "learning_rate": 0.00011709386619984496,
+      "loss": 1.0059,
+      "step": 11355
+    },
+    {
+      "epoch": 0.8734712714406584,
+      "learning_rate": 0.00011695351034835227,
+      "loss": 1.4836,
+      "step": 11356
+    },
+    {
+      "epoch": 0.8735481886008769,
+      "learning_rate": 0.00011681323525438542,
+      "loss": 1.098,
+      "step": 11357
+    },
+    {
+      "epoch": 0.8736251057610953,
+      "learning_rate": 0.00011667304092613534,
+      "loss": 0.8327,
+      "step": 11358
+    },
+    {
+      "epoch": 0.8737020229213137,
+      "learning_rate": 0.000116532927371788,
+      "loss": 0.9604,
+      "step": 11359
+    },
+    {
+      "epoch": 0.8737789400815322,
+      "learning_rate": 0.00011639289459952467,
+      "loss": 1.1297,
+      "step": 11360
+    },
+    {
+      "epoch": 0.8738558572417506,
+      "learning_rate": 0.00011625294261752268,
+      "loss": 1.4265,
+      "step": 11361
+    },
+    {
+      "epoch": 0.873932774401969,
+      "learning_rate": 0.0001161130714339531,
+      "loss": 1.3408,
+      "step": 11362
+    },
+    {
+      "epoch": 0.8740096915621876,
+      "learning_rate": 0.00011597328105698329,
+      "loss": 1.0941,
+      "step": 11363
+    },
+    {
+      "epoch": 0.874086608722406,
+      "learning_rate": 0.00011583357149477636,
+      "loss": 0.908,
+      "step": 11364
+    },
+    {
+      "epoch": 0.8741635258826244,
+      "learning_rate": 0.00011569394275548978,
+      "loss": 1.6905,
+      "step": 11365
+    },
+    {
+      "epoch": 0.8742404430428429,
+      "learning_rate": 0.00011555439484727654,
+      "loss": 0.9637,
+      "step": 11366
+    },
+    {
+      "epoch": 0.8743173602030613,
+      "learning_rate": 0.00011541492777828493,
+      "loss": 1.2012,
+      "step": 11367
+    },
+    {
+      "epoch": 0.8743942773632798,
+      "learning_rate": 0.00011527554155665848,
+      "loss": 0.9998,
+      "step": 11368
+    },
+    {
+      "epoch": 0.8744711945234982,
+      "learning_rate": 0.00011513623619053648,
+      "loss": 1.4104,
+      "step": 11369
+    },
+    {
+      "epoch": 0.8745481116837166,
+      "learning_rate": 0.00011499701168805326,
+      "loss": 1.2468,
+      "step": 11370
+    },
+    {
+      "epoch": 0.8746250288439351,
+      "learning_rate": 0.00011485786805733733,
+      "loss": 0.9052,
+      "step": 11371
+    },
+    {
+      "epoch": 0.8747019460041535,
+      "learning_rate": 0.00011471880530651435,
+      "loss": 1.192,
+      "step": 11372
+    },
+    {
+      "epoch": 0.8747788631643719,
+      "learning_rate": 0.00011457982344370399,
+      "loss": 1.2228,
+      "step": 11373
+    },
+    {
+      "epoch": 0.8748557803245904,
+      "learning_rate": 0.00011444092247702159,
+      "loss": 1.0131,
+      "step": 11374
+    },
+    {
+      "epoch": 0.8749326974848088,
+      "learning_rate": 0.00011430210241457784,
+      "loss": 1.2772,
+      "step": 11375
+    },
+    {
+      "epoch": 0.8750096146450274,
+      "learning_rate": 0.00011416336326447807,
+      "loss": 1.3133,
+      "step": 11376
+    },
+    {
+      "epoch": 0.8750865318052458,
+      "learning_rate": 0.00011402470503482415,
+      "loss": 1.3592,
+      "step": 11377
+    },
+    {
+      "epoch": 0.8751634489654642,
+      "learning_rate": 0.0001138861277337121,
+      "loss": 1.0208,
+      "step": 11378
+    },
+    {
+      "epoch": 0.8752403661256827,
+      "learning_rate": 0.00011374763136923377,
+      "loss": 1.2365,
+      "step": 11379
+    },
+    {
+      "epoch": 0.8753172832859011,
+      "learning_rate": 0.00011360921594947571,
+      "loss": 1.454,
+      "step": 11380
+    },
+    {
+      "epoch": 0.8753942004461195,
+      "learning_rate": 0.0001134708814825206,
+      "loss": 1.2529,
+      "step": 11381
+    },
+    {
+      "epoch": 0.875471117606338,
+      "learning_rate": 0.00011333262797644533,
+      "loss": 1.3665,
+      "step": 11382
+    },
+    {
+      "epoch": 0.8755480347665564,
+      "learning_rate": 0.00011319445543932361,
+      "loss": 1.1139,
+      "step": 11383
+    },
+    {
+      "epoch": 0.8756249519267748,
+      "learning_rate": 0.00011305636387922263,
+      "loss": 1.2085,
+      "step": 11384
+    },
+    {
+      "epoch": 0.8757018690869933,
+      "learning_rate": 0.00011291835330420596,
+      "loss": 1.2508,
+      "step": 11385
+    },
+    {
+      "epoch": 0.8757787862472117,
+      "learning_rate": 0.00011278042372233233,
+      "loss": 1.2565,
+      "step": 11386
+    },
+    {
+      "epoch": 0.8758557034074302,
+      "learning_rate": 0.00011264257514165543,
+      "loss": 1.2357,
+      "step": 11387
+    },
+    {
+      "epoch": 0.8759326205676486,
+      "learning_rate": 0.00011250480757022452,
+      "loss": 1.2803,
+      "step": 11388
+    },
+    {
+      "epoch": 0.876009537727867,
+      "learning_rate": 0.00011236712101608382,
+      "loss": 1.3607,
+      "step": 11389
+    },
+    {
+      "epoch": 0.8760864548880856,
+      "learning_rate": 0.00011222951548727289,
+      "loss": 0.8425,
+      "step": 11390
+    },
+    {
+      "epoch": 0.876163372048304,
+      "learning_rate": 0.00011209199099182715,
+      "loss": 1.1194,
+      "step": 11391
+    },
+    {
+      "epoch": 0.8762402892085224,
+      "learning_rate": 0.00011195454753777668,
+      "loss": 1.0913,
+      "step": 11392
+    },
+    {
+      "epoch": 0.8763172063687409,
+      "learning_rate": 0.0001118171851331462,
+      "loss": 1.1493,
+      "step": 11393
+    },
+    {
+      "epoch": 0.8763941235289593,
+      "learning_rate": 0.00011167990378595733,
+      "loss": 1.5724,
+      "step": 11394
+    },
+    {
+      "epoch": 0.8764710406891778,
+      "learning_rate": 0.00011154270350422547,
+      "loss": 1.3541,
+      "step": 11395
+    },
+    {
+      "epoch": 0.8765479578493962,
+      "learning_rate": 0.00011140558429596271,
+      "loss": 1.1884,
+      "step": 11396
+    },
+    {
+      "epoch": 0.8766248750096146,
+      "learning_rate": 0.00011126854616917481,
+      "loss": 1.22,
+      "step": 11397
+    },
+    {
+      "epoch": 0.8767017921698331,
+      "learning_rate": 0.00011113158913186355,
+      "loss": 1.2094,
+      "step": 11398
+    },
+    {
+      "epoch": 0.8767787093300515,
+      "learning_rate": 0.00011099471319202653,
+      "loss": 1.1837,
+      "step": 11399
+    },
+    {
+      "epoch": 0.8768556264902699,
+      "learning_rate": 0.00011085791835765568,
+      "loss": 1.4606,
+      "step": 11400
+    },
+    {
+      "epoch": 0.8769325436504884,
+      "learning_rate": 0.00011072120463673879,
+      "loss": 0.9544,
+      "step": 11401
+    },
+    {
+      "epoch": 0.8770094608107069,
+      "learning_rate": 0.00011058457203725864,
+      "loss": 0.907,
+      "step": 11402
+    },
+    {
+      "epoch": 0.8770863779709254,
+      "learning_rate": 0.00011044802056719299,
+      "loss": 1.0953,
+      "step": 11403
+    },
+    {
+      "epoch": 0.8771632951311438,
+      "learning_rate": 0.00011031155023451584,
+      "loss": 1.5112,
+      "step": 11404
+    },
+    {
+      "epoch": 0.8772402122913622,
+      "learning_rate": 0.00011017516104719593,
+      "loss": 1.0694,
+      "step": 11405
+    },
+    {
+      "epoch": 0.8773171294515807,
+      "learning_rate": 0.00011003885301319627,
+      "loss": 0.8511,
+      "step": 11406
+    },
+    {
+      "epoch": 0.8773940466117991,
+      "learning_rate": 0.00010990262614047681,
+      "loss": 0.9416,
+      "step": 11407
+    },
+    {
+      "epoch": 0.8774709637720175,
+      "learning_rate": 0.00010976648043699167,
+      "loss": 1.1338,
+      "step": 11408
+    },
+    {
+      "epoch": 0.877547880932236,
+      "learning_rate": 0.00010963041591069067,
+      "loss": 0.8955,
+      "step": 11409
+    },
+    {
+      "epoch": 0.8776247980924544,
+      "learning_rate": 0.00010949443256951864,
+      "loss": 1.1223,
+      "step": 11410
+    },
+    {
+      "epoch": 0.8777017152526728,
+      "learning_rate": 0.00010935853042141586,
+      "loss": 0.9867,
+      "step": 11411
+    },
+    {
+      "epoch": 0.8777786324128913,
+      "learning_rate": 0.00010922270947431767,
+      "loss": 0.6568,
+      "step": 11412
+    },
+    {
+      "epoch": 0.8778555495731097,
+      "learning_rate": 0.00010908696973615506,
+      "loss": 1.3113,
+      "step": 11413
+    },
+    {
+      "epoch": 0.8779324667333283,
+      "learning_rate": 0.00010895131121485369,
+      "loss": 1.2506,
+      "step": 11414
+    },
+    {
+      "epoch": 0.8780093838935467,
+      "learning_rate": 0.00010881573391833521,
+      "loss": 1.4292,
+      "step": 11415
+    },
+    {
+      "epoch": 0.8780863010537651,
+      "learning_rate": 0.00010868023785451581,
+      "loss": 0.8695,
+      "step": 11416
+    },
+    {
+      "epoch": 0.8781632182139836,
+      "learning_rate": 0.00010854482303130698,
+      "loss": 1.0954,
+      "step": 11417
+    },
+    {
+      "epoch": 0.878240135374202,
+      "learning_rate": 0.00010840948945661671,
+      "loss": 1.2705,
+      "step": 11418
+    },
+    {
+      "epoch": 0.8783170525344204,
+      "learning_rate": 0.00010827423713834638,
+      "loss": 1.3253,
+      "step": 11419
+    },
+    {
+      "epoch": 0.8783939696946389,
+      "learning_rate": 0.00010813906608439333,
+      "loss": 0.9382,
+      "step": 11420
+    },
+    {
+      "epoch": 0.8784708868548573,
+      "learning_rate": 0.00010800397630265125,
+      "loss": 1.2581,
+      "step": 11421
+    },
+    {
+      "epoch": 0.8785478040150758,
+      "learning_rate": 0.00010786896780100763,
+      "loss": 1.7604,
+      "step": 11422
+    },
+    {
+      "epoch": 0.8786247211752942,
+      "learning_rate": 0.00010773404058734571,
+      "loss": 1.1944,
+      "step": 11423
+    },
+    {
+      "epoch": 0.8787016383355126,
+      "learning_rate": 0.00010759919466954432,
+      "loss": 1.0712,
+      "step": 11424
+    },
+    {
+      "epoch": 0.8787785554957311,
+      "learning_rate": 0.00010746443005547668,
+      "loss": 1.0611,
+      "step": 11425
+    },
+    {
+      "epoch": 0.8788554726559495,
+      "learning_rate": 0.0001073297467530125,
+      "loss": 1.0331,
+      "step": 11426
+    },
+    {
+      "epoch": 0.878932389816168,
+      "learning_rate": 0.00010719514477001613,
+      "loss": 0.9628,
+      "step": 11427
+    },
+    {
+      "epoch": 0.8790093069763865,
+      "learning_rate": 0.00010706062411434631,
+      "loss": 1.0546,
+      "step": 11428
+    },
+    {
+      "epoch": 0.8790862241366049,
+      "learning_rate": 0.00010692618479385857,
+      "loss": 1.1125,
+      "step": 11429
+    },
+    {
+      "epoch": 0.8791631412968233,
+      "learning_rate": 0.00010679182681640264,
+      "loss": 1.4027,
+      "step": 11430
+    },
+    {
+      "epoch": 0.8792400584570418,
+      "learning_rate": 0.00010665755018982392,
+      "loss": 1.3157,
+      "step": 11431
+    },
+    {
+      "epoch": 0.8793169756172602,
+      "learning_rate": 0.00010652335492196314,
+      "loss": 0.9309,
+      "step": 11432
+    },
+    {
+      "epoch": 0.8793938927774787,
+      "learning_rate": 0.0001063892410206555,
+      "loss": 1.253,
+      "step": 11433
+    },
+    {
+      "epoch": 0.8794708099376971,
+      "learning_rate": 0.00010625520849373277,
+      "loss": 0.9708,
+      "step": 11434
+    },
+    {
+      "epoch": 0.8795477270979155,
+      "learning_rate": 0.00010612125734902083,
+      "loss": 1.188,
+      "step": 11435
+    },
+    {
+      "epoch": 0.879624644258134,
+      "learning_rate": 0.00010598738759434146,
+      "loss": 1.3302,
+      "step": 11436
+    },
+    {
+      "epoch": 0.8797015614183524,
+      "learning_rate": 0.00010585359923751137,
+      "loss": 1.0724,
+      "step": 11437
+    },
+    {
+      "epoch": 0.8797784785785708,
+      "learning_rate": 0.00010571989228634249,
+      "loss": 1.0742,
+      "step": 11438
+    },
+    {
+      "epoch": 0.8798553957387893,
+      "learning_rate": 0.00010558626674864191,
+      "loss": 0.8101,
+      "step": 11439
+    },
+    {
+      "epoch": 0.8799323128990078,
+      "learning_rate": 0.00010545272263221289,
+      "loss": 0.9729,
+      "step": 11440
+    },
+    {
+      "epoch": 0.8800092300592263,
+      "learning_rate": 0.00010531925994485236,
+      "loss": 1.0727,
+      "step": 11441
+    },
+    {
+      "epoch": 0.8800861472194447,
+      "learning_rate": 0.00010518587869435392,
+      "loss": 1.1956,
+      "step": 11442
+    },
+    {
+      "epoch": 0.8801630643796631,
+      "learning_rate": 0.00010505257888850584,
+      "loss": 1.2086,
+      "step": 11443
+    },
+    {
+      "epoch": 0.8802399815398816,
+      "learning_rate": 0.00010491936053509122,
+      "loss": 1.3478,
+      "step": 11444
+    },
+    {
+      "epoch": 0.8803168987001,
+      "learning_rate": 0.00010478622364188917,
+      "loss": 1.0753,
+      "step": 11445
+    },
+    {
+      "epoch": 0.8803938158603184,
+      "learning_rate": 0.00010465316821667364,
+      "loss": 1.152,
+      "step": 11446
+    },
+    {
+      "epoch": 0.8804707330205369,
+      "learning_rate": 0.00010452019426721344,
+      "loss": 1.4289,
+      "step": 11447
+    },
+    {
+      "epoch": 0.8805476501807553,
+      "learning_rate": 0.00010438730180127382,
+      "loss": 1.1304,
+      "step": 11448
+    },
+    {
+      "epoch": 0.8806245673409737,
+      "learning_rate": 0.00010425449082661425,
+      "loss": 1.1484,
+      "step": 11449
+    },
+    {
+      "epoch": 0.8807014845011922,
+      "learning_rate": 0.00010412176135098905,
+      "loss": 1.1992,
+      "step": 11450
+    },
+    {
+      "epoch": 0.8807784016614106,
+      "learning_rate": 0.00010398911338214934,
+      "loss": 0.9903,
+      "step": 11451
+    },
+    {
+      "epoch": 0.8808553188216292,
+      "learning_rate": 0.00010385654692783991,
+      "loss": 1.2172,
+      "step": 11452
+    },
+    {
+      "epoch": 0.8809322359818476,
+      "learning_rate": 0.00010372406199580209,
+      "loss": 0.9563,
+      "step": 11453
+    },
+    {
+      "epoch": 0.881009153142066,
+      "learning_rate": 0.00010359165859377135,
+      "loss": 1.0242,
+      "step": 11454
+    },
+    {
+      "epoch": 0.8810860703022845,
+      "learning_rate": 0.00010345933672947866,
+      "loss": 1.1362,
+      "step": 11455
+    },
+    {
+      "epoch": 0.8811629874625029,
+      "learning_rate": 0.00010332709641065102,
+      "loss": 1.1323,
+      "step": 11456
+    },
+    {
+      "epoch": 0.8812399046227213,
+      "learning_rate": 0.00010319493764500976,
+      "loss": 1.2717,
+      "step": 11457
+    },
+    {
+      "epoch": 0.8813168217829398,
+      "learning_rate": 0.00010306286044027168,
+      "loss": 1.0881,
+      "step": 11458
+    },
+    {
+      "epoch": 0.8813937389431582,
+      "learning_rate": 0.00010293086480414932,
+      "loss": 0.8135,
+      "step": 11459
+    },
+    {
+      "epoch": 0.8814706561033767,
+      "learning_rate": 0.00010279895074434931,
+      "loss": 1.1207,
+      "step": 11460
+    },
+    {
+      "epoch": 0.8815475732635951,
+      "learning_rate": 0.000102667118268575,
+      "loss": 1.0214,
+      "step": 11461
+    },
+    {
+      "epoch": 0.8816244904238135,
+      "learning_rate": 0.0001025353673845239,
+      "loss": 1.2395,
+      "step": 11462
+    },
+    {
+      "epoch": 0.881701407584032,
+      "learning_rate": 0.00010240369809988903,
+      "loss": 1.1762,
+      "step": 11463
+    },
+    {
+      "epoch": 0.8817783247442504,
+      "learning_rate": 0.0001022721104223589,
+      "loss": 0.9741,
+      "step": 11464
+    },
+    {
+      "epoch": 0.8818552419044688,
+      "learning_rate": 0.00010214060435961686,
+      "loss": 1.0985,
+      "step": 11465
+    },
+    {
+      "epoch": 0.8819321590646874,
+      "learning_rate": 0.00010200917991934177,
+      "loss": 1.1406,
+      "step": 11466
+    },
+    {
+      "epoch": 0.8820090762249058,
+      "learning_rate": 0.0001018778371092075,
+      "loss": 1.1153,
+      "step": 11467
+    },
+    {
+      "epoch": 0.8820859933851242,
+      "learning_rate": 0.00010174657593688358,
+      "loss": 0.7835,
+      "step": 11468
+    },
+    {
+      "epoch": 0.8821629105453427,
+      "learning_rate": 0.00010161539641003404,
+      "loss": 1.6216,
+      "step": 11469
+    },
+    {
+      "epoch": 0.8822398277055611,
+      "learning_rate": 0.00010148429853631907,
+      "loss": 1.2589,
+      "step": 11470
+    },
+    {
+      "epoch": 0.8823167448657796,
+      "learning_rate": 0.00010135328232339341,
+      "loss": 1.2109,
+      "step": 11471
+    },
+    {
+      "epoch": 0.882393662025998,
+      "learning_rate": 0.0001012223477789071,
+      "loss": 0.8953,
+      "step": 11472
+    },
+    {
+      "epoch": 0.8824705791862164,
+      "learning_rate": 0.00010109149491050585,
+      "loss": 1.0965,
+      "step": 11473
+    },
+    {
+      "epoch": 0.8825474963464349,
+      "learning_rate": 0.00010096072372582987,
+      "loss": 0.8941,
+      "step": 11474
+    },
+    {
+      "epoch": 0.8826244135066533,
+      "learning_rate": 0.0001008300342325154,
+      "loss": 0.9229,
+      "step": 11475
+    },
+    {
+      "epoch": 0.8827013306668717,
+      "learning_rate": 0.00010069942643819368,
+      "loss": 1.3066,
+      "step": 11476
+    },
+    {
+      "epoch": 0.8827782478270902,
+      "learning_rate": 0.00010056890035049043,
+      "loss": 1.253,
+      "step": 11477
+    },
+    {
+      "epoch": 0.8828551649873086,
+      "learning_rate": 0.00010043845597702756,
+      "loss": 1.2333,
+      "step": 11478
+    },
+    {
+      "epoch": 0.8829320821475272,
+      "learning_rate": 0.00010030809332542195,
+      "loss": 1.166,
+      "step": 11479
+    },
+    {
+      "epoch": 0.8830089993077456,
+      "learning_rate": 0.00010017781240328555,
+      "loss": 1.142,
+      "step": 11480
+    },
+    {
+      "epoch": 0.883085916467964,
+      "learning_rate": 0.00010004761321822559,
+      "loss": 1.2751,
+      "step": 11481
+    },
+    {
+      "epoch": 0.8831628336281825,
+      "learning_rate": 9.991749577784415e-05,
+      "loss": 0.9672,
+      "step": 11482
+    },
+    {
+      "epoch": 0.8832397507884009,
+      "learning_rate": 9.978746008973949e-05,
+      "loss": 0.7696,
+      "step": 11483
+    },
+    {
+      "epoch": 0.8833166679486193,
+      "learning_rate": 9.96575061615042e-05,
+      "loss": 0.9291,
+      "step": 11484
+    },
+    {
+      "epoch": 0.8833935851088378,
+      "learning_rate": 9.952763400072673e-05,
+      "loss": 1.2045,
+      "step": 11485
+    },
+    {
+      "epoch": 0.8834705022690562,
+      "learning_rate": 9.939784361499016e-05,
+      "loss": 1.2746,
+      "step": 11486
+    },
+    {
+      "epoch": 0.8835474194292746,
+      "learning_rate": 9.92681350118731e-05,
+      "loss": 0.9339,
+      "step": 11487
+    },
+    {
+      "epoch": 0.8836243365894931,
+      "learning_rate": 9.913850819894931e-05,
+      "loss": 0.9674,
+      "step": 11488
+    },
+    {
+      "epoch": 0.8837012537497115,
+      "learning_rate": 9.900896318378844e-05,
+      "loss": 1.3145,
+      "step": 11489
+    },
+    {
+      "epoch": 0.88377817090993,
+      "learning_rate": 9.887949997395373e-05,
+      "loss": 1.014,
+      "step": 11490
+    },
+    {
+      "epoch": 0.8838550880701485,
+      "learning_rate": 9.87501185770055e-05,
+      "loss": 0.8902,
+      "step": 11491
+    },
+    {
+      "epoch": 0.8839320052303669,
+      "learning_rate": 9.862081900049819e-05,
+      "loss": 0.862,
+      "step": 11492
+    },
+    {
+      "epoch": 0.8840089223905854,
+      "learning_rate": 9.849160125198192e-05,
+      "loss": 1.4363,
+      "step": 11493
+    },
+    {
+      "epoch": 0.8840858395508038,
+      "learning_rate": 9.836246533900151e-05,
+      "loss": 1.3161,
+      "step": 11494
+    },
+    {
+      "epoch": 0.8841627567110222,
+      "learning_rate": 9.823341126909741e-05,
+      "loss": 1.1359,
+      "step": 11495
+    },
+    {
+      "epoch": 0.8842396738712407,
+      "learning_rate": 9.810443904980525e-05,
+      "loss": 1.0161,
+      "step": 11496
+    },
+    {
+      "epoch": 0.8843165910314591,
+      "learning_rate": 9.797554868865604e-05,
+      "loss": 1.4108,
+      "step": 11497
+    },
+    {
+      "epoch": 0.8843935081916776,
+      "learning_rate": 9.784674019317574e-05,
+      "loss": 1.1903,
+      "step": 11498
+    },
+    {
+      "epoch": 0.884470425351896,
+      "learning_rate": 9.77180135708855e-05,
+      "loss": 1.2703,
+      "step": 11499
+    },
+    {
+      "epoch": 0.8845473425121144,
+      "learning_rate": 9.758936882930197e-05,
+      "loss": 1.2174,
+      "step": 11500
+    },
+    {
+      "epoch": 0.8846242596723329,
+      "learning_rate": 9.746080597593648e-05,
+      "loss": 1.3184,
+      "step": 11501
+    },
+    {
+      "epoch": 0.8847011768325513,
+      "learning_rate": 9.73323250182967e-05,
+      "loss": 1.0426,
+      "step": 11502
+    },
+    {
+      "epoch": 0.8847780939927697,
+      "learning_rate": 9.720392596388395e-05,
+      "loss": 1.0595,
+      "step": 11503
+    },
+    {
+      "epoch": 0.8848550111529883,
+      "learning_rate": 9.707560882019589e-05,
+      "loss": 1.3072,
+      "step": 11504
+    },
+    {
+      "epoch": 0.8849319283132067,
+      "learning_rate": 9.694737359472522e-05,
+      "loss": 1.1604,
+      "step": 11505
+    },
+    {
+      "epoch": 0.8850088454734252,
+      "learning_rate": 9.68192202949596e-05,
+      "loss": 1.2572,
+      "step": 11506
+    },
+    {
+      "epoch": 0.8850857626336436,
+      "learning_rate": 9.669114892838221e-05,
+      "loss": 1.3691,
+      "step": 11507
+    },
+    {
+      "epoch": 0.885162679793862,
+      "learning_rate": 9.656315950247108e-05,
+      "loss": 1.0542,
+      "step": 11508
+    },
+    {
+      "epoch": 0.8852395969540805,
+      "learning_rate": 9.643525202469938e-05,
+      "loss": 1.2763,
+      "step": 11509
+    },
+    {
+      "epoch": 0.8853165141142989,
+      "learning_rate": 9.630742650253632e-05,
+      "loss": 1.1088,
+      "step": 11510
+    },
+    {
+      "epoch": 0.8853934312745173,
+      "learning_rate": 9.617968294344576e-05,
+      "loss": 1.1865,
+      "step": 11511
+    },
+    {
+      "epoch": 0.8854703484347358,
+      "learning_rate": 9.605202135488605e-05,
+      "loss": 1.0677,
+      "step": 11512
+    },
+    {
+      "epoch": 0.8855472655949542,
+      "learning_rate": 9.592444174431226e-05,
+      "loss": 1.1806,
+      "step": 11513
+    },
+    {
+      "epoch": 0.8856241827551726,
+      "learning_rate": 9.579694411917373e-05,
+      "loss": 0.933,
+      "step": 11514
+    },
+    {
+      "epoch": 0.8857010999153911,
+      "learning_rate": 9.566952848691484e-05,
+      "loss": 1.099,
+      "step": 11515
+    },
+    {
+      "epoch": 0.8857780170756095,
+      "learning_rate": 9.554219485497584e-05,
+      "loss": 1.0346,
+      "step": 11516
+    },
+    {
+      "epoch": 0.8858549342358281,
+      "learning_rate": 9.541494323079159e-05,
+      "loss": 1.0292,
+      "step": 11517
+    },
+    {
+      "epoch": 0.8859318513960465,
+      "learning_rate": 9.52877736217928e-05,
+      "loss": 0.9372,
+      "step": 11518
+    },
+    {
+      "epoch": 0.8860087685562649,
+      "learning_rate": 9.516068603540473e-05,
+      "loss": 1.1542,
+      "step": 11519
+    },
+    {
+      "epoch": 0.8860856857164834,
+      "learning_rate": 9.503368047904842e-05,
+      "loss": 1.0551,
+      "step": 11520
+    },
+    {
+      "epoch": 0.8861626028767018,
+      "learning_rate": 9.490675696013978e-05,
+      "loss": 0.8332,
+      "step": 11521
+    },
+    {
+      "epoch": 0.8862395200369202,
+      "learning_rate": 9.477991548608989e-05,
+      "loss": 1.1166,
+      "step": 11522
+    },
+    {
+      "epoch": 0.8863164371971387,
+      "learning_rate": 9.465315606430513e-05,
+      "loss": 0.8214,
+      "step": 11523
+    },
+    {
+      "epoch": 0.8863933543573571,
+      "learning_rate": 9.452647870218745e-05,
+      "loss": 1.0597,
+      "step": 11524
+    },
+    {
+      "epoch": 0.8864702715175756,
+      "learning_rate": 9.439988340713306e-05,
+      "loss": 1.2656,
+      "step": 11525
+    },
+    {
+      "epoch": 0.886547188677794,
+      "learning_rate": 9.42733701865346e-05,
+      "loss": 1.2766,
+      "step": 11526
+    },
+    {
+      "epoch": 0.8866241058380124,
+      "learning_rate": 9.41469390477791e-05,
+      "loss": 1.1914,
+      "step": 11527
+    },
+    {
+      "epoch": 0.886701022998231,
+      "learning_rate": 9.402058999824886e-05,
+      "loss": 1.3831,
+      "step": 11528
+    },
+    {
+      "epoch": 0.8867779401584494,
+      "learning_rate": 9.389432304532181e-05,
+      "loss": 1.2206,
+      "step": 11529
+    },
+    {
+      "epoch": 0.8868548573186678,
+      "learning_rate": 9.376813819637054e-05,
+      "loss": 1.1589,
+      "step": 11530
+    },
+    {
+      "epoch": 0.8869317744788863,
+      "learning_rate": 9.364203545876298e-05,
+      "loss": 1.1878,
+      "step": 11531
+    },
+    {
+      "epoch": 0.8870086916391047,
+      "learning_rate": 9.351601483986294e-05,
+      "loss": 1.0108,
+      "step": 11532
+    },
+    {
+      "epoch": 0.8870856087993231,
+      "learning_rate": 9.339007634702867e-05,
+      "loss": 0.888,
+      "step": 11533
+    },
+    {
+      "epoch": 0.8871625259595416,
+      "learning_rate": 9.326421998761347e-05,
+      "loss": 1.2003,
+      "step": 11534
+    },
+    {
+      "epoch": 0.88723944311976,
+      "learning_rate": 9.313844576896662e-05,
+      "loss": 1.4812,
+      "step": 11535
+    },
+    {
+      "epoch": 0.8873163602799785,
+      "learning_rate": 9.30127536984321e-05,
+      "loss": 0.8595,
+      "step": 11536
+    },
+    {
+      "epoch": 0.8873932774401969,
+      "learning_rate": 9.28871437833495e-05,
+      "loss": 1.125,
+      "step": 11537
+    },
+    {
+      "epoch": 0.8874701946004153,
+      "learning_rate": 9.276161603105282e-05,
+      "loss": 1.3038,
+      "step": 11538
+    },
+    {
+      "epoch": 0.8875471117606338,
+      "learning_rate": 9.263617044887185e-05,
+      "loss": 1.1381,
+      "step": 11539
+    },
+    {
+      "epoch": 0.8876240289208522,
+      "learning_rate": 9.251080704413173e-05,
+      "loss": 0.8655,
+      "step": 11540
+    },
+    {
+      "epoch": 0.8877009460810706,
+      "learning_rate": 9.23855258241526e-05,
+      "loss": 1.2277,
+      "step": 11541
+    },
+    {
+      "epoch": 0.8877778632412892,
+      "learning_rate": 9.226032679624946e-05,
+      "loss": 0.9171,
+      "step": 11542
+    },
+    {
+      "epoch": 0.8878547804015076,
+      "learning_rate": 9.213520996773295e-05,
+      "loss": 1.4198,
+      "step": 11543
+    },
+    {
+      "epoch": 0.8879316975617261,
+      "learning_rate": 9.201017534590855e-05,
+      "loss": 1.0412,
+      "step": 11544
+    },
+    {
+      "epoch": 0.8880086147219445,
+      "learning_rate": 9.18852229380776e-05,
+      "loss": 1.0927,
+      "step": 11545
+    },
+    {
+      "epoch": 0.8880855318821629,
+      "learning_rate": 9.176035275153626e-05,
+      "loss": 1.1522,
+      "step": 11546
+    },
+    {
+      "epoch": 0.8881624490423814,
+      "learning_rate": 9.163556479357504e-05,
+      "loss": 1.1021,
+      "step": 11547
+    },
+    {
+      "epoch": 0.8882393662025998,
+      "learning_rate": 9.151085907148127e-05,
+      "loss": 1.2182,
+      "step": 11548
+    },
+    {
+      "epoch": 0.8883162833628182,
+      "learning_rate": 9.138623559253612e-05,
+      "loss": 0.8963,
+      "step": 11549
+    },
+    {
+      "epoch": 0.8883932005230367,
+      "learning_rate": 9.126169436401693e-05,
+      "loss": 0.8112,
+      "step": 11550
+    },
+    {
+      "epoch": 0.8884701176832551,
+      "learning_rate": 9.11372353931954e-05,
+      "loss": 0.9991,
+      "step": 11551
+    },
+    {
+      "epoch": 0.8885470348434735,
+      "learning_rate": 9.101285868733905e-05,
+      "loss": 1.1995,
+      "step": 11552
+    },
+    {
+      "epoch": 0.888623952003692,
+      "learning_rate": 9.088856425371005e-05,
+      "loss": 1.1579,
+      "step": 11553
+    },
+    {
+      "epoch": 0.8887008691639104,
+      "learning_rate": 9.076435209956663e-05,
+      "loss": 0.8164,
+      "step": 11554
+    },
+    {
+      "epoch": 0.888777786324129,
+      "learning_rate": 9.064022223216129e-05,
+      "loss": 1.2062,
+      "step": 11555
+    },
+    {
+      "epoch": 0.8888547034843474,
+      "learning_rate": 9.051617465874223e-05,
+      "loss": 1.3546,
+      "step": 11556
+    },
+    {
+      "epoch": 0.8889316206445658,
+      "learning_rate": 9.039220938655268e-05,
+      "loss": 1.2306,
+      "step": 11557
+    },
+    {
+      "epoch": 0.8890085378047843,
+      "learning_rate": 9.026832642283067e-05,
+      "loss": 0.76,
+      "step": 11558
+    },
+    {
+      "epoch": 0.8890854549650027,
+      "learning_rate": 9.014452577481092e-05,
+      "loss": 1.4175,
+      "step": 11559
+    },
+    {
+      "epoch": 0.8891623721252211,
+      "learning_rate": 9.00208074497213e-05,
+      "loss": 1.4658,
+      "step": 11560
+    },
+    {
+      "epoch": 0.8892392892854396,
+      "learning_rate": 8.989717145478587e-05,
+      "loss": 1.3656,
+      "step": 11561
+    },
+    {
+      "epoch": 0.889316206445658,
+      "learning_rate": 8.977361779722453e-05,
+      "loss": 1.3049,
+      "step": 11562
+    },
+    {
+      "epoch": 0.8893931236058765,
+      "learning_rate": 8.965014648425135e-05,
+      "loss": 1.3541,
+      "step": 11563
+    },
+    {
+      "epoch": 0.8894700407660949,
+      "learning_rate": 8.952675752307588e-05,
+      "loss": 1.1139,
+      "step": 11564
+    },
+    {
+      "epoch": 0.8895469579263133,
+      "learning_rate": 8.940345092090318e-05,
+      "loss": 1.5293,
+      "step": 11565
+    },
+    {
+      "epoch": 0.8896238750865318,
+      "learning_rate": 8.928022668493269e-05,
+      "loss": 1.3007,
+      "step": 11566
+    },
+    {
+      "epoch": 0.8897007922467502,
+      "learning_rate": 8.915708482236013e-05,
+      "loss": 0.9564,
+      "step": 11567
+    },
+    {
+      "epoch": 0.8897777094069687,
+      "learning_rate": 8.903402534037608e-05,
+      "loss": 1.2317,
+      "step": 11568
+    },
+    {
+      "epoch": 0.8898546265671872,
+      "learning_rate": 8.89110482461653e-05,
+      "loss": 1.1855,
+      "step": 11569
+    },
+    {
+      "epoch": 0.8899315437274056,
+      "learning_rate": 8.878815354690923e-05,
+      "loss": 1.3336,
+      "step": 11570
+    },
+    {
+      "epoch": 0.890008460887624,
+      "learning_rate": 8.866534124978359e-05,
+      "loss": 1.3226,
+      "step": 11571
+    },
+    {
+      "epoch": 0.8900853780478425,
+      "learning_rate": 8.854261136195952e-05,
+      "loss": 1.1299,
+      "step": 11572
+    },
+    {
+      "epoch": 0.8901622952080609,
+      "learning_rate": 8.841996389060325e-05,
+      "loss": 0.9352,
+      "step": 11573
+    },
+    {
+      "epoch": 0.8902392123682794,
+      "learning_rate": 8.829739884287641e-05,
+      "loss": 0.9524,
+      "step": 11574
+    },
+    {
+      "epoch": 0.8903161295284978,
+      "learning_rate": 8.817491622593576e-05,
+      "loss": 0.7863,
+      "step": 11575
+    },
+    {
+      "epoch": 0.8903930466887162,
+      "learning_rate": 8.805251604693326e-05,
+      "loss": 1.3396,
+      "step": 11576
+    },
+    {
+      "epoch": 0.8904699638489347,
+      "learning_rate": 8.793019831301569e-05,
+      "loss": 1.0705,
+      "step": 11577
+    },
+    {
+      "epoch": 0.8905468810091531,
+      "learning_rate": 8.780796303132548e-05,
+      "loss": 0.8763,
+      "step": 11578
+    },
+    {
+      "epoch": 0.8906237981693715,
+      "learning_rate": 8.768581020900012e-05,
+      "loss": 1.0995,
+      "step": 11579
+    },
+    {
+      "epoch": 0.89070071532959,
+      "learning_rate": 8.756373985317189e-05,
+      "loss": 1.2277,
+      "step": 11580
+    },
+    {
+      "epoch": 0.8907776324898085,
+      "learning_rate": 8.744175197096943e-05,
+      "loss": 1.2304,
+      "step": 11581
+    },
+    {
+      "epoch": 0.890854549650027,
+      "learning_rate": 8.73198465695147e-05,
+      "loss": 1.4348,
+      "step": 11582
+    },
+    {
+      "epoch": 0.8909314668102454,
+      "learning_rate": 8.719802365592667e-05,
+      "loss": 1.0804,
+      "step": 11583
+    },
+    {
+      "epoch": 0.8910083839704638,
+      "learning_rate": 8.707628323731848e-05,
+      "loss": 1.4491,
+      "step": 11584
+    },
+    {
+      "epoch": 0.8910853011306823,
+      "learning_rate": 8.695462532079846e-05,
+      "loss": 1.4276,
+      "step": 11585
+    },
+    {
+      "epoch": 0.8911622182909007,
+      "learning_rate": 8.683304991347057e-05,
+      "loss": 1.1899,
+      "step": 11586
+    },
+    {
+      "epoch": 0.8912391354511191,
+      "learning_rate": 8.671155702243366e-05,
+      "loss": 0.8688,
+      "step": 11587
+    },
+    {
+      "epoch": 0.8913160526113376,
+      "learning_rate": 8.659014665478171e-05,
+      "loss": 1.2816,
+      "step": 11588
+    },
+    {
+      "epoch": 0.891392969771556,
+      "learning_rate": 8.646881881760421e-05,
+      "loss": 0.819,
+      "step": 11589
+    },
+    {
+      "epoch": 0.8914698869317744,
+      "learning_rate": 8.634757351798583e-05,
+      "loss": 1.4842,
+      "step": 11590
+    },
+    {
+      "epoch": 0.8915468040919929,
+      "learning_rate": 8.622641076300541e-05,
+      "loss": 1.0664,
+      "step": 11591
+    },
+    {
+      "epoch": 0.8916237212522113,
+      "learning_rate": 8.610533055973847e-05,
+      "loss": 0.8228,
+      "step": 11592
+    },
+    {
+      "epoch": 0.8917006384124299,
+      "learning_rate": 8.598433291525464e-05,
+      "loss": 0.9508,
+      "step": 11593
+    },
+    {
+      "epoch": 0.8917775555726483,
+      "learning_rate": 8.586341783661966e-05,
+      "loss": 0.8014,
+      "step": 11594
+    },
+    {
+      "epoch": 0.8918544727328667,
+      "learning_rate": 8.574258533089335e-05,
+      "loss": 1.1235,
+      "step": 11595
+    },
+    {
+      "epoch": 0.8919313898930852,
+      "learning_rate": 8.562183540513107e-05,
+      "loss": 1.075,
+      "step": 11596
+    },
+    {
+      "epoch": 0.8920083070533036,
+      "learning_rate": 8.550116806638403e-05,
+      "loss": 1.0194,
+      "step": 11597
+    },
+    {
+      "epoch": 0.892085224213522,
+      "learning_rate": 8.538058332169807e-05,
+      "loss": 1.0437,
+      "step": 11598
+    },
+    {
+      "epoch": 0.8921621413737405,
+      "learning_rate": 8.526008117811407e-05,
+      "loss": 1.1943,
+      "step": 11599
+    },
+    {
+      "epoch": 0.8922390585339589,
+      "learning_rate": 8.51396616426684e-05,
+      "loss": 0.6849,
+      "step": 11600
+    },
+    {
+      "epoch": 0.8923159756941774,
+      "learning_rate": 8.50193247223921e-05,
+      "loss": 1.1121,
+      "step": 11601
+    },
+    {
+      "epoch": 0.8923928928543958,
+      "learning_rate": 8.489907042431238e-05,
+      "loss": 1.0803,
+      "step": 11602
+    },
+    {
+      "epoch": 0.8924698100146142,
+      "learning_rate": 8.47788987554508e-05,
+      "loss": 1.3998,
+      "step": 11603
+    },
+    {
+      "epoch": 0.8925467271748327,
+      "learning_rate": 8.465880972282391e-05,
+      "loss": 0.9228,
+      "step": 11604
+    },
+    {
+      "epoch": 0.8926236443350511,
+      "learning_rate": 8.453880333344444e-05,
+      "loss": 1.2237,
+      "step": 11605
+    },
+    {
+      "epoch": 0.8927005614952696,
+      "learning_rate": 8.441887959431926e-05,
+      "loss": 1.1221,
+      "step": 11606
+    },
+    {
+      "epoch": 0.8927774786554881,
+      "learning_rate": 8.429903851245114e-05,
+      "loss": 1.1364,
+      "step": 11607
+    },
+    {
+      "epoch": 0.8928543958157065,
+      "learning_rate": 8.417928009483744e-05,
+      "loss": 1.4891,
+      "step": 11608
+    },
+    {
+      "epoch": 0.8929313129759249,
+      "learning_rate": 8.405960434847109e-05,
+      "loss": 1.404,
+      "step": 11609
+    },
+    {
+      "epoch": 0.8930082301361434,
+      "learning_rate": 8.394001128034001e-05,
+      "loss": 1.1013,
+      "step": 11610
+    },
+    {
+      "epoch": 0.8930851472963618,
+      "learning_rate": 8.38205008974276e-05,
+      "loss": 0.9517,
+      "step": 11611
+    },
+    {
+      "epoch": 0.8931620644565803,
+      "learning_rate": 8.370107320671194e-05,
+      "loss": 0.8864,
+      "step": 11612
+    },
+    {
+      "epoch": 0.8932389816167987,
+      "learning_rate": 8.35817282151668e-05,
+      "loss": 0.9827,
+      "step": 11613
+    },
+    {
+      "epoch": 0.8933158987770171,
+      "learning_rate": 8.346246592976075e-05,
+      "loss": 1.2023,
+      "step": 11614
+    },
+    {
+      "epoch": 0.8933928159372356,
+      "learning_rate": 8.334328635745724e-05,
+      "loss": 0.9427,
+      "step": 11615
+    },
+    {
+      "epoch": 0.893469733097454,
+      "learning_rate": 8.322418950521605e-05,
+      "loss": 1.4364,
+      "step": 11616
+    },
+    {
+      "epoch": 0.8935466502576724,
+      "learning_rate": 8.310517537999091e-05,
+      "loss": 1.4936,
+      "step": 11617
+    },
+    {
+      "epoch": 0.893623567417891,
+      "learning_rate": 8.298624398873095e-05,
+      "loss": 1.0511,
+      "step": 11618
+    },
+    {
+      "epoch": 0.8937004845781094,
+      "learning_rate": 8.286739533838111e-05,
+      "loss": 1.1745,
+      "step": 11619
+    },
+    {
+      "epoch": 0.8937774017383279,
+      "learning_rate": 8.274862943588102e-05,
+      "loss": 1.005,
+      "step": 11620
+    },
+    {
+      "epoch": 0.8938543188985463,
+      "learning_rate": 8.262994628816545e-05,
+      "loss": 1.0272,
+      "step": 11621
+    },
+    {
+      "epoch": 0.8939312360587647,
+      "learning_rate": 8.251134590216452e-05,
+      "loss": 0.984,
+      "step": 11622
+    },
+    {
+      "epoch": 0.8940081532189832,
+      "learning_rate": 8.23928282848032e-05,
+      "loss": 1.2201,
+      "step": 11623
+    },
+    {
+      "epoch": 0.8940850703792016,
+      "learning_rate": 8.227439344300214e-05,
+      "loss": 1.3198,
+      "step": 11624
+    },
+    {
+      "epoch": 0.89416198753942,
+      "learning_rate": 8.215604138367678e-05,
+      "loss": 0.9837,
+      "step": 11625
+    },
+    {
+      "epoch": 0.8942389046996385,
+      "learning_rate": 8.203777211373792e-05,
+      "loss": 1.4332,
+      "step": 11626
+    },
+    {
+      "epoch": 0.8943158218598569,
+      "learning_rate": 8.191958564009122e-05,
+      "loss": 1.2763,
+      "step": 11627
+    },
+    {
+      "epoch": 0.8943927390200754,
+      "learning_rate": 8.18014819696375e-05,
+      "loss": 0.8833,
+      "step": 11628
+    },
+    {
+      "epoch": 0.8944696561802938,
+      "learning_rate": 8.168346110927388e-05,
+      "loss": 1.1425,
+      "step": 11629
+    },
+    {
+      "epoch": 0.8945465733405122,
+      "learning_rate": 8.156552306589071e-05,
+      "loss": 1.0878,
+      "step": 11630
+    },
+    {
+      "epoch": 0.8946234905007308,
+      "learning_rate": 8.144766784637481e-05,
+      "loss": 0.8387,
+      "step": 11631
+    },
+    {
+      "epoch": 0.8947004076609492,
+      "learning_rate": 8.132989545760816e-05,
+      "loss": 1.0278,
+      "step": 11632
+    },
+    {
+      "epoch": 0.8947773248211676,
+      "learning_rate": 8.121220590646727e-05,
+      "loss": 1.01,
+      "step": 11633
+    },
+    {
+      "epoch": 0.8948542419813861,
+      "learning_rate": 8.109459919982448e-05,
+      "loss": 1.0479,
+      "step": 11634
+    },
+    {
+      "epoch": 0.8949311591416045,
+      "learning_rate": 8.097707534454678e-05,
+      "loss": 1.2038,
+      "step": 11635
+    },
+    {
+      "epoch": 0.8950080763018229,
+      "learning_rate": 8.085963434749639e-05,
+      "loss": 0.6738,
+      "step": 11636
+    },
+    {
+      "epoch": 0.8950849934620414,
+      "learning_rate": 8.074227621553076e-05,
+      "loss": 0.9903,
+      "step": 11637
+    },
+    {
+      "epoch": 0.8951619106222598,
+      "learning_rate": 8.062500095550313e-05,
+      "loss": 1.1692,
+      "step": 11638
+    },
+    {
+      "epoch": 0.8952388277824783,
+      "learning_rate": 8.050780857426082e-05,
+      "loss": 1.1571,
+      "step": 11639
+    },
+    {
+      "epoch": 0.8953157449426967,
+      "learning_rate": 8.039069907864688e-05,
+      "loss": 1.0482,
+      "step": 11640
+    },
+    {
+      "epoch": 0.8953926621029151,
+      "learning_rate": 8.027367247549949e-05,
+      "loss": 1.3399,
+      "step": 11641
+    },
+    {
+      "epoch": 0.8954695792631336,
+      "learning_rate": 8.015672877165187e-05,
+      "loss": 1.1453,
+      "step": 11642
+    },
+    {
+      "epoch": 0.895546496423352,
+      "learning_rate": 8.003986797393286e-05,
+      "loss": 1.1411,
+      "step": 11643
+    },
+    {
+      "epoch": 0.8956234135835704,
+      "learning_rate": 7.99230900891657e-05,
+      "loss": 1.2096,
+      "step": 11644
+    },
+    {
+      "epoch": 0.895700330743789,
+      "learning_rate": 7.980639512416893e-05,
+      "loss": 0.8934,
+      "step": 11645
+    },
+    {
+      "epoch": 0.8957772479040074,
+      "learning_rate": 7.968978308575725e-05,
+      "loss": 0.8719,
+      "step": 11646
+    },
+    {
+      "epoch": 0.8958541650642259,
+      "learning_rate": 7.957325398073922e-05,
+      "loss": 1.1567,
+      "step": 11647
+    },
+    {
+      "epoch": 0.8959310822244443,
+      "learning_rate": 7.945680781591924e-05,
+      "loss": 1.2281,
+      "step": 11648
+    },
+    {
+      "epoch": 0.8960079993846627,
+      "learning_rate": 7.93404445980967e-05,
+      "loss": 0.8987,
+      "step": 11649
+    },
+    {
+      "epoch": 0.8960849165448812,
+      "learning_rate": 7.922416433406598e-05,
+      "loss": 1.1464,
+      "step": 11650
+    },
+    {
+      "epoch": 0.8961618337050996,
+      "learning_rate": 7.910796703061734e-05,
+      "loss": 1.4386,
+      "step": 11651
+    },
+    {
+      "epoch": 0.896238750865318,
+      "learning_rate": 7.899185269453534e-05,
+      "loss": 1.4999,
+      "step": 11652
+    },
+    {
+      "epoch": 0.8963156680255365,
+      "learning_rate": 7.887582133259974e-05,
+      "loss": 1.1821,
+      "step": 11653
+    },
+    {
+      "epoch": 0.8963925851857549,
+      "learning_rate": 7.87598729515861e-05,
+      "loss": 1.2184,
+      "step": 11654
+    },
+    {
+      "epoch": 0.8964695023459733,
+      "learning_rate": 7.864400755826467e-05,
+      "loss": 1.1649,
+      "step": 11655
+    },
+    {
+      "epoch": 0.8965464195061918,
+      "learning_rate": 7.85282251594009e-05,
+      "loss": 1.0316,
+      "step": 11656
+    },
+    {
+      "epoch": 0.8966233366664103,
+      "learning_rate": 7.841252576175551e-05,
+      "loss": 1.9651,
+      "step": 11657
+    },
+    {
+      "epoch": 0.8967002538266288,
+      "learning_rate": 7.829690937208412e-05,
+      "loss": 1.2488,
+      "step": 11658
+    },
+    {
+      "epoch": 0.8967771709868472,
+      "learning_rate": 7.818137599713798e-05,
+      "loss": 0.976,
+      "step": 11659
+    },
+    {
+      "epoch": 0.8968540881470656,
+      "learning_rate": 7.806592564366322e-05,
+      "loss": 0.83,
+      "step": 11660
+    },
+    {
+      "epoch": 0.8969310053072841,
+      "learning_rate": 7.795055831840092e-05,
+      "loss": 1.1996,
+      "step": 11661
+    },
+    {
+      "epoch": 0.8970079224675025,
+      "learning_rate": 7.783527402808737e-05,
+      "loss": 1.0164,
+      "step": 11662
+    },
+    {
+      "epoch": 0.8970848396277209,
+      "learning_rate": 7.772007277945454e-05,
+      "loss": 1.0251,
+      "step": 11663
+    },
+    {
+      "epoch": 0.8971617567879394,
+      "learning_rate": 7.760495457922868e-05,
+      "loss": 1.5952,
+      "step": 11664
+    },
+    {
+      "epoch": 0.8972386739481578,
+      "learning_rate": 7.748991943413225e-05,
+      "loss": 0.7782,
+      "step": 11665
+    },
+    {
+      "epoch": 0.8973155911083763,
+      "learning_rate": 7.737496735088156e-05,
+      "loss": 1.1199,
+      "step": 11666
+    },
+    {
+      "epoch": 0.8973925082685947,
+      "learning_rate": 7.726009833618925e-05,
+      "loss": 1.1331,
+      "step": 11667
+    },
+    {
+      "epoch": 0.8974694254288131,
+      "learning_rate": 7.714531239676259e-05,
+      "loss": 1.1998,
+      "step": 11668
+    },
+    {
+      "epoch": 0.8975463425890317,
+      "learning_rate": 7.703060953930408e-05,
+      "loss": 1.2644,
+      "step": 11669
+    },
+    {
+      "epoch": 0.8976232597492501,
+      "learning_rate": 7.691598977051117e-05,
+      "loss": 1.1396,
+      "step": 11670
+    },
+    {
+      "epoch": 0.8977001769094685,
+      "learning_rate": 7.680145309707687e-05,
+      "loss": 1.3195,
+      "step": 11671
+    },
+    {
+      "epoch": 0.897777094069687,
+      "learning_rate": 7.668699952568863e-05,
+      "loss": 1.2649,
+      "step": 11672
+    },
+    {
+      "epoch": 0.8978540112299054,
+      "learning_rate": 7.657262906302998e-05,
+      "loss": 0.6796,
+      "step": 11673
+    },
+    {
+      "epoch": 0.8979309283901238,
+      "learning_rate": 7.645834171577936e-05,
+      "loss": 1.3104,
+      "step": 11674
+    },
+    {
+      "epoch": 0.8980078455503423,
+      "learning_rate": 7.634413749060915e-05,
+      "loss": 1.3297,
+      "step": 11675
+    },
+    {
+      "epoch": 0.8980847627105607,
+      "learning_rate": 7.62300163941888e-05,
+      "loss": 1.1521,
+      "step": 11676
+    },
+    {
+      "epoch": 0.8981616798707792,
+      "learning_rate": 7.611597843318118e-05,
+      "loss": 1.1726,
+      "step": 11677
+    },
+    {
+      "epoch": 0.8982385970309976,
+      "learning_rate": 7.600202361424613e-05,
+      "loss": 1.2127,
+      "step": 11678
+    },
+    {
+      "epoch": 0.898315514191216,
+      "learning_rate": 7.588815194403681e-05,
+      "loss": 0.9094,
+      "step": 11679
+    },
+    {
+      "epoch": 0.8983924313514345,
+      "learning_rate": 7.577436342920208e-05,
+      "loss": 1.1092,
+      "step": 11680
+    },
+    {
+      "epoch": 0.8984693485116529,
+      "learning_rate": 7.566065807638694e-05,
+      "loss": 1.2423,
+      "step": 11681
+    },
+    {
+      "epoch": 0.8985462656718713,
+      "learning_rate": 7.554703589223028e-05,
+      "loss": 1.2819,
+      "step": 11682
+    },
+    {
+      "epoch": 0.8986231828320899,
+      "learning_rate": 7.543349688336676e-05,
+      "loss": 0.9137,
+      "step": 11683
+    },
+    {
+      "epoch": 0.8987000999923083,
+      "learning_rate": 7.532004105642609e-05,
+      "loss": 0.9771,
+      "step": 11684
+    },
+    {
+      "epoch": 0.8987770171525268,
+      "learning_rate": 7.520666841803265e-05,
+      "loss": 1.3371,
+      "step": 11685
+    },
+    {
+      "epoch": 0.8988539343127452,
+      "learning_rate": 7.509337897480695e-05,
+      "loss": 1.3606,
+      "step": 11686
+    },
+    {
+      "epoch": 0.8989308514729636,
+      "learning_rate": 7.498017273336405e-05,
+      "loss": 1.3945,
+      "step": 11687
+    },
+    {
+      "epoch": 0.8990077686331821,
+      "learning_rate": 7.486704970031366e-05,
+      "loss": 0.8943,
+      "step": 11688
+    },
+    {
+      "epoch": 0.8990846857934005,
+      "learning_rate": 7.475400988226166e-05,
+      "loss": 1.0878,
+      "step": 11689
+    },
+    {
+      "epoch": 0.8991616029536189,
+      "learning_rate": 7.464105328580828e-05,
+      "loss": 1.2263,
+      "step": 11690
+    },
+    {
+      "epoch": 0.8992385201138374,
+      "learning_rate": 7.45281799175494e-05,
+      "loss": 1.3543,
+      "step": 11691
+    },
+    {
+      "epoch": 0.8993154372740558,
+      "learning_rate": 7.441538978407558e-05,
+      "loss": 0.594,
+      "step": 11692
+    },
+    {
+      "epoch": 0.8993923544342742,
+      "learning_rate": 7.430268289197289e-05,
+      "loss": 1.0057,
+      "step": 11693
+    },
+    {
+      "epoch": 0.8994692715944927,
+      "learning_rate": 7.419005924782224e-05,
+      "loss": 1.1454,
+      "step": 11694
+    },
+    {
+      "epoch": 0.8995461887547112,
+      "learning_rate": 7.407751885820002e-05,
+      "loss": 1.2633,
+      "step": 11695
+    },
+    {
+      "epoch": 0.8996231059149297,
+      "learning_rate": 7.396506172967766e-05,
+      "loss": 1.1363,
+      "step": 11696
+    },
+    {
+      "epoch": 0.8997000230751481,
+      "learning_rate": 7.385268786882154e-05,
+      "loss": 1.1124,
+      "step": 11697
+    },
+    {
+      "epoch": 0.8997769402353665,
+      "learning_rate": 7.374039728219329e-05,
+      "loss": 1.1851,
+      "step": 11698
+    },
+    {
+      "epoch": 0.899853857395585,
+      "learning_rate": 7.362818997634929e-05,
+      "loss": 1.4856,
+      "step": 11699
+    },
+    {
+      "epoch": 0.8999307745558034,
+      "learning_rate": 7.351606595784233e-05,
+      "loss": 0.9974,
+      "step": 11700
+    },
+    {
+      "epoch": 0.9000076917160218,
+      "learning_rate": 7.340402523321865e-05,
+      "loss": 1.0596,
+      "step": 11701
+    },
+    {
+      "epoch": 0.9000846088762403,
+      "learning_rate": 7.329206780902054e-05,
+      "loss": 1.3564,
+      "step": 11702
+    },
+    {
+      "epoch": 0.9001615260364587,
+      "learning_rate": 7.318019369178574e-05,
+      "loss": 1.1683,
+      "step": 11703
+    },
+    {
+      "epoch": 0.9002384431966772,
+      "learning_rate": 7.30684028880464e-05,
+      "loss": 1.2756,
+      "step": 11704
+    },
+    {
+      "epoch": 0.9003153603568956,
+      "learning_rate": 7.295669540433009e-05,
+      "loss": 1.3974,
+      "step": 11705
+    },
+    {
+      "epoch": 0.900392277517114,
+      "learning_rate": 7.28450712471596e-05,
+      "loss": 1.5267,
+      "step": 11706
+    },
+    {
+      "epoch": 0.9004691946773326,
+      "learning_rate": 7.273353042305241e-05,
+      "loss": 1.3843,
+      "step": 11707
+    },
+    {
+      "epoch": 0.900546111837551,
+      "learning_rate": 7.262207293852214e-05,
+      "loss": 0.9572,
+      "step": 11708
+    },
+    {
+      "epoch": 0.9006230289977694,
+      "learning_rate": 7.251069880007688e-05,
+      "loss": 1.2914,
+      "step": 11709
+    },
+    {
+      "epoch": 0.9006999461579879,
+      "learning_rate": 7.239940801421913e-05,
+      "loss": 1.0633,
+      "step": 11710
+    },
+    {
+      "epoch": 0.9007768633182063,
+      "learning_rate": 7.2288200587448e-05,
+      "loss": 0.968,
+      "step": 11711
+    },
+    {
+      "epoch": 0.9008537804784247,
+      "learning_rate": 7.217707652625666e-05,
+      "loss": 1.3555,
+      "step": 11712
+    },
+    {
+      "epoch": 0.9009306976386432,
+      "learning_rate": 7.206603583713405e-05,
+      "loss": 0.9122,
+      "step": 11713
+    },
+    {
+      "epoch": 0.9010076147988616,
+      "learning_rate": 7.19550785265637e-05,
+      "loss": 1.3127,
+      "step": 11714
+    },
+    {
+      "epoch": 0.9010845319590801,
+      "learning_rate": 7.18442046010242e-05,
+      "loss": 1.1678,
+      "step": 11715
+    },
+    {
+      "epoch": 0.9011614491192985,
+      "learning_rate": 7.173341406699041e-05,
+      "loss": 1.2193,
+      "step": 11716
+    },
+    {
+      "epoch": 0.9012383662795169,
+      "learning_rate": 7.162270693093098e-05,
+      "loss": 1.2674,
+      "step": 11717
+    },
+    {
+      "epoch": 0.9013152834397354,
+      "learning_rate": 7.151208319931024e-05,
+      "loss": 1.3066,
+      "step": 11718
+    },
+    {
+      "epoch": 0.9013922005999538,
+      "learning_rate": 7.140154287858786e-05,
+      "loss": 1.194,
+      "step": 11719
+    },
+    {
+      "epoch": 0.9014691177601722,
+      "learning_rate": 7.129108597521816e-05,
+      "loss": 1.1326,
+      "step": 11720
+    },
+    {
+      "epoch": 0.9015460349203908,
+      "learning_rate": 7.118071249565067e-05,
+      "loss": 1.4602,
+      "step": 11721
+    },
+    {
+      "epoch": 0.9016229520806092,
+      "learning_rate": 7.107042244633088e-05,
+      "loss": 1.2625,
+      "step": 11722
+    },
+    {
+      "epoch": 0.9016998692408277,
+      "learning_rate": 7.0960215833698e-05,
+      "loss": 1.1909,
+      "step": 11723
+    },
+    {
+      "epoch": 0.9017767864010461,
+      "learning_rate": 7.085009266418752e-05,
+      "loss": 0.9271,
+      "step": 11724
+    },
+    {
+      "epoch": 0.9018537035612645,
+      "learning_rate": 7.074005294422964e-05,
+      "loss": 0.9104,
+      "step": 11725
+    },
+    {
+      "epoch": 0.901930620721483,
+      "learning_rate": 7.063009668024956e-05,
+      "loss": 1.2579,
+      "step": 11726
+    },
+    {
+      "epoch": 0.9020075378817014,
+      "learning_rate": 7.05202238786678e-05,
+      "loss": 0.9298,
+      "step": 11727
+    },
+    {
+      "epoch": 0.9020844550419198,
+      "learning_rate": 7.04104345458999e-05,
+      "loss": 1.3275,
+      "step": 11728
+    },
+    {
+      "epoch": 0.9021613722021383,
+      "learning_rate": 7.030072868835641e-05,
+      "loss": 1.109,
+      "step": 11729
+    },
+    {
+      "epoch": 0.9022382893623567,
+      "learning_rate": 7.019110631244369e-05,
+      "loss": 1.2055,
+      "step": 11730
+    },
+    {
+      "epoch": 0.9023152065225752,
+      "learning_rate": 7.008156742456229e-05,
+      "loss": 0.8952,
+      "step": 11731
+    },
+    {
+      "epoch": 0.9023921236827936,
+      "learning_rate": 6.997211203110825e-05,
+      "loss": 1.0172,
+      "step": 11732
+    },
+    {
+      "epoch": 0.902469040843012,
+      "learning_rate": 6.986274013847316e-05,
+      "loss": 0.9172,
+      "step": 11733
+    },
+    {
+      "epoch": 0.9025459580032306,
+      "learning_rate": 6.975345175304287e-05,
+      "loss": 1.2287,
+      "step": 11734
+    },
+    {
+      "epoch": 0.902622875163449,
+      "learning_rate": 6.964424688119947e-05,
+      "loss": 1.2064,
+      "step": 11735
+    },
+    {
+      "epoch": 0.9026997923236674,
+      "learning_rate": 6.953512552931901e-05,
+      "loss": 1.1301,
+      "step": 11736
+    },
+    {
+      "epoch": 0.9027767094838859,
+      "learning_rate": 6.942608770377307e-05,
+      "loss": 1.4835,
+      "step": 11737
+    },
+    {
+      "epoch": 0.9028536266441043,
+      "learning_rate": 6.931713341092905e-05,
+      "loss": 1.1486,
+      "step": 11738
+    },
+    {
+      "epoch": 0.9029305438043227,
+      "learning_rate": 6.920826265714852e-05,
+      "loss": 1.1691,
+      "step": 11739
+    },
+    {
+      "epoch": 0.9030074609645412,
+      "learning_rate": 6.909947544878875e-05,
+      "loss": 1.2209,
+      "step": 11740
+    },
+    {
+      "epoch": 0.9030843781247596,
+      "learning_rate": 6.899077179220198e-05,
+      "loss": 1.6455,
+      "step": 11741
+    },
+    {
+      "epoch": 0.9031612952849781,
+      "learning_rate": 6.888215169373512e-05,
+      "loss": 1.069,
+      "step": 11742
+    },
+    {
+      "epoch": 0.9032382124451965,
+      "learning_rate": 6.877361515973096e-05,
+      "loss": 1.3897,
+      "step": 11743
+    },
+    {
+      "epoch": 0.9033151296054149,
+      "learning_rate": 6.86651621965274e-05,
+      "loss": 1.392,
+      "step": 11744
+    },
+    {
+      "epoch": 0.9033920467656334,
+      "learning_rate": 6.855679281045624e-05,
+      "loss": 1.4806,
+      "step": 11745
+    },
+    {
+      "epoch": 0.9034689639258519,
+      "learning_rate": 6.844850700784588e-05,
+      "loss": 1.0319,
+      "step": 11746
+    },
+    {
+      "epoch": 0.9035458810860703,
+      "learning_rate": 6.834030479501929e-05,
+      "loss": 1.1348,
+      "step": 11747
+    },
+    {
+      "epoch": 0.9036227982462888,
+      "learning_rate": 6.823218617829424e-05,
+      "loss": 1.0737,
+      "step": 11748
+    },
+    {
+      "epoch": 0.9036997154065072,
+      "learning_rate": 6.812415116398401e-05,
+      "loss": 1.2461,
+      "step": 11749
+    },
+    {
+      "epoch": 0.9037766325667257,
+      "learning_rate": 6.801619975839673e-05,
+      "loss": 0.7658,
+      "step": 11750
+    },
+    {
+      "epoch": 0.9038535497269441,
+      "learning_rate": 6.790833196783602e-05,
+      "loss": 1.1775,
+      "step": 11751
+    },
+    {
+      "epoch": 0.9039304668871625,
+      "learning_rate": 6.780054779860034e-05,
+      "loss": 1.1126,
+      "step": 11752
+    },
+    {
+      "epoch": 0.904007384047381,
+      "learning_rate": 6.769284725698316e-05,
+      "loss": 1.1478,
+      "step": 11753
+    },
+    {
+      "epoch": 0.9040843012075994,
+      "learning_rate": 6.758523034927344e-05,
+      "loss": 1.1646,
+      "step": 11754
+    },
+    {
+      "epoch": 0.9041612183678178,
+      "learning_rate": 6.7477697081755e-05,
+      "loss": 1.466,
+      "step": 11755
+    },
+    {
+      "epoch": 0.9042381355280363,
+      "learning_rate": 6.737024746070647e-05,
+      "loss": 1.5361,
+      "step": 11756
+    },
+    {
+      "epoch": 0.9043150526882547,
+      "learning_rate": 6.726288149240284e-05,
+      "loss": 0.8641,
+      "step": 11757
+    },
+    {
+      "epoch": 0.9043919698484731,
+      "learning_rate": 6.715559918311226e-05,
+      "loss": 1.3044,
+      "step": 11758
+    },
+    {
+      "epoch": 0.9044688870086917,
+      "learning_rate": 6.704840053909955e-05,
+      "loss": 1.1893,
+      "step": 11759
+    },
+    {
+      "epoch": 0.9045458041689101,
+      "learning_rate": 6.694128556662422e-05,
+      "loss": 0.8801,
+      "step": 11760
+    },
+    {
+      "epoch": 0.9046227213291286,
+      "learning_rate": 6.683425427194057e-05,
+      "loss": 1.366,
+      "step": 11761
+    },
+    {
+      "epoch": 0.904699638489347,
+      "learning_rate": 6.672730666129861e-05,
+      "loss": 1.1066,
+      "step": 11762
+    },
+    {
+      "epoch": 0.9047765556495654,
+      "learning_rate": 6.662044274094286e-05,
+      "loss": 1.1511,
+      "step": 11763
+    },
+    {
+      "epoch": 0.9048534728097839,
+      "learning_rate": 6.651366251711316e-05,
+      "loss": 1.4392,
+      "step": 11764
+    },
+    {
+      "epoch": 0.9049303899700023,
+      "learning_rate": 6.640696599604485e-05,
+      "loss": 1.2444,
+      "step": 11765
+    },
+    {
+      "epoch": 0.9050073071302207,
+      "learning_rate": 6.630035318396776e-05,
+      "loss": 1.4095,
+      "step": 11766
+    },
+    {
+      "epoch": 0.9050842242904392,
+      "learning_rate": 6.619382408710728e-05,
+      "loss": 0.9544,
+      "step": 11767
+    },
+    {
+      "epoch": 0.9051611414506576,
+      "learning_rate": 6.608737871168357e-05,
+      "loss": 1.0499,
+      "step": 11768
+    },
+    {
+      "epoch": 0.9052380586108761,
+      "learning_rate": 6.598101706391219e-05,
+      "loss": 1.0724,
+      "step": 11769
+    },
+    {
+      "epoch": 0.9053149757710945,
+      "learning_rate": 6.587473915000397e-05,
+      "loss": 1.0502,
+      "step": 11770
+    },
+    {
+      "epoch": 0.905391892931313,
+      "learning_rate": 6.57685449761643e-05,
+      "loss": 0.8725,
+      "step": 11771
+    },
+    {
+      "epoch": 0.9054688100915315,
+      "learning_rate": 6.566243454859372e-05,
+      "loss": 1.0236,
+      "step": 11772
+    },
+    {
+      "epoch": 0.9055457272517499,
+      "learning_rate": 6.55564078734886e-05,
+      "loss": 0.8038,
+      "step": 11773
+    },
+    {
+      "epoch": 0.9056226444119683,
+      "learning_rate": 6.545046495703966e-05,
+      "loss": 1.3477,
+      "step": 11774
+    },
+    {
+      "epoch": 0.9056995615721868,
+      "learning_rate": 6.534460580543327e-05,
+      "loss": 1.1728,
+      "step": 11775
+    },
+    {
+      "epoch": 0.9057764787324052,
+      "learning_rate": 6.523883042485052e-05,
+      "loss": 1.3801,
+      "step": 11776
+    },
+    {
+      "epoch": 0.9058533958926236,
+      "learning_rate": 6.51331388214676e-05,
+      "loss": 1.0779,
+      "step": 11777
+    },
+    {
+      "epoch": 0.9059303130528421,
+      "learning_rate": 6.502753100145609e-05,
+      "loss": 1.2253,
+      "step": 11778
+    },
+    {
+      "epoch": 0.9060072302130605,
+      "learning_rate": 6.492200697098255e-05,
+      "loss": 0.7492,
+      "step": 11779
+    },
+    {
+      "epoch": 0.906084147373279,
+      "learning_rate": 6.481656673620873e-05,
+      "loss": 1.4086,
+      "step": 11780
+    },
+    {
+      "epoch": 0.9061610645334974,
+      "learning_rate": 6.471121030329119e-05,
+      "loss": 1.1758,
+      "step": 11781
+    },
+    {
+      "epoch": 0.9062379816937158,
+      "learning_rate": 6.460593767838203e-05,
+      "loss": 1.1186,
+      "step": 11782
+    },
+    {
+      "epoch": 0.9063148988539343,
+      "learning_rate": 6.450074886762813e-05,
+      "loss": 1.094,
+      "step": 11783
+    },
+    {
+      "epoch": 0.9063918160141528,
+      "learning_rate": 6.439564387717145e-05,
+      "loss": 1.0366,
+      "step": 11784
+    },
+    {
+      "epoch": 0.9064687331743712,
+      "learning_rate": 6.429062271314923e-05,
+      "loss": 1.0036,
+      "step": 11785
+    },
+    {
+      "epoch": 0.9065456503345897,
+      "learning_rate": 6.418568538169372e-05,
+      "loss": 1.1917,
+      "step": 11786
+    },
+    {
+      "epoch": 0.9066225674948081,
+      "learning_rate": 6.408083188893255e-05,
+      "loss": 1.2543,
+      "step": 11787
+    },
+    {
+      "epoch": 0.9066994846550266,
+      "learning_rate": 6.397606224098812e-05,
+      "loss": 1.7143,
+      "step": 11788
+    },
+    {
+      "epoch": 0.906776401815245,
+      "learning_rate": 6.387137644397789e-05,
+      "loss": 1.221,
+      "step": 11789
+    },
+    {
+      "epoch": 0.9068533189754634,
+      "learning_rate": 6.376677450401481e-05,
+      "loss": 1.5333,
+      "step": 11790
+    },
+    {
+      "epoch": 0.9069302361356819,
+      "learning_rate": 6.366225642720629e-05,
+      "loss": 1.2567,
+      "step": 11791
+    },
+    {
+      "epoch": 0.9070071532959003,
+      "learning_rate": 6.355782221965583e-05,
+      "loss": 1.2335,
+      "step": 11792
+    },
+    {
+      "epoch": 0.9070840704561187,
+      "learning_rate": 6.34534718874612e-05,
+      "loss": 1.305,
+      "step": 11793
+    },
+    {
+      "epoch": 0.9071609876163372,
+      "learning_rate": 6.334920543671519e-05,
+      "loss": 0.6841,
+      "step": 11794
+    },
+    {
+      "epoch": 0.9072379047765556,
+      "learning_rate": 6.324502287350659e-05,
+      "loss": 1.4732,
+      "step": 11795
+    },
+    {
+      "epoch": 0.907314821936774,
+      "learning_rate": 6.314092420391837e-05,
+      "loss": 1.0183,
+      "step": 11796
+    },
+    {
+      "epoch": 0.9073917390969926,
+      "learning_rate": 6.303690943402902e-05,
+      "loss": 1.175,
+      "step": 11797
+    },
+    {
+      "epoch": 0.907468656257211,
+      "learning_rate": 6.293297856991214e-05,
+      "loss": 1.0436,
+      "step": 11798
+    },
+    {
+      "epoch": 0.9075455734174295,
+      "learning_rate": 6.282913161763609e-05,
+      "loss": 0.7594,
+      "step": 11799
+    },
+    {
+      "epoch": 0.9076224905776479,
+      "learning_rate": 6.272536858326499e-05,
+      "loss": 0.7875,
+      "step": 11800
+    },
+    {
+      "epoch": 0.9076994077378663,
+      "learning_rate": 6.262168947285768e-05,
+      "loss": 1.0691,
+      "step": 11801
+    },
+    {
+      "epoch": 0.9077763248980848,
+      "learning_rate": 6.25180942924678e-05,
+      "loss": 1.4212,
+      "step": 11802
+    },
+    {
+      "epoch": 0.9078532420583032,
+      "learning_rate": 6.241458304814434e-05,
+      "loss": 0.9927,
+      "step": 11803
+    },
+    {
+      "epoch": 0.9079301592185216,
+      "learning_rate": 6.231115574593182e-05,
+      "loss": 1.0773,
+      "step": 11804
+    },
+    {
+      "epoch": 0.9080070763787401,
+      "learning_rate": 6.220781239186906e-05,
+      "loss": 1.1155,
+      "step": 11805
+    },
+    {
+      "epoch": 0.9080839935389585,
+      "learning_rate": 6.210455299199075e-05,
+      "loss": 1.5424,
+      "step": 11806
+    },
+    {
+      "epoch": 0.908160910699177,
+      "learning_rate": 6.200137755232587e-05,
+      "loss": 1.1167,
+      "step": 11807
+    },
+    {
+      "epoch": 0.9082378278593954,
+      "learning_rate": 6.189828607889947e-05,
+      "loss": 1.1444,
+      "step": 11808
+    },
+    {
+      "epoch": 0.9083147450196138,
+      "learning_rate": 6.179527857773071e-05,
+      "loss": 0.7555,
+      "step": 11809
+    },
+    {
+      "epoch": 0.9083916621798324,
+      "learning_rate": 6.169235505483462e-05,
+      "loss": 1.2785,
+      "step": 11810
+    },
+    {
+      "epoch": 0.9084685793400508,
+      "learning_rate": 6.158951551622088e-05,
+      "loss": 1.3092,
+      "step": 11811
+    },
+    {
+      "epoch": 0.9085454965002692,
+      "learning_rate": 6.148675996789437e-05,
+      "loss": 1.1576,
+      "step": 11812
+    },
+    {
+      "epoch": 0.9086224136604877,
+      "learning_rate": 6.138408841585497e-05,
+      "loss": 1.1901,
+      "step": 11813
+    },
+    {
+      "epoch": 0.9086993308207061,
+      "learning_rate": 6.128150086609819e-05,
+      "loss": 1.3538,
+      "step": 11814
+    },
+    {
+      "epoch": 0.9087762479809245,
+      "learning_rate": 6.117899732461408e-05,
+      "loss": 0.8202,
+      "step": 11815
+    },
+    {
+      "epoch": 0.908853165141143,
+      "learning_rate": 6.107657779738735e-05,
+      "loss": 0.7978,
+      "step": 11816
+    },
+    {
+      "epoch": 0.9089300823013614,
+      "learning_rate": 6.097424229039922e-05,
+      "loss": 1.192,
+      "step": 11817
+    },
+    {
+      "epoch": 0.9090069994615799,
+      "learning_rate": 6.087199080962458e-05,
+      "loss": 1.1199,
+      "step": 11818
+    },
+    {
+      "epoch": 0.9090839166217983,
+      "learning_rate": 6.0769823361034484e-05,
+      "loss": 1.6453,
+      "step": 11819
+    },
+    {
+      "epoch": 0.9091608337820167,
+      "learning_rate": 6.066773995059416e-05,
+      "loss": 1.581,
+      "step": 11820
+    },
+    {
+      "epoch": 0.9092377509422352,
+      "learning_rate": 6.056574058426451e-05,
+      "loss": 1.4239,
+      "step": 11821
+    },
+    {
+      "epoch": 0.9093146681024536,
+      "learning_rate": 6.0463825268001425e-05,
+      "loss": 1.0648,
+      "step": 11822
+    },
+    {
+      "epoch": 0.909391585262672,
+      "learning_rate": 6.0361994007755826e-05,
+      "loss": 1.4888,
+      "step": 11823
+    },
+    {
+      "epoch": 0.9094685024228906,
+      "learning_rate": 6.026024680947379e-05,
+      "loss": 0.6561,
+      "step": 11824
+    },
+    {
+      "epoch": 0.909545419583109,
+      "learning_rate": 6.0158583679096547e-05,
+      "loss": 1.6584,
+      "step": 11825
+    },
+    {
+      "epoch": 0.9096223367433275,
+      "learning_rate": 6.005700462255986e-05,
+      "loss": 1.0234,
+      "step": 11826
+    },
+    {
+      "epoch": 0.9096992539035459,
+      "learning_rate": 5.995550964579549e-05,
+      "loss": 0.9035,
+      "step": 11827
+    },
+    {
+      "epoch": 0.9097761710637643,
+      "learning_rate": 5.985409875472986e-05,
+      "loss": 1.1757,
+      "step": 11828
+    },
+    {
+      "epoch": 0.9098530882239828,
+      "learning_rate": 5.975277195528406e-05,
+      "loss": 1.0951,
+      "step": 11829
+    },
+    {
+      "epoch": 0.9099300053842012,
+      "learning_rate": 5.965152925337503e-05,
+      "loss": 1.3039,
+      "step": 11830
+    },
+    {
+      "epoch": 0.9100069225444196,
+      "learning_rate": 5.9550370654914195e-05,
+      "loss": 1.0342,
+      "step": 11831
+    },
+    {
+      "epoch": 0.9100838397046381,
+      "learning_rate": 5.944929616580852e-05,
+      "loss": 1.0479,
+      "step": 11832
+    },
+    {
+      "epoch": 0.9101607568648565,
+      "learning_rate": 5.934830579195977e-05,
+      "loss": 0.9659,
+      "step": 11833
+    },
+    {
+      "epoch": 0.9102376740250749,
+      "learning_rate": 5.924739953926472e-05,
+      "loss": 1.0638,
+      "step": 11834
+    },
+    {
+      "epoch": 0.9103145911852935,
+      "learning_rate": 5.914657741361551e-05,
+      "loss": 1.254,
+      "step": 11835
+    },
+    {
+      "epoch": 0.9103915083455119,
+      "learning_rate": 5.904583942089925e-05,
+      "loss": 1.3537,
+      "step": 11836
+    },
+    {
+      "epoch": 0.9104684255057304,
+      "learning_rate": 5.8945185566998405e-05,
+      "loss": 1.1721,
+      "step": 11837
+    },
+    {
+      "epoch": 0.9105453426659488,
+      "learning_rate": 5.8844615857789774e-05,
+      "loss": 1.312,
+      "step": 11838
+    },
+    {
+      "epoch": 0.9106222598261672,
+      "learning_rate": 5.874413029914616e-05,
+      "loss": 0.9252,
+      "step": 11839
+    },
+    {
+      "epoch": 0.9106991769863857,
+      "learning_rate": 5.864372889693454e-05,
+      "loss": 1.2886,
+      "step": 11840
+    },
+    {
+      "epoch": 0.9107760941466041,
+      "learning_rate": 5.854341165701804e-05,
+      "loss": 1.248,
+      "step": 11841
+    },
+    {
+      "epoch": 0.9108530113068225,
+      "learning_rate": 5.844317858525383e-05,
+      "loss": 1.3042,
+      "step": 11842
+    },
+    {
+      "epoch": 0.910929928467041,
+      "learning_rate": 5.834302968749472e-05,
+      "loss": 1.1956,
+      "step": 11843
+    },
+    {
+      "epoch": 0.9110068456272594,
+      "learning_rate": 5.824296496958853e-05,
+      "loss": 1.3257,
+      "step": 11844
+    },
+    {
+      "epoch": 0.9110837627874779,
+      "learning_rate": 5.8142984437378424e-05,
+      "loss": 1.1969,
+      "step": 11845
+    },
+    {
+      "epoch": 0.9111606799476963,
+      "learning_rate": 5.804308809670189e-05,
+      "loss": 1.0234,
+      "step": 11846
+    },
+    {
+      "epoch": 0.9112375971079147,
+      "learning_rate": 5.794327595339227e-05,
+      "loss": 1.3848,
+      "step": 11847
+    },
+    {
+      "epoch": 0.9113145142681333,
+      "learning_rate": 5.784354801327757e-05,
+      "loss": 1.2557,
+      "step": 11848
+    },
+    {
+      "epoch": 0.9113914314283517,
+      "learning_rate": 5.7743904282181293e-05,
+      "loss": 1.4429,
+      "step": 11849
+    },
+    {
+      "epoch": 0.9114683485885701,
+      "learning_rate": 5.764434476592162e-05,
+      "loss": 1.1427,
+      "step": 11850
+    },
+    {
+      "epoch": 0.9115452657487886,
+      "learning_rate": 5.75448694703114e-05,
+      "loss": 0.9271,
+      "step": 11851
+    },
+    {
+      "epoch": 0.911622182909007,
+      "learning_rate": 5.744547840115982e-05,
+      "loss": 1.1305,
+      "step": 11852
+    },
+    {
+      "epoch": 0.9116991000692255,
+      "learning_rate": 5.7346171564269914e-05,
+      "loss": 1.236,
+      "step": 11853
+    },
+    {
+      "epoch": 0.9117760172294439,
+      "learning_rate": 5.7246948965441025e-05,
+      "loss": 1.1155,
+      "step": 11854
+    },
+    {
+      "epoch": 0.9118529343896623,
+      "learning_rate": 5.714781061046603e-05,
+      "loss": 1.3916,
+      "step": 11855
+    },
+    {
+      "epoch": 0.9119298515498808,
+      "learning_rate": 5.704875650513414e-05,
+      "loss": 1.1753,
+      "step": 11856
+    },
+    {
+      "epoch": 0.9120067687100992,
+      "learning_rate": 5.694978665522921e-05,
+      "loss": 1.5762,
+      "step": 11857
+    },
+    {
+      "epoch": 0.9120836858703176,
+      "learning_rate": 5.6850901066530126e-05,
+      "loss": 1.3602,
+      "step": 11858
+    },
+    {
+      "epoch": 0.9121606030305361,
+      "learning_rate": 5.6752099744810937e-05,
+      "loss": 1.1882,
+      "step": 11859
+    },
+    {
+      "epoch": 0.9122375201907545,
+      "learning_rate": 5.6653382695840696e-05,
+      "loss": 1.444,
+      "step": 11860
+    },
+    {
+      "epoch": 0.912314437350973,
+      "learning_rate": 5.655474992538378e-05,
+      "loss": 0.9654,
+      "step": 11861
+    },
+    {
+      "epoch": 0.9123913545111915,
+      "learning_rate": 5.6456201439199105e-05,
+      "loss": 1.1835,
+      "step": 11862
+    },
+    {
+      "epoch": 0.9124682716714099,
+      "learning_rate": 5.635773724304172e-05,
+      "loss": 1.0858,
+      "step": 11863
+    },
+    {
+      "epoch": 0.9125451888316284,
+      "learning_rate": 5.6259357342660024e-05,
+      "loss": 0.9491,
+      "step": 11864
+    },
+    {
+      "epoch": 0.9126221059918468,
+      "learning_rate": 5.6161061743799424e-05,
+      "loss": 1.1276,
+      "step": 11865
+    },
+    {
+      "epoch": 0.9126990231520652,
+      "learning_rate": 5.606285045219917e-05,
+      "loss": 1.0894,
+      "step": 11866
+    },
+    {
+      "epoch": 0.9127759403122837,
+      "learning_rate": 5.5964723473593826e-05,
+      "loss": 1.0319,
+      "step": 11867
+    },
+    {
+      "epoch": 0.9128528574725021,
+      "learning_rate": 5.5866680813713324e-05,
+      "loss": 1.1197,
+      "step": 11868
+    },
+    {
+      "epoch": 0.9129297746327205,
+      "learning_rate": 5.576872247828224e-05,
+      "loss": 1.1269,
+      "step": 11869
+    },
+    {
+      "epoch": 0.913006691792939,
+      "learning_rate": 5.567084847302051e-05,
+      "loss": 1.0615,
+      "step": 11870
+    },
+    {
+      "epoch": 0.9130836089531574,
+      "learning_rate": 5.557305880364322e-05,
+      "loss": 1.0588,
+      "step": 11871
+    },
+    {
+      "epoch": 0.913160526113376,
+      "learning_rate": 5.5475353475860466e-05,
+      "loss": 1.1178,
+      "step": 11872
+    },
+    {
+      "epoch": 0.9132374432735944,
+      "learning_rate": 5.5377732495377364e-05,
+      "loss": 1.5532,
+      "step": 11873
+    },
+    {
+      "epoch": 0.9133143604338128,
+      "learning_rate": 5.5280195867893843e-05,
+      "loss": 1.2373,
+      "step": 11874
+    },
+    {
+      "epoch": 0.9133912775940313,
+      "learning_rate": 5.518274359910519e-05,
+      "loss": 1.2619,
+      "step": 11875
+    },
+    {
+      "epoch": 0.9134681947542497,
+      "learning_rate": 5.508537569470234e-05,
+      "loss": 1.2821,
+      "step": 11876
+    },
+    {
+      "epoch": 0.9135451119144681,
+      "learning_rate": 5.49880921603701e-05,
+      "loss": 1.1162,
+      "step": 11877
+    },
+    {
+      "epoch": 0.9136220290746866,
+      "learning_rate": 5.489089300178873e-05,
+      "loss": 0.8705,
+      "step": 11878
+    },
+    {
+      "epoch": 0.913698946234905,
+      "learning_rate": 5.479377822463455e-05,
+      "loss": 1.3552,
+      "step": 11879
+    },
+    {
+      "epoch": 0.9137758633951234,
+      "learning_rate": 5.4696747834577846e-05,
+      "loss": 0.9738,
+      "step": 11880
+    },
+    {
+      "epoch": 0.9138527805553419,
+      "learning_rate": 5.459980183728408e-05,
+      "loss": 1.042,
+      "step": 11881
+    },
+    {
+      "epoch": 0.9139296977155603,
+      "learning_rate": 5.45029402384144e-05,
+      "loss": 1.2623,
+      "step": 11882
+    },
+    {
+      "epoch": 0.9140066148757788,
+      "learning_rate": 5.440616304362428e-05,
+      "loss": 1.2096,
+      "step": 11883
+    },
+    {
+      "epoch": 0.9140835320359972,
+      "learning_rate": 5.430947025856503e-05,
+      "loss": 1.2831,
+      "step": 11884
+    },
+    {
+      "epoch": 0.9141604491962156,
+      "learning_rate": 5.42128618888828e-05,
+      "loss": 1.2196,
+      "step": 11885
+    },
+    {
+      "epoch": 0.9142373663564342,
+      "learning_rate": 5.4116337940217756e-05,
+      "loss": 0.8838,
+      "step": 11886
+    },
+    {
+      "epoch": 0.9143142835166526,
+      "learning_rate": 5.401989841820687e-05,
+      "loss": 1.0429,
+      "step": 11887
+    },
+    {
+      "epoch": 0.914391200676871,
+      "learning_rate": 5.392354332848115e-05,
+      "loss": 1.365,
+      "step": 11888
+    },
+    {
+      "epoch": 0.9144681178370895,
+      "learning_rate": 5.382727267666676e-05,
+      "loss": 0.9185,
+      "step": 11889
+    },
+    {
+      "epoch": 0.9145450349973079,
+      "learning_rate": 5.3731086468385035e-05,
+      "loss": 1.2586,
+      "step": 11890
+    },
+    {
+      "epoch": 0.9146219521575264,
+      "learning_rate": 5.3634984709252156e-05,
+      "loss": 1.1925,
+      "step": 11891
+    },
+    {
+      "epoch": 0.9146988693177448,
+      "learning_rate": 5.353896740488029e-05,
+      "loss": 1.3063,
+      "step": 11892
+    },
+    {
+      "epoch": 0.9147757864779632,
+      "learning_rate": 5.344303456087546e-05,
+      "loss": 1.0249,
+      "step": 11893
+    },
+    {
+      "epoch": 0.9148527036381817,
+      "learning_rate": 5.3347186182839344e-05,
+      "loss": 1.1116,
+      "step": 11894
+    },
+    {
+      "epoch": 0.9149296207984001,
+      "learning_rate": 5.325142227636881e-05,
+      "loss": 1.2574,
+      "step": 11895
+    },
+    {
+      "epoch": 0.9150065379586185,
+      "learning_rate": 5.3155742847055544e-05,
+      "loss": 0.8567,
+      "step": 11896
+    },
+    {
+      "epoch": 0.915083455118837,
+      "learning_rate": 5.306014790048608e-05,
+      "loss": 1.3659,
+      "step": 11897
+    },
+    {
+      "epoch": 0.9151603722790554,
+      "learning_rate": 5.296463744224278e-05,
+      "loss": 1.1399,
+      "step": 11898
+    },
+    {
+      "epoch": 0.9152372894392738,
+      "learning_rate": 5.2869211477902354e-05,
+      "loss": 1.3039,
+      "step": 11899
+    },
+    {
+      "epoch": 0.9153142065994924,
+      "learning_rate": 5.277387001303652e-05,
+      "loss": 0.8714,
+      "step": 11900
+    },
+    {
+      "epoch": 0.9153911237597108,
+      "learning_rate": 5.267861305321298e-05,
+      "loss": 1.0065,
+      "step": 11901
+    },
+    {
+      "epoch": 0.9154680409199293,
+      "learning_rate": 5.258344060399345e-05,
+      "loss": 1.0591,
+      "step": 11902
+    },
+    {
+      "epoch": 0.9155449580801477,
+      "learning_rate": 5.2488352670935325e-05,
+      "loss": 1.2966,
+      "step": 11903
+    },
+    {
+      "epoch": 0.9156218752403661,
+      "learning_rate": 5.239334925959099e-05,
+      "loss": 1.0625,
+      "step": 11904
+    },
+    {
+      "epoch": 0.9156987924005846,
+      "learning_rate": 5.2298430375507354e-05,
+      "loss": 1.1937,
+      "step": 11905
+    },
+    {
+      "epoch": 0.915775709560803,
+      "learning_rate": 5.220359602422731e-05,
+      "loss": 0.8867,
+      "step": 11906
+    },
+    {
+      "epoch": 0.9158526267210214,
+      "learning_rate": 5.210884621128825e-05,
+      "loss": 1.236,
+      "step": 11907
+    },
+    {
+      "epoch": 0.9159295438812399,
+      "learning_rate": 5.2014180942222435e-05,
+      "loss": 1.1658,
+      "step": 11908
+    },
+    {
+      "epoch": 0.9160064610414583,
+      "learning_rate": 5.191960022255776e-05,
+      "loss": 1.125,
+      "step": 11909
+    },
+    {
+      "epoch": 0.9160833782016768,
+      "learning_rate": 5.182510405781649e-05,
+      "loss": 1.4948,
+      "step": 11910
+    },
+    {
+      "epoch": 0.9161602953618952,
+      "learning_rate": 5.173069245351719e-05,
+      "loss": 1.0294,
+      "step": 11911
+    },
+    {
+      "epoch": 0.9162372125221137,
+      "learning_rate": 5.163636541517181e-05,
+      "loss": 1.3724,
+      "step": 11912
+    },
+    {
+      "epoch": 0.9163141296823322,
+      "learning_rate": 5.154212294828825e-05,
+      "loss": 0.9865,
+      "step": 11913
+    },
+    {
+      "epoch": 0.9163910468425506,
+      "learning_rate": 5.144796505836996e-05,
+      "loss": 1.3347,
+      "step": 11914
+    },
+    {
+      "epoch": 0.916467964002769,
+      "learning_rate": 5.13538917509147e-05,
+      "loss": 1.3131,
+      "step": 11915
+    },
+    {
+      "epoch": 0.9165448811629875,
+      "learning_rate": 5.1259903031415254e-05,
+      "loss": 1.2446,
+      "step": 11916
+    },
+    {
+      "epoch": 0.9166217983232059,
+      "learning_rate": 5.1165998905360055e-05,
+      "loss": 1.2039,
+      "step": 11917
+    },
+    {
+      "epoch": 0.9166987154834243,
+      "learning_rate": 5.107217937823205e-05,
+      "loss": 1.0967,
+      "step": 11918
+    },
+    {
+      "epoch": 0.9167756326436428,
+      "learning_rate": 5.097844445550953e-05,
+      "loss": 1.3681,
+      "step": 11919
+    },
+    {
+      "epoch": 0.9168525498038612,
+      "learning_rate": 5.0884794142665944e-05,
+      "loss": 0.7595,
+      "step": 11920
+    },
+    {
+      "epoch": 0.9169294669640797,
+      "learning_rate": 5.079122844516909e-05,
+      "loss": 1.2948,
+      "step": 11921
+    },
+    {
+      "epoch": 0.9170063841242981,
+      "learning_rate": 5.069774736848309e-05,
+      "loss": 1.2925,
+      "step": 11922
+    },
+    {
+      "epoch": 0.9170833012845165,
+      "learning_rate": 5.060435091806592e-05,
+      "loss": 0.9102,
+      "step": 11923
+    },
+    {
+      "epoch": 0.917160218444735,
+      "learning_rate": 5.051103909937121e-05,
+      "loss": 1.0557,
+      "step": 11924
+    },
+    {
+      "epoch": 0.9172371356049535,
+      "learning_rate": 5.041781191784761e-05,
+      "loss": 1.4557,
+      "step": 11925
+    },
+    {
+      "epoch": 0.9173140527651719,
+      "learning_rate": 5.032466937893859e-05,
+      "loss": 0.9919,
+      "step": 11926
+    },
+    {
+      "epoch": 0.9173909699253904,
+      "learning_rate": 5.0231611488082806e-05,
+      "loss": 1.1075,
+      "step": 11927
+    },
+    {
+      "epoch": 0.9174678870856088,
+      "learning_rate": 5.0138638250714406e-05,
+      "loss": 1.0142,
+      "step": 11928
+    },
+    {
+      "epoch": 0.9175448042458273,
+      "learning_rate": 5.004574967226172e-05,
+      "loss": 0.9889,
+      "step": 11929
+    },
+    {
+      "epoch": 0.9176217214060457,
+      "learning_rate": 4.995294575814907e-05,
+      "loss": 0.9621,
+      "step": 11930
+    },
+    {
+      "epoch": 0.9176986385662641,
+      "learning_rate": 4.986022651379496e-05,
+      "loss": 1.2588,
+      "step": 11931
+    },
+    {
+      "epoch": 0.9177755557264826,
+      "learning_rate": 4.976759194461339e-05,
+      "loss": 0.7398,
+      "step": 11932
+    },
+    {
+      "epoch": 0.917852472886701,
+      "learning_rate": 4.9675042056013695e-05,
+      "loss": 1.2111,
+      "step": 11933
+    },
+    {
+      "epoch": 0.9179293900469194,
+      "learning_rate": 4.9582576853399894e-05,
+      "loss": 1.2539,
+      "step": 11934
+    },
+    {
+      "epoch": 0.9180063072071379,
+      "learning_rate": 4.9490196342170836e-05,
+      "loss": 1.0612,
+      "step": 11935
+    },
+    {
+      "epoch": 0.9180832243673563,
+      "learning_rate": 4.939790052772103e-05,
+      "loss": 1.5774,
+      "step": 11936
+    },
+    {
+      "epoch": 0.9181601415275747,
+      "learning_rate": 4.93056894154395e-05,
+      "loss": 0.9043,
+      "step": 11937
+    },
+    {
+      "epoch": 0.9182370586877933,
+      "learning_rate": 4.9213563010710614e-05,
+      "loss": 1.3238,
+      "step": 11938
+    },
+    {
+      "epoch": 0.9183139758480117,
+      "learning_rate": 4.91215213189139e-05,
+      "loss": 1.0755,
+      "step": 11939
+    },
+    {
+      "epoch": 0.9183908930082302,
+      "learning_rate": 4.902956434542338e-05,
+      "loss": 1.2284,
+      "step": 11940
+    },
+    {
+      "epoch": 0.9184678101684486,
+      "learning_rate": 4.893769209560894e-05,
+      "loss": 0.9964,
+      "step": 11941
+    },
+    {
+      "epoch": 0.918544727328667,
+      "learning_rate": 4.8845904574834956e-05,
+      "loss": 1.2363,
+      "step": 11942
+    },
+    {
+      "epoch": 0.9186216444888855,
+      "learning_rate": 4.875420178846079e-05,
+      "loss": 1.29,
+      "step": 11943
+    },
+    {
+      "epoch": 0.9186985616491039,
+      "learning_rate": 4.8662583741841515e-05,
+      "loss": 0.6637,
+      "step": 11944
+    },
+    {
+      "epoch": 0.9187754788093223,
+      "learning_rate": 4.8571050440326333e-05,
+      "loss": 0.9039,
+      "step": 11945
+    },
+    {
+      "epoch": 0.9188523959695408,
+      "learning_rate": 4.8479601889259984e-05,
+      "loss": 1.3058,
+      "step": 11946
+    },
+    {
+      "epoch": 0.9189293131297592,
+      "learning_rate": 4.838823809398285e-05,
+      "loss": 1.5151,
+      "step": 11947
+    },
+    {
+      "epoch": 0.9190062302899777,
+      "learning_rate": 4.829695905982884e-05,
+      "loss": 0.7713,
+      "step": 11948
+    },
+    {
+      "epoch": 0.9190831474501961,
+      "learning_rate": 4.820576479212868e-05,
+      "loss": 0.9511,
+      "step": 11949
+    },
+    {
+      "epoch": 0.9191600646104146,
+      "learning_rate": 4.811465529620679e-05,
+      "loss": 1.5938,
+      "step": 11950
+    },
+    {
+      "epoch": 0.9192369817706331,
+      "learning_rate": 4.8023630577383404e-05,
+      "loss": 0.6984,
+      "step": 11951
+    },
+    {
+      "epoch": 0.9193138989308515,
+      "learning_rate": 4.793269064097344e-05,
+      "loss": 1.1737,
+      "step": 11952
+    },
+    {
+      "epoch": 0.9193908160910699,
+      "learning_rate": 4.7841835492286823e-05,
+      "loss": 1.3361,
+      "step": 11953
+    },
+    {
+      "epoch": 0.9194677332512884,
+      "learning_rate": 4.77510651366288e-05,
+      "loss": 1.1423,
+      "step": 11954
+    },
+    {
+      "epoch": 0.9195446504115068,
+      "learning_rate": 4.76603795792998e-05,
+      "loss": 1.0514,
+      "step": 11955
+    },
+    {
+      "epoch": 0.9196215675717253,
+      "learning_rate": 4.7569778825594604e-05,
+      "loss": 0.7971,
+      "step": 11956
+    },
+    {
+      "epoch": 0.9196984847319437,
+      "learning_rate": 4.7479262880803955e-05,
+      "loss": 1.0356,
+      "step": 11957
+    },
+    {
+      "epoch": 0.9197754018921621,
+      "learning_rate": 4.73888317502128e-05,
+      "loss": 1.0134,
+      "step": 11958
+    },
+    {
+      "epoch": 0.9198523190523806,
+      "learning_rate": 4.7298485439101426e-05,
+      "loss": 1.2458,
+      "step": 11959
+    },
+    {
+      "epoch": 0.919929236212599,
+      "learning_rate": 4.7208223952745935e-05,
+      "loss": 1.1579,
+      "step": 11960
+    },
+    {
+      "epoch": 0.9200061533728174,
+      "learning_rate": 4.711804729641611e-05,
+      "loss": 1.464,
+      "step": 11961
+    },
+    {
+      "epoch": 0.920083070533036,
+      "learning_rate": 4.7027955475377416e-05,
+      "loss": 1.223,
+      "step": 11962
+    },
+    {
+      "epoch": 0.9201599876932544,
+      "learning_rate": 4.6937948494890805e-05,
+      "loss": 1.3812,
+      "step": 11963
+    },
+    {
+      "epoch": 0.9202369048534728,
+      "learning_rate": 4.684802636021174e-05,
+      "loss": 1.089,
+      "step": 11964
+    },
+    {
+      "epoch": 0.9203138220136913,
+      "learning_rate": 4.6758189076591015e-05,
+      "loss": 1.0512,
+      "step": 11965
+    },
+    {
+      "epoch": 0.9203907391739097,
+      "learning_rate": 4.666843664927395e-05,
+      "loss": 0.7737,
+      "step": 11966
+    },
+    {
+      "epoch": 0.9204676563341282,
+      "learning_rate": 4.65787690835015e-05,
+      "loss": 0.8502,
+      "step": 11967
+    },
+    {
+      "epoch": 0.9205445734943466,
+      "learning_rate": 4.64891863845095e-05,
+      "loss": 1.0651,
+      "step": 11968
+    },
+    {
+      "epoch": 0.920621490654565,
+      "learning_rate": 4.6399688557528907e-05,
+      "loss": 1.2638,
+      "step": 11969
+    },
+    {
+      "epoch": 0.9206984078147835,
+      "learning_rate": 4.631027560778506e-05,
+      "loss": 1.1519,
+      "step": 11970
+    },
+    {
+      "epoch": 0.9207753249750019,
+      "learning_rate": 4.6220947540499433e-05,
+      "loss": 1.3492,
+      "step": 11971
+    },
+    {
+      "epoch": 0.9208522421352203,
+      "learning_rate": 4.6131704360887706e-05,
+      "loss": 1.1754,
+      "step": 11972
+    },
+    {
+      "epoch": 0.9209291592954388,
+      "learning_rate": 4.604254607416103e-05,
+      "loss": 1.3276,
+      "step": 11973
+    },
+    {
+      "epoch": 0.9210060764556572,
+      "learning_rate": 4.595347268552524e-05,
+      "loss": 1.0822,
+      "step": 11974
+    },
+    {
+      "epoch": 0.9210829936158758,
+      "learning_rate": 4.586448420018152e-05,
+      "loss": 1.2038,
+      "step": 11975
+    },
+    {
+      "epoch": 0.9211599107760942,
+      "learning_rate": 4.577558062332604e-05,
+      "loss": 1.0735,
+      "step": 11976
+    },
+    {
+      "epoch": 0.9212368279363126,
+      "learning_rate": 4.568676196015015e-05,
+      "loss": 0.9305,
+      "step": 11977
+    },
+    {
+      "epoch": 0.9213137450965311,
+      "learning_rate": 4.559802821583969e-05,
+      "loss": 1.2265,
+      "step": 11978
+    },
+    {
+      "epoch": 0.9213906622567495,
+      "learning_rate": 4.55093793955762e-05,
+      "loss": 1.3947,
+      "step": 11979
+    },
+    {
+      "epoch": 0.9214675794169679,
+      "learning_rate": 4.542081550453603e-05,
+      "loss": 1.1411,
+      "step": 11980
+    },
+    {
+      "epoch": 0.9215444965771864,
+      "learning_rate": 4.533233654789004e-05,
+      "loss": 1.0006,
+      "step": 11981
+    },
+    {
+      "epoch": 0.9216214137374048,
+      "learning_rate": 4.524394253080527e-05,
+      "loss": 0.8626,
+      "step": 11982
+    },
+    {
+      "epoch": 0.9216983308976232,
+      "learning_rate": 4.515563345844259e-05,
+      "loss": 0.9308,
+      "step": 11983
+    },
+    {
+      "epoch": 0.9217752480578417,
+      "learning_rate": 4.506740933595854e-05,
+      "loss": 0.9866,
+      "step": 11984
+    },
+    {
+      "epoch": 0.9218521652180601,
+      "learning_rate": 4.4979270168505004e-05,
+      "loss": 1.2695,
+      "step": 11985
+    },
+    {
+      "epoch": 0.9219290823782786,
+      "learning_rate": 4.489121596122819e-05,
+      "loss": 1.0283,
+      "step": 11986
+    },
+    {
+      "epoch": 0.922005999538497,
+      "learning_rate": 4.480324671926983e-05,
+      "loss": 1.3385,
+      "step": 11987
+    },
+    {
+      "epoch": 0.9220829166987154,
+      "learning_rate": 4.471536244776647e-05,
+      "loss": 0.9881,
+      "step": 11988
+    },
+    {
+      "epoch": 0.922159833858934,
+      "learning_rate": 4.462756315184951e-05,
+      "loss": 1.2224,
+      "step": 11989
+    },
+    {
+      "epoch": 0.9222367510191524,
+      "learning_rate": 4.453984883664603e-05,
+      "loss": 1.3394,
+      "step": 11990
+    },
+    {
+      "epoch": 0.9223136681793708,
+      "learning_rate": 4.44522195072779e-05,
+      "loss": 1.0757,
+      "step": 11991
+    },
+    {
+      "epoch": 0.9223905853395893,
+      "learning_rate": 4.4364675168861226e-05,
+      "loss": 1.2237,
+      "step": 11992
+    },
+    {
+      "epoch": 0.9224675024998077,
+      "learning_rate": 4.42772158265084e-05,
+      "loss": 0.7563,
+      "step": 11993
+    },
+    {
+      "epoch": 0.9225444196600262,
+      "learning_rate": 4.4189841485325844e-05,
+      "loss": 1.928,
+      "step": 11994
+    },
+    {
+      "epoch": 0.9226213368202446,
+      "learning_rate": 4.410255215041598e-05,
+      "loss": 1.156,
+      "step": 11995
+    },
+    {
+      "epoch": 0.922698253980463,
+      "learning_rate": 4.401534782687522e-05,
+      "loss": 1.1637,
+      "step": 11996
+    },
+    {
+      "epoch": 0.9227751711406815,
+      "learning_rate": 4.39282285197955e-05,
+      "loss": 0.9752,
+      "step": 11997
+    },
+    {
+      "epoch": 0.9228520883008999,
+      "learning_rate": 4.384119423426425e-05,
+      "loss": 1.1092,
+      "step": 11998
+    },
+    {
+      "epoch": 0.9229290054611183,
+      "learning_rate": 4.375424497536323e-05,
+      "loss": 1.0258,
+      "step": 11999
+    },
+    {
+      "epoch": 0.9230059226213368,
+      "learning_rate": 4.3667380748169396e-05,
+      "loss": 1.3356,
+      "step": 12000
+    },
+    {
+      "epoch": 0.9230828397815553,
+      "learning_rate": 4.3580601557755004e-05,
+      "loss": 0.8829,
+      "step": 12001
+    },
+    {
+      "epoch": 0.9231597569417737,
+      "learning_rate": 4.349390740918718e-05,
+      "loss": 1.0759,
+      "step": 12002
+    },
+    {
+      "epoch": 0.9232366741019922,
+      "learning_rate": 4.3407298307527876e-05,
+      "loss": 1.1812,
+      "step": 12003
+    },
+    {
+      "epoch": 0.9233135912622106,
+      "learning_rate": 4.3320774257834704e-05,
+      "loss": 1.2493,
+      "step": 12004
+    },
+    {
+      "epoch": 0.9233905084224291,
+      "learning_rate": 4.3234335265159466e-05,
+      "loss": 0.9774,
+      "step": 12005
+    },
+    {
+      "epoch": 0.9234674255826475,
+      "learning_rate": 4.314798133454961e-05,
+      "loss": 0.9717,
+      "step": 12006
+    },
+    {
+      "epoch": 0.9235443427428659,
+      "learning_rate": 4.306171247104746e-05,
+      "loss": 1.1435,
+      "step": 12007
+    },
+    {
+      "epoch": 0.9236212599030844,
+      "learning_rate": 4.297552867969029e-05,
+      "loss": 1.2968,
+      "step": 12008
+    },
+    {
+      "epoch": 0.9236981770633028,
+      "learning_rate": 4.28894299655106e-05,
+      "loss": 1.1827,
+      "step": 12009
+    },
+    {
+      "epoch": 0.9237750942235212,
+      "learning_rate": 4.2803416333535514e-05,
+      "loss": 1.2274,
+      "step": 12010
+    },
+    {
+      "epoch": 0.9238520113837397,
+      "learning_rate": 4.2717487788787536e-05,
+      "loss": 1.2271,
+      "step": 12011
+    },
+    {
+      "epoch": 0.9239289285439581,
+      "learning_rate": 4.2631644336284136e-05,
+      "loss": 1.1826,
+      "step": 12012
+    },
+    {
+      "epoch": 0.9240058457041767,
+      "learning_rate": 4.254588598103815e-05,
+      "loss": 1.0768,
+      "step": 12013
+    },
+    {
+      "epoch": 0.9240827628643951,
+      "learning_rate": 4.2460212728056555e-05,
+      "loss": 0.9962,
+      "step": 12014
+    },
+    {
+      "epoch": 0.9241596800246135,
+      "learning_rate": 4.2374624582342366e-05,
+      "loss": 1.0629,
+      "step": 12015
+    },
+    {
+      "epoch": 0.924236597184832,
+      "learning_rate": 4.228912154889258e-05,
+      "loss": 0.6099,
+      "step": 12016
+    },
+    {
+      "epoch": 0.9243135143450504,
+      "learning_rate": 4.2203703632700705e-05,
+      "loss": 0.9834,
+      "step": 12017
+    },
+    {
+      "epoch": 0.9243904315052688,
+      "learning_rate": 4.2118370838753584e-05,
+      "loss": 0.7054,
+      "step": 12018
+    },
+    {
+      "epoch": 0.9244673486654873,
+      "learning_rate": 4.2033123172033895e-05,
+      "loss": 1.1603,
+      "step": 12019
+    },
+    {
+      "epoch": 0.9245442658257057,
+      "learning_rate": 4.194796063751999e-05,
+      "loss": 0.8002,
+      "step": 12020
+    },
+    {
+      "epoch": 0.9246211829859241,
+      "learning_rate": 4.1862883240184055e-05,
+      "loss": 1.2631,
+      "step": 12021
+    },
+    {
+      "epoch": 0.9246981001461426,
+      "learning_rate": 4.177789098499396e-05,
+      "loss": 1.0248,
+      "step": 12022
+    },
+    {
+      "epoch": 0.924775017306361,
+      "learning_rate": 4.169298387691273e-05,
+      "loss": 0.9804,
+      "step": 12023
+    },
+    {
+      "epoch": 0.9248519344665795,
+      "learning_rate": 4.160816192089772e-05,
+      "loss": 1.2527,
+      "step": 12024
+    },
+    {
+      "epoch": 0.9249288516267979,
+      "learning_rate": 4.152342512190216e-05,
+      "loss": 0.9706,
+      "step": 12025
+    },
+    {
+      "epoch": 0.9250057687870163,
+      "learning_rate": 4.143877348487391e-05,
+      "loss": 1.1578,
+      "step": 12026
+    },
+    {
+      "epoch": 0.9250826859472349,
+      "learning_rate": 4.135420701475551e-05,
+      "loss": 0.9385,
+      "step": 12027
+    },
+    {
+      "epoch": 0.9251596031074533,
+      "learning_rate": 4.1269725716485184e-05,
+      "loss": 1.1869,
+      "step": 12028
+    },
+    {
+      "epoch": 0.9252365202676717,
+      "learning_rate": 4.118532959499599e-05,
+      "loss": 1.2315,
+      "step": 12029
+    },
+    {
+      "epoch": 0.9253134374278902,
+      "learning_rate": 4.110101865521565e-05,
+      "loss": 1.4298,
+      "step": 12030
+    },
+    {
+      "epoch": 0.9253903545881086,
+      "learning_rate": 4.101679290206722e-05,
+      "loss": 1.627,
+      "step": 12031
+    },
+    {
+      "epoch": 0.9254672717483271,
+      "learning_rate": 4.093265234046878e-05,
+      "loss": 0.9546,
+      "step": 12032
+    },
+    {
+      "epoch": 0.9255441889085455,
+      "learning_rate": 4.084859697533338e-05,
+      "loss": 1.1561,
+      "step": 12033
+    },
+    {
+      "epoch": 0.9256211060687639,
+      "learning_rate": 4.076462681156912e-05,
+      "loss": 0.9854,
+      "step": 12034
+    },
+    {
+      "epoch": 0.9256980232289824,
+      "learning_rate": 4.068074185407905e-05,
+      "loss": 0.8717,
+      "step": 12035
+    },
+    {
+      "epoch": 0.9257749403892008,
+      "learning_rate": 4.0596942107761435e-05,
+      "loss": 1.3097,
+      "step": 12036
+    },
+    {
+      "epoch": 0.9258518575494192,
+      "learning_rate": 4.0513227577509195e-05,
+      "loss": 1.1436,
+      "step": 12037
+    },
+    {
+      "epoch": 0.9259287747096377,
+      "learning_rate": 4.0429598268210575e-05,
+      "loss": 1.1507,
+      "step": 12038
+    },
+    {
+      "epoch": 0.9260056918698562,
+      "learning_rate": 4.0346054184749016e-05,
+      "loss": 1.0428,
+      "step": 12039
+    },
+    {
+      "epoch": 0.9260826090300746,
+      "learning_rate": 4.02625953320026e-05,
+      "loss": 1.1619,
+      "step": 12040
+    },
+    {
+      "epoch": 0.9261595261902931,
+      "learning_rate": 4.017922171484428e-05,
+      "loss": 1.3877,
+      "step": 12041
+    },
+    {
+      "epoch": 0.9262364433505115,
+      "learning_rate": 4.009593333814282e-05,
+      "loss": 1.326,
+      "step": 12042
+    },
+    {
+      "epoch": 0.92631336051073,
+      "learning_rate": 4.001273020676116e-05,
+      "loss": 1.2347,
+      "step": 12043
+    },
+    {
+      "epoch": 0.9263902776709484,
+      "learning_rate": 3.992961232555792e-05,
+      "loss": 1.2249,
+      "step": 12044
+    },
+    {
+      "epoch": 0.9264671948311668,
+      "learning_rate": 3.9846579699386046e-05,
+      "loss": 1.1475,
+      "step": 12045
+    },
+    {
+      "epoch": 0.9265441119913853,
+      "learning_rate": 3.9763632333094156e-05,
+      "loss": 1.2705,
+      "step": 12046
+    },
+    {
+      "epoch": 0.9266210291516037,
+      "learning_rate": 3.968077023152572e-05,
+      "loss": 1.0513,
+      "step": 12047
+    },
+    {
+      "epoch": 0.9266979463118221,
+      "learning_rate": 3.959799339951903e-05,
+      "loss": 1.1733,
+      "step": 12048
+    },
+    {
+      "epoch": 0.9267748634720406,
+      "learning_rate": 3.951530184190738e-05,
+      "loss": 1.2158,
+      "step": 12049
+    },
+    {
+      "epoch": 0.926851780632259,
+      "learning_rate": 3.943269556351942e-05,
+      "loss": 1.0553,
+      "step": 12050
+    },
+    {
+      "epoch": 0.9269286977924776,
+      "learning_rate": 3.9350174569178273e-05,
+      "loss": 1.2771,
+      "step": 12051
+    },
+    {
+      "epoch": 0.927005614952696,
+      "learning_rate": 3.9267738863703105e-05,
+      "loss": 1.1422,
+      "step": 12052
+    },
+    {
+      "epoch": 0.9270825321129144,
+      "learning_rate": 3.918538845190689e-05,
+      "loss": 0.8328,
+      "step": 12053
+    },
+    {
+      "epoch": 0.9271594492731329,
+      "learning_rate": 3.910312333859794e-05,
+      "loss": 1.0746,
+      "step": 12054
+    },
+    {
+      "epoch": 0.9272363664333513,
+      "learning_rate": 3.902094352858043e-05,
+      "loss": 1.0455,
+      "step": 12055
+    },
+    {
+      "epoch": 0.9273132835935697,
+      "learning_rate": 3.893884902665268e-05,
+      "loss": 1.1229,
+      "step": 12056
+    },
+    {
+      "epoch": 0.9273902007537882,
+      "learning_rate": 3.8856839837608015e-05,
+      "loss": 1.0004,
+      "step": 12057
+    },
+    {
+      "epoch": 0.9274671179140066,
+      "learning_rate": 3.877491596623545e-05,
+      "loss": 1.0069,
+      "step": 12058
+    },
+    {
+      "epoch": 0.9275440350742251,
+      "learning_rate": 3.8693077417318145e-05,
+      "loss": 0.9788,
+      "step": 12059
+    },
+    {
+      "epoch": 0.9276209522344435,
+      "learning_rate": 3.861132419563512e-05,
+      "loss": 1.388,
+      "step": 12060
+    },
+    {
+      "epoch": 0.9276978693946619,
+      "learning_rate": 3.852965630596006e-05,
+      "loss": 1.3543,
+      "step": 12061
+    },
+    {
+      "epoch": 0.9277747865548804,
+      "learning_rate": 3.844807375306114e-05,
+      "loss": 1.283,
+      "step": 12062
+    },
+    {
+      "epoch": 0.9278517037150988,
+      "learning_rate": 3.836657654170256e-05,
+      "loss": 1.1871,
+      "step": 12063
+    },
+    {
+      "epoch": 0.9279286208753172,
+      "learning_rate": 3.8285164676642836e-05,
+      "loss": 0.9242,
+      "step": 12064
+    },
+    {
+      "epoch": 0.9280055380355358,
+      "learning_rate": 3.820383816263584e-05,
+      "loss": 0.9492,
+      "step": 12065
+    },
+    {
+      "epoch": 0.9280824551957542,
+      "learning_rate": 3.81225970044301e-05,
+      "loss": 1.2078,
+      "step": 12066
+    },
+    {
+      "epoch": 0.9281593723559726,
+      "learning_rate": 3.804144120676933e-05,
+      "loss": 1.0509,
+      "step": 12067
+    },
+    {
+      "epoch": 0.9282362895161911,
+      "learning_rate": 3.7960370774392406e-05,
+      "loss": 0.9027,
+      "step": 12068
+    },
+    {
+      "epoch": 0.9283132066764095,
+      "learning_rate": 3.7879385712033196e-05,
+      "loss": 0.9869,
+      "step": 12069
+    },
+    {
+      "epoch": 0.928390123836628,
+      "learning_rate": 3.77984860244206e-05,
+      "loss": 0.9502,
+      "step": 12070
+    },
+    {
+      "epoch": 0.9284670409968464,
+      "learning_rate": 3.771767171627816e-05,
+      "loss": 1.239,
+      "step": 12071
+    },
+    {
+      "epoch": 0.9285439581570648,
+      "learning_rate": 3.763694279232477e-05,
+      "loss": 1.1828,
+      "step": 12072
+    },
+    {
+      "epoch": 0.9286208753172833,
+      "learning_rate": 3.755629925727433e-05,
+      "loss": 0.8113,
+      "step": 12073
+    },
+    {
+      "epoch": 0.9286977924775017,
+      "learning_rate": 3.7475741115835906e-05,
+      "loss": 1.0285,
+      "step": 12074
+    },
+    {
+      "epoch": 0.9287747096377201,
+      "learning_rate": 3.739526837271307e-05,
+      "loss": 1.148,
+      "step": 12075
+    },
+    {
+      "epoch": 0.9288516267979386,
+      "learning_rate": 3.7314881032604553e-05,
+      "loss": 1.3272,
+      "step": 12076
+    },
+    {
+      "epoch": 0.928928543958157,
+      "learning_rate": 3.723457910020478e-05,
+      "loss": 1.1356,
+      "step": 12077
+    },
+    {
+      "epoch": 0.9290054611183756,
+      "learning_rate": 3.715436258020233e-05,
+      "loss": 1.0096,
+      "step": 12078
+    },
+    {
+      "epoch": 0.929082378278594,
+      "learning_rate": 3.707423147728112e-05,
+      "loss": 1.2124,
+      "step": 12079
+    },
+    {
+      "epoch": 0.9291592954388124,
+      "learning_rate": 3.699418579612024e-05,
+      "loss": 0.8465,
+      "step": 12080
+    },
+    {
+      "epoch": 0.9292362125990309,
+      "learning_rate": 3.6914225541393456e-05,
+      "loss": 1.2877,
+      "step": 12081
+    },
+    {
+      "epoch": 0.9293131297592493,
+      "learning_rate": 3.683435071776986e-05,
+      "loss": 0.923,
+      "step": 12082
+    },
+    {
+      "epoch": 0.9293900469194677,
+      "learning_rate": 3.675456132991339e-05,
+      "loss": 1.2139,
+      "step": 12083
+    },
+    {
+      "epoch": 0.9294669640796862,
+      "learning_rate": 3.667485738248316e-05,
+      "loss": 1.0502,
+      "step": 12084
+    },
+    {
+      "epoch": 0.9295438812399046,
+      "learning_rate": 3.6595238880132944e-05,
+      "loss": 1.1556,
+      "step": 12085
+    },
+    {
+      "epoch": 0.929620798400123,
+      "learning_rate": 3.6515705827512016e-05,
+      "loss": 1.2585,
+      "step": 12086
+    },
+    {
+      "epoch": 0.9296977155603415,
+      "learning_rate": 3.6436258229263996e-05,
+      "loss": 0.7302,
+      "step": 12087
+    },
+    {
+      "epoch": 0.9297746327205599,
+      "learning_rate": 3.635689609002835e-05,
+      "loss": 1.4461,
+      "step": 12088
+    },
+    {
+      "epoch": 0.9298515498807784,
+      "learning_rate": 3.627761941443869e-05,
+      "loss": 1.3529,
+      "step": 12089
+    },
+    {
+      "epoch": 0.9299284670409969,
+      "learning_rate": 3.619842820712449e-05,
+      "loss": 1.0811,
+      "step": 12090
+    },
+    {
+      "epoch": 0.9300053842012153,
+      "learning_rate": 3.611932247270955e-05,
+      "loss": 1.2806,
+      "step": 12091
+    },
+    {
+      "epoch": 0.9300823013614338,
+      "learning_rate": 3.6040302215813004e-05,
+      "loss": 1.1022,
+      "step": 12092
+    },
+    {
+      "epoch": 0.9301592185216522,
+      "learning_rate": 3.596136744104883e-05,
+      "loss": 1.0145,
+      "step": 12093
+    },
+    {
+      "epoch": 0.9302361356818706,
+      "learning_rate": 3.5882518153026344e-05,
+      "loss": 1.3617,
+      "step": 12094
+    },
+    {
+      "epoch": 0.9303130528420891,
+      "learning_rate": 3.580375435634919e-05,
+      "loss": 1.2117,
+      "step": 12095
+    },
+    {
+      "epoch": 0.9303899700023075,
+      "learning_rate": 3.572507605561703e-05,
+      "loss": 0.9597,
+      "step": 12096
+    },
+    {
+      "epoch": 0.930466887162526,
+      "learning_rate": 3.564648325542369e-05,
+      "loss": 0.8889,
+      "step": 12097
+    },
+    {
+      "epoch": 0.9305438043227444,
+      "learning_rate": 3.5567975960358326e-05,
+      "loss": 1.2961,
+      "step": 12098
+    },
+    {
+      "epoch": 0.9306207214829628,
+      "learning_rate": 3.5489554175004936e-05,
+      "loss": 1.1012,
+      "step": 12099
+    },
+    {
+      "epoch": 0.9306976386431813,
+      "learning_rate": 3.5411217903942696e-05,
+      "loss": 1.0175,
+      "step": 12100
+    },
+    {
+      "epoch": 0.9307745558033997,
+      "learning_rate": 3.533296715174611e-05,
+      "loss": 1.2612,
+      "step": 12101
+    },
+    {
+      "epoch": 0.9308514729636181,
+      "learning_rate": 3.525480192298369e-05,
+      "loss": 1.1085,
+      "step": 12102
+    },
+    {
+      "epoch": 0.9309283901238367,
+      "learning_rate": 3.517672222221996e-05,
+      "loss": 0.8248,
+      "step": 12103
+    },
+    {
+      "epoch": 0.9310053072840551,
+      "learning_rate": 3.50987280540141e-05,
+      "loss": 1.152,
+      "step": 12104
+    },
+    {
+      "epoch": 0.9310822244442735,
+      "learning_rate": 3.502081942292012e-05,
+      "loss": 1.1902,
+      "step": 12105
+    },
+    {
+      "epoch": 0.931159141604492,
+      "learning_rate": 3.494299633348741e-05,
+      "loss": 1.0832,
+      "step": 12106
+    },
+    {
+      "epoch": 0.9312360587647104,
+      "learning_rate": 3.486525879025998e-05,
+      "loss": 1.0917,
+      "step": 12107
+    },
+    {
+      "epoch": 0.9313129759249289,
+      "learning_rate": 3.47876067977767e-05,
+      "loss": 1.348,
+      "step": 12108
+    },
+    {
+      "epoch": 0.9313898930851473,
+      "learning_rate": 3.471004036057229e-05,
+      "loss": 1.3364,
+      "step": 12109
+    },
+    {
+      "epoch": 0.9314668102453657,
+      "learning_rate": 3.463255948317595e-05,
+      "loss": 1.5034,
+      "step": 12110
+    },
+    {
+      "epoch": 0.9315437274055842,
+      "learning_rate": 3.455516417011123e-05,
+      "loss": 1.4563,
+      "step": 12111
+    },
+    {
+      "epoch": 0.9316206445658026,
+      "learning_rate": 3.447785442589802e-05,
+      "loss": 1.2429,
+      "step": 12112
+    },
+    {
+      "epoch": 0.931697561726021,
+      "learning_rate": 3.440063025505002e-05,
+      "loss": 1.0759,
+      "step": 12113
+    },
+    {
+      "epoch": 0.9317744788862395,
+      "learning_rate": 3.4323491662076825e-05,
+      "loss": 0.9891,
+      "step": 12114
+    },
+    {
+      "epoch": 0.931851396046458,
+      "learning_rate": 3.42464386514823e-05,
+      "loss": 1.1364,
+      "step": 12115
+    },
+    {
+      "epoch": 0.9319283132066765,
+      "learning_rate": 3.41694712277657e-05,
+      "loss": 0.8841,
+      "step": 12116
+    },
+    {
+      "epoch": 0.9320052303668949,
+      "learning_rate": 3.4092589395421424e-05,
+      "loss": 0.8101,
+      "step": 12117
+    },
+    {
+      "epoch": 0.9320821475271133,
+      "learning_rate": 3.4015793158938544e-05,
+      "loss": 1.2486,
+      "step": 12118
+    },
+    {
+      "epoch": 0.9321590646873318,
+      "learning_rate": 3.393908252280148e-05,
+      "loss": 1.1403,
+      "step": 12119
+    },
+    {
+      "epoch": 0.9322359818475502,
+      "learning_rate": 3.386245749148914e-05,
+      "loss": 0.958,
+      "step": 12120
+    },
+    {
+      "epoch": 0.9323128990077686,
+      "learning_rate": 3.3785918069475953e-05,
+      "loss": 1.4061,
+      "step": 12121
+    },
+    {
+      "epoch": 0.9323898161679871,
+      "learning_rate": 3.370946426123101e-05,
+      "loss": 1.1242,
+      "step": 12122
+    },
+    {
+      "epoch": 0.9324667333282055,
+      "learning_rate": 3.3633096071218724e-05,
+      "loss": 1.0106,
+      "step": 12123
+    },
+    {
+      "epoch": 0.9325436504884239,
+      "learning_rate": 3.3556813503898056e-05,
+      "loss": 0.9884,
+      "step": 12124
+    },
+    {
+      "epoch": 0.9326205676486424,
+      "learning_rate": 3.348061656372309e-05,
+      "loss": 0.9873,
+      "step": 12125
+    },
+    {
+      "epoch": 0.9326974848088608,
+      "learning_rate": 3.3404505255143605e-05,
+      "loss": 1.1096,
+      "step": 12126
+    },
+    {
+      "epoch": 0.9327744019690793,
+      "learning_rate": 3.3328479582603376e-05,
+      "loss": 1.4038,
+      "step": 12127
+    },
+    {
+      "epoch": 0.9328513191292978,
+      "learning_rate": 3.325253955054186e-05,
+      "loss": 1.0945,
+      "step": 12128
+    },
+    {
+      "epoch": 0.9329282362895162,
+      "learning_rate": 3.317668516339317e-05,
+      "loss": 1.1887,
+      "step": 12129
+    },
+    {
+      "epoch": 0.9330051534497347,
+      "learning_rate": 3.310091642558627e-05,
+      "loss": 1.9605,
+      "step": 12130
+    },
+    {
+      "epoch": 0.9330820706099531,
+      "learning_rate": 3.3025233341545945e-05,
+      "loss": 1.4124,
+      "step": 12131
+    },
+    {
+      "epoch": 0.9331589877701715,
+      "learning_rate": 3.2949635915691e-05,
+      "loss": 1.2597,
+      "step": 12132
+    },
+    {
+      "epoch": 0.93323590493039,
+      "learning_rate": 3.287412415243557e-05,
+      "loss": 0.9069,
+      "step": 12133
+    },
+    {
+      "epoch": 0.9333128220906084,
+      "learning_rate": 3.279869805618929e-05,
+      "loss": 1.1168,
+      "step": 12134
+    },
+    {
+      "epoch": 0.9333897392508269,
+      "learning_rate": 3.272335763135581e-05,
+      "loss": 0.9069,
+      "step": 12135
+    },
+    {
+      "epoch": 0.9334666564110453,
+      "learning_rate": 3.264810288233494e-05,
+      "loss": 1.2822,
+      "step": 12136
+    },
+    {
+      "epoch": 0.9335435735712637,
+      "learning_rate": 3.25729338135205e-05,
+      "loss": 1.0019,
+      "step": 12137
+    },
+    {
+      "epoch": 0.9336204907314822,
+      "learning_rate": 3.2497850429301643e-05,
+      "loss": 1.2547,
+      "step": 12138
+    },
+    {
+      "epoch": 0.9336974078917006,
+      "learning_rate": 3.24228527340627e-05,
+      "loss": 1.4441,
+      "step": 12139
+    },
+    {
+      "epoch": 0.933774325051919,
+      "learning_rate": 3.2347940732183e-05,
+      "loss": 1.1251,
+      "step": 12140
+    },
+    {
+      "epoch": 0.9338512422121376,
+      "learning_rate": 3.2273114428036546e-05,
+      "loss": 1.2194,
+      "step": 12141
+    },
+    {
+      "epoch": 0.933928159372356,
+      "learning_rate": 3.21983738259925e-05,
+      "loss": 1.4413,
+      "step": 12142
+    },
+    {
+      "epoch": 0.9340050765325744,
+      "learning_rate": 3.212371893041521e-05,
+      "loss": 1.2906,
+      "step": 12143
+    },
+    {
+      "epoch": 0.9340819936927929,
+      "learning_rate": 3.204914974566353e-05,
+      "loss": 1.2077,
+      "step": 12144
+    },
+    {
+      "epoch": 0.9341589108530113,
+      "learning_rate": 3.1974666276092135e-05,
+      "loss": 1.3196,
+      "step": 12145
+    },
+    {
+      "epoch": 0.9342358280132298,
+      "learning_rate": 3.190026852604971e-05,
+      "loss": 1.1493,
+      "step": 12146
+    },
+    {
+      "epoch": 0.9343127451734482,
+      "learning_rate": 3.182595649988079e-05,
+      "loss": 1.0146,
+      "step": 12147
+    },
+    {
+      "epoch": 0.9343896623336666,
+      "learning_rate": 3.175173020192424e-05,
+      "loss": 1.3035,
+      "step": 12148
+    },
+    {
+      "epoch": 0.9344665794938851,
+      "learning_rate": 3.16775896365144e-05,
+      "loss": 1.1528,
+      "step": 12149
+    },
+    {
+      "epoch": 0.9345434966541035,
+      "learning_rate": 3.16035348079805e-05,
+      "loss": 1.3201,
+      "step": 12150
+    },
+    {
+      "epoch": 0.9346204138143219,
+      "learning_rate": 3.152956572064625e-05,
+      "loss": 1.0597,
+      "step": 12151
+    },
+    {
+      "epoch": 0.9346973309745404,
+      "learning_rate": 3.145568237883117e-05,
+      "loss": 1.223,
+      "step": 12152
+    },
+    {
+      "epoch": 0.9347742481347588,
+      "learning_rate": 3.138188478684917e-05,
+      "loss": 1.2984,
+      "step": 12153
+    },
+    {
+      "epoch": 0.9348511652949774,
+      "learning_rate": 3.130817294900962e-05,
+      "loss": 1.4366,
+      "step": 12154
+    },
+    {
+      "epoch": 0.9349280824551958,
+      "learning_rate": 3.123454686961657e-05,
+      "loss": 1.1027,
+      "step": 12155
+    },
+    {
+      "epoch": 0.9350049996154142,
+      "learning_rate": 3.116100655296894e-05,
+      "loss": 1.3012,
+      "step": 12156
+    },
+    {
+      "epoch": 0.9350819167756327,
+      "learning_rate": 3.1087552003360775e-05,
+      "loss": 1.1954,
+      "step": 12157
+    },
+    {
+      "epoch": 0.9351588339358511,
+      "learning_rate": 3.101418322508165e-05,
+      "loss": 1.4599,
+      "step": 12158
+    },
+    {
+      "epoch": 0.9352357510960695,
+      "learning_rate": 3.094090022241497e-05,
+      "loss": 0.9842,
+      "step": 12159
+    },
+    {
+      "epoch": 0.935312668256288,
+      "learning_rate": 3.086770299964031e-05,
+      "loss": 1.1034,
+      "step": 12160
+    },
+    {
+      "epoch": 0.9353895854165064,
+      "learning_rate": 3.0794591561031426e-05,
+      "loss": 1.6368,
+      "step": 12161
+    },
+    {
+      "epoch": 0.9354665025767248,
+      "learning_rate": 3.072156591085773e-05,
+      "loss": 1.0797,
+      "step": 12162
+    },
+    {
+      "epoch": 0.9355434197369433,
+      "learning_rate": 3.064862605338281e-05,
+      "loss": 1.1395,
+      "step": 12163
+    },
+    {
+      "epoch": 0.9356203368971617,
+      "learning_rate": 3.0575771992866095e-05,
+      "loss": 1.2444,
+      "step": 12164
+    },
+    {
+      "epoch": 0.9356972540573802,
+      "learning_rate": 3.050300373356135e-05,
+      "loss": 1.0648,
+      "step": 12165
+    },
+    {
+      "epoch": 0.9357741712175986,
+      "learning_rate": 3.043032127971784e-05,
+      "loss": 1.4334,
+      "step": 12166
+    },
+    {
+      "epoch": 0.935851088377817,
+      "learning_rate": 3.0357724635579342e-05,
+      "loss": 1.1942,
+      "step": 12167
+    },
+    {
+      "epoch": 0.9359280055380356,
+      "learning_rate": 3.0285213805384793e-05,
+      "loss": 1.1077,
+      "step": 12168
+    },
+    {
+      "epoch": 0.936004922698254,
+      "learning_rate": 3.0212788793368474e-05,
+      "loss": 0.946,
+      "step": 12169
+    },
+    {
+      "epoch": 0.9360818398584724,
+      "learning_rate": 3.0140449603759167e-05,
+      "loss": 1.0302,
+      "step": 12170
+    },
+    {
+      "epoch": 0.9361587570186909,
+      "learning_rate": 3.006819624078083e-05,
+      "loss": 1.0565,
+      "step": 12171
+    },
+    {
+      "epoch": 0.9362356741789093,
+      "learning_rate": 2.999602870865259e-05,
+      "loss": 0.9015,
+      "step": 12172
+    },
+    {
+      "epoch": 0.9363125913391278,
+      "learning_rate": 2.9923947011587904e-05,
+      "loss": 1.0329,
+      "step": 12173
+    },
+    {
+      "epoch": 0.9363895084993462,
+      "learning_rate": 2.985195115379624e-05,
+      "loss": 1.229,
+      "step": 12174
+    },
+    {
+      "epoch": 0.9364664256595646,
+      "learning_rate": 2.9780041139481405e-05,
+      "loss": 1.1141,
+      "step": 12175
+    },
+    {
+      "epoch": 0.9365433428197831,
+      "learning_rate": 2.9708216972842038e-05,
+      "loss": 1.3029,
+      "step": 12176
+    },
+    {
+      "epoch": 0.9366202599800015,
+      "learning_rate": 2.9636478658072287e-05,
+      "loss": 1.1784,
+      "step": 12177
+    },
+    {
+      "epoch": 0.9366971771402199,
+      "learning_rate": 2.9564826199360795e-05,
+      "loss": 1.3269,
+      "step": 12178
+    },
+    {
+      "epoch": 0.9367740943004385,
+      "learning_rate": 2.9493259600891555e-05,
+      "loss": 1.0073,
+      "step": 12179
+    },
+    {
+      "epoch": 0.9368510114606569,
+      "learning_rate": 2.942177886684355e-05,
+      "loss": 1.264,
+      "step": 12180
+    },
+    {
+      "epoch": 0.9369279286208754,
+      "learning_rate": 2.9350384001390285e-05,
+      "loss": 0.7756,
+      "step": 12181
+    },
+    {
+      "epoch": 0.9370048457810938,
+      "learning_rate": 2.9279075008700916e-05,
+      "loss": 0.9171,
+      "step": 12182
+    },
+    {
+      "epoch": 0.9370817629413122,
+      "learning_rate": 2.920785189293912e-05,
+      "loss": 1.203,
+      "step": 12183
+    },
+    {
+      "epoch": 0.9371586801015307,
+      "learning_rate": 2.9136714658263564e-05,
+      "loss": 1.1214,
+      "step": 12184
+    },
+    {
+      "epoch": 0.9372355972617491,
+      "learning_rate": 2.906566330882826e-05,
+      "loss": 1.3194,
+      "step": 12185
+    },
+    {
+      "epoch": 0.9373125144219675,
+      "learning_rate": 2.8994697848781725e-05,
+      "loss": 1.1999,
+      "step": 12186
+    },
+    {
+      "epoch": 0.937389431582186,
+      "learning_rate": 2.8923818282267812e-05,
+      "loss": 0.9972,
+      "step": 12187
+    },
+    {
+      "epoch": 0.9374663487424044,
+      "learning_rate": 2.8853024613425372e-05,
+      "loss": 1.1405,
+      "step": 12188
+    },
+    {
+      "epoch": 0.9375432659026228,
+      "learning_rate": 2.8782316846388102e-05,
+      "loss": 1.2609,
+      "step": 12189
+    },
+    {
+      "epoch": 0.9376201830628413,
+      "learning_rate": 2.8711694985284366e-05,
+      "loss": 0.7842,
+      "step": 12190
+    },
+    {
+      "epoch": 0.9376971002230597,
+      "learning_rate": 2.8641159034238363e-05,
+      "loss": 1.2734,
+      "step": 12191
+    },
+    {
+      "epoch": 0.9377740173832783,
+      "learning_rate": 2.85707089973683e-05,
+      "loss": 1.3886,
+      "step": 12192
+    },
+    {
+      "epoch": 0.9378509345434967,
+      "learning_rate": 2.8500344878788222e-05,
+      "loss": 1.0924,
+      "step": 12193
+    },
+    {
+      "epoch": 0.9379278517037151,
+      "learning_rate": 2.8430066682606337e-05,
+      "loss": 1.0277,
+      "step": 12194
+    },
+    {
+      "epoch": 0.9380047688639336,
+      "learning_rate": 2.8359874412926534e-05,
+      "loss": 1.2736,
+      "step": 12195
+    },
+    {
+      "epoch": 0.938081686024152,
+      "learning_rate": 2.828976807384753e-05,
+      "loss": 1.0204,
+      "step": 12196
+    },
+    {
+      "epoch": 0.9381586031843704,
+      "learning_rate": 2.8219747669462558e-05,
+      "loss": 1.2011,
+      "step": 12197
+    },
+    {
+      "epoch": 0.9382355203445889,
+      "learning_rate": 2.8149813203860507e-05,
+      "loss": 1.3007,
+      "step": 12198
+    },
+    {
+      "epoch": 0.9383124375048073,
+      "learning_rate": 2.807996468112478e-05,
+      "loss": 1.2816,
+      "step": 12199
+    },
+    {
+      "epoch": 0.9383893546650258,
+      "learning_rate": 2.8010202105333782e-05,
+      "loss": 0.9556,
+      "step": 12200
+    },
+    {
+      "epoch": 0.9384662718252442,
+      "learning_rate": 2.7940525480561252e-05,
+      "loss": 0.8778,
+      "step": 12201
+    },
+    {
+      "epoch": 0.9385431889854626,
+      "learning_rate": 2.787093481087577e-05,
+      "loss": 0.861,
+      "step": 12202
+    },
+    {
+      "epoch": 0.9386201061456811,
+      "learning_rate": 2.780143010034025e-05,
+      "loss": 1.3557,
+      "step": 12203
+    },
+    {
+      "epoch": 0.9386970233058995,
+      "learning_rate": 2.7732011353013786e-05,
+      "loss": 1.0216,
+      "step": 12204
+    },
+    {
+      "epoch": 0.938773940466118,
+      "learning_rate": 2.7662678572949462e-05,
+      "loss": 1.2426,
+      "step": 12205
+    },
+    {
+      "epoch": 0.9388508576263365,
+      "learning_rate": 2.7593431764195707e-05,
+      "loss": 0.9678,
+      "step": 12206
+    },
+    {
+      "epoch": 0.9389277747865549,
+      "learning_rate": 2.7524270930796125e-05,
+      "loss": 1.3174,
+      "step": 12207
+    },
+    {
+      "epoch": 0.9390046919467733,
+      "learning_rate": 2.7455196076788812e-05,
+      "loss": 0.9357,
+      "step": 12208
+    },
+    {
+      "epoch": 0.9390816091069918,
+      "learning_rate": 2.7386207206207214e-05,
+      "loss": 1.1123,
+      "step": 12209
+    },
+    {
+      "epoch": 0.9391585262672102,
+      "learning_rate": 2.7317304323079773e-05,
+      "loss": 1.0222,
+      "step": 12210
+    },
+    {
+      "epoch": 0.9392354434274287,
+      "learning_rate": 2.724848743142977e-05,
+      "loss": 1.2877,
+      "step": 12211
+    },
+    {
+      "epoch": 0.9393123605876471,
+      "learning_rate": 2.7179756535275325e-05,
+      "loss": 1.5601,
+      "step": 12212
+    },
+    {
+      "epoch": 0.9393892777478655,
+      "learning_rate": 2.711111163863006e-05,
+      "loss": 1.0236,
+      "step": 12213
+    },
+    {
+      "epoch": 0.939466194908084,
+      "learning_rate": 2.7042552745501605e-05,
+      "loss": 1.4442,
+      "step": 12214
+    },
+    {
+      "epoch": 0.9395431120683024,
+      "learning_rate": 2.6974079859894084e-05,
+      "loss": 0.8903,
+      "step": 12215
+    },
+    {
+      "epoch": 0.9396200292285208,
+      "learning_rate": 2.690569298580481e-05,
+      "loss": 0.8188,
+      "step": 12216
+    },
+    {
+      "epoch": 0.9396969463887394,
+      "learning_rate": 2.683739212722741e-05,
+      "loss": 0.9695,
+      "step": 12217
+    },
+    {
+      "epoch": 0.9397738635489578,
+      "learning_rate": 2.6769177288150036e-05,
+      "loss": 1.0914,
+      "step": 12218
+    },
+    {
+      "epoch": 0.9398507807091763,
+      "learning_rate": 2.6701048472555834e-05,
+      "loss": 1.0811,
+      "step": 12219
+    },
+    {
+      "epoch": 0.9399276978693947,
+      "learning_rate": 2.6633005684422785e-05,
+      "loss": 1.045,
+      "step": 12220
+    },
+    {
+      "epoch": 0.9400046150296131,
+      "learning_rate": 2.6565048927724045e-05,
+      "loss": 0.731,
+      "step": 12221
+    },
+    {
+      "epoch": 0.9400815321898316,
+      "learning_rate": 2.6497178206427443e-05,
+      "loss": 1.2308,
+      "step": 12222
+    },
+    {
+      "epoch": 0.94015844935005,
+      "learning_rate": 2.642939352449647e-05,
+      "loss": 0.9671,
+      "step": 12223
+    },
+    {
+      "epoch": 0.9402353665102684,
+      "learning_rate": 2.6361694885888967e-05,
+      "loss": 1.104,
+      "step": 12224
+    },
+    {
+      "epoch": 0.9403122836704869,
+      "learning_rate": 2.6294082294557764e-05,
+      "loss": 1.1925,
+      "step": 12225
+    },
+    {
+      "epoch": 0.9403892008307053,
+      "learning_rate": 2.622655575445104e-05,
+      "loss": 1.2977,
+      "step": 12226
+    },
+    {
+      "epoch": 0.9404661179909237,
+      "learning_rate": 2.6159115269511812e-05,
+      "loss": 1.1346,
+      "step": 12227
+    },
+    {
+      "epoch": 0.9405430351511422,
+      "learning_rate": 2.6091760843677593e-05,
+      "loss": 0.9481,
+      "step": 12228
+    },
+    {
+      "epoch": 0.9406199523113606,
+      "learning_rate": 2.6024492480881735e-05,
+      "loss": 0.8181,
+      "step": 12229
+    },
+    {
+      "epoch": 0.9406968694715792,
+      "learning_rate": 2.5957310185051597e-05,
+      "loss": 0.9931,
+      "step": 12230
+    },
+    {
+      "epoch": 0.9407737866317976,
+      "learning_rate": 2.5890213960110542e-05,
+      "loss": 1.4884,
+      "step": 12231
+    },
+    {
+      "epoch": 0.940850703792016,
+      "learning_rate": 2.5823203809976105e-05,
+      "loss": 1.0598,
+      "step": 12232
+    },
+    {
+      "epoch": 0.9409276209522345,
+      "learning_rate": 2.5756279738561317e-05,
+      "loss": 1.04,
+      "step": 12233
+    },
+    {
+      "epoch": 0.9410045381124529,
+      "learning_rate": 2.5689441749773555e-05,
+      "loss": 1.4693,
+      "step": 12234
+    },
+    {
+      "epoch": 0.9410814552726713,
+      "learning_rate": 2.562268984751587e-05,
+      "loss": 1.1455,
+      "step": 12235
+    },
+    {
+      "epoch": 0.9411583724328898,
+      "learning_rate": 2.55560240356858e-05,
+      "loss": 0.9774,
+      "step": 12236
+    },
+    {
+      "epoch": 0.9412352895931082,
+      "learning_rate": 2.5489444318176235e-05,
+      "loss": 1.2256,
+      "step": 12237
+    },
+    {
+      "epoch": 0.9413122067533267,
+      "learning_rate": 2.5422950698874735e-05,
+      "loss": 1.2458,
+      "step": 12238
+    },
+    {
+      "epoch": 0.9413891239135451,
+      "learning_rate": 2.535654318166386e-05,
+      "loss": 1.0736,
+      "step": 12239
+    },
+    {
+      "epoch": 0.9414660410737635,
+      "learning_rate": 2.5290221770421173e-05,
+      "loss": 1.4101,
+      "step": 12240
+    },
+    {
+      "epoch": 0.941542958233982,
+      "learning_rate": 2.522398646901941e-05,
+      "loss": 0.8386,
+      "step": 12241
+    },
+    {
+      "epoch": 0.9416198753942004,
+      "learning_rate": 2.5157837281326147e-05,
+      "loss": 0.9314,
+      "step": 12242
+    },
+    {
+      "epoch": 0.9416967925544188,
+      "learning_rate": 2.5091774211203787e-05,
+      "loss": 1.0781,
+      "step": 12243
+    },
+    {
+      "epoch": 0.9417737097146374,
+      "learning_rate": 2.5025797262509587e-05,
+      "loss": 1.1547,
+      "step": 12244
+    },
+    {
+      "epoch": 0.9418506268748558,
+      "learning_rate": 2.4959906439096626e-05,
+      "loss": 0.8614,
+      "step": 12245
+    },
+    {
+      "epoch": 0.9419275440350742,
+      "learning_rate": 2.4894101744811824e-05,
+      "loss": 1.0271,
+      "step": 12246
+    },
+    {
+      "epoch": 0.9420044611952927,
+      "learning_rate": 2.4828383183497617e-05,
+      "loss": 0.7347,
+      "step": 12247
+    },
+    {
+      "epoch": 0.9420813783555111,
+      "learning_rate": 2.476275075899176e-05,
+      "loss": 1.1688,
+      "step": 12248
+    },
+    {
+      "epoch": 0.9421582955157296,
+      "learning_rate": 2.4697204475126024e-05,
+      "loss": 1.2839,
+      "step": 12249
+    },
+    {
+      "epoch": 0.942235212675948,
+      "learning_rate": 2.463174433572818e-05,
+      "loss": 0.9427,
+      "step": 12250
+    },
+    {
+      "epoch": 0.9423121298361664,
+      "learning_rate": 2.456637034462067e-05,
+      "loss": 1.2026,
+      "step": 12251
+    },
+    {
+      "epoch": 0.9423890469963849,
+      "learning_rate": 2.4501082505620108e-05,
+      "loss": 0.7465,
+      "step": 12252
+    },
+    {
+      "epoch": 0.9424659641566033,
+      "learning_rate": 2.4435880822539116e-05,
+      "loss": 1.1824,
+      "step": 12253
+    },
+    {
+      "epoch": 0.9425428813168217,
+      "learning_rate": 2.4370765299184805e-05,
+      "loss": 1.3512,
+      "step": 12254
+    },
+    {
+      "epoch": 0.9426197984770402,
+      "learning_rate": 2.4305735939359473e-05,
+      "loss": 1.1434,
+      "step": 12255
+    },
+    {
+      "epoch": 0.9426967156372587,
+      "learning_rate": 2.424079274686025e-05,
+      "loss": 1.169,
+      "step": 12256
+    },
+    {
+      "epoch": 0.9427736327974772,
+      "learning_rate": 2.4175935725478938e-05,
+      "loss": 1.2582,
+      "step": 12257
+    },
+    {
+      "epoch": 0.9428505499576956,
+      "learning_rate": 2.4111164879002832e-05,
+      "loss": 1.2718,
+      "step": 12258
+    },
+    {
+      "epoch": 0.942927467117914,
+      "learning_rate": 2.4046480211214082e-05,
+      "loss": 0.9659,
+      "step": 12259
+    },
+    {
+      "epoch": 0.9430043842781325,
+      "learning_rate": 2.3981881725889332e-05,
+      "loss": 1.3946,
+      "step": 12260
+    },
+    {
+      "epoch": 0.9430813014383509,
+      "learning_rate": 2.391736942680106e-05,
+      "loss": 1.2674,
+      "step": 12261
+    },
+    {
+      "epoch": 0.9431582185985693,
+      "learning_rate": 2.3852943317715757e-05,
+      "loss": 1.0531,
+      "step": 12262
+    },
+    {
+      "epoch": 0.9432351357587878,
+      "learning_rate": 2.3788603402395415e-05,
+      "loss": 0.9711,
+      "step": 12263
+    },
+    {
+      "epoch": 0.9433120529190062,
+      "learning_rate": 2.372434968459719e-05,
+      "loss": 1.3543,
+      "step": 12264
+    },
+    {
+      "epoch": 0.9433889700792246,
+      "learning_rate": 2.3660182168072585e-05,
+      "loss": 0.9723,
+      "step": 12265
+    },
+    {
+      "epoch": 0.9434658872394431,
+      "learning_rate": 2.3596100856568603e-05,
+      "loss": 0.8284,
+      "step": 12266
+    },
+    {
+      "epoch": 0.9435428043996615,
+      "learning_rate": 2.3532105753826917e-05,
+      "loss": 1.4464,
+      "step": 12267
+    },
+    {
+      "epoch": 0.94361972155988,
+      "learning_rate": 2.3468196863584367e-05,
+      "loss": 0.8575,
+      "step": 12268
+    },
+    {
+      "epoch": 0.9436966387200985,
+      "learning_rate": 2.3404374189572643e-05,
+      "loss": 1.0575,
+      "step": 12269
+    },
+    {
+      "epoch": 0.9437735558803169,
+      "learning_rate": 2.3340637735518254e-05,
+      "loss": 0.663,
+      "step": 12270
+    },
+    {
+      "epoch": 0.9438504730405354,
+      "learning_rate": 2.32769875051429e-05,
+      "loss": 0.9797,
+      "step": 12271
+    },
+    {
+      "epoch": 0.9439273902007538,
+      "learning_rate": 2.3213423502163432e-05,
+      "loss": 0.8766,
+      "step": 12272
+    },
+    {
+      "epoch": 0.9440043073609722,
+      "learning_rate": 2.314994573029139e-05,
+      "loss": 0.9139,
+      "step": 12273
+    },
+    {
+      "epoch": 0.9440812245211907,
+      "learning_rate": 2.30865541932328e-05,
+      "loss": 1.2268,
+      "step": 12274
+    },
+    {
+      "epoch": 0.9441581416814091,
+      "learning_rate": 2.302324889468971e-05,
+      "loss": 1.0848,
+      "step": 12275
+    },
+    {
+      "epoch": 0.9442350588416276,
+      "learning_rate": 2.2960029838358154e-05,
+      "loss": 1.016,
+      "step": 12276
+    },
+    {
+      "epoch": 0.944311976001846,
+      "learning_rate": 2.2896897027930186e-05,
+      "loss": 1.0007,
+      "step": 12277
+    },
+    {
+      "epoch": 0.9443888931620644,
+      "learning_rate": 2.2833850467091523e-05,
+      "loss": 1.1234,
+      "step": 12278
+    },
+    {
+      "epoch": 0.9444658103222829,
+      "learning_rate": 2.2770890159523882e-05,
+      "loss": 1.1582,
+      "step": 12279
+    },
+    {
+      "epoch": 0.9445427274825013,
+      "learning_rate": 2.2708016108903494e-05,
+      "loss": 1.3735,
+      "step": 12280
+    },
+    {
+      "epoch": 0.9446196446427197,
+      "learning_rate": 2.264522831890159e-05,
+      "loss": 1.0893,
+      "step": 12281
+    },
+    {
+      "epoch": 0.9446965618029383,
+      "learning_rate": 2.2582526793184566e-05,
+      "loss": 1.427,
+      "step": 12282
+    },
+    {
+      "epoch": 0.9447734789631567,
+      "learning_rate": 2.251991153541333e-05,
+      "loss": 1.382,
+      "step": 12283
+    },
+    {
+      "epoch": 0.9448503961233752,
+      "learning_rate": 2.2457382549244288e-05,
+      "loss": 1.1619,
+      "step": 12284
+    },
+    {
+      "epoch": 0.9449273132835936,
+      "learning_rate": 2.239493983832869e-05,
+      "loss": 1.1343,
+      "step": 12285
+    },
+    {
+      "epoch": 0.945004230443812,
+      "learning_rate": 2.233258340631261e-05,
+      "loss": 0.9671,
+      "step": 12286
+    },
+    {
+      "epoch": 0.9450811476040305,
+      "learning_rate": 2.227031325683665e-05,
+      "loss": 0.9059,
+      "step": 12287
+    },
+    {
+      "epoch": 0.9451580647642489,
+      "learning_rate": 2.220812939353739e-05,
+      "loss": 1.4422,
+      "step": 12288
+    },
+    {
+      "epoch": 0.9452349819244673,
+      "learning_rate": 2.214603182004543e-05,
+      "loss": 1.3011,
+      "step": 12289
+    },
+    {
+      "epoch": 0.9453118990846858,
+      "learning_rate": 2.208402053998687e-05,
+      "loss": 1.0817,
+      "step": 12290
+    },
+    {
+      "epoch": 0.9453888162449042,
+      "learning_rate": 2.2022095556982645e-05,
+      "loss": 1.4671,
+      "step": 12291
+    },
+    {
+      "epoch": 0.9454657334051226,
+      "learning_rate": 2.1960256874648532e-05,
+      "loss": 0.9501,
+      "step": 12292
+    },
+    {
+      "epoch": 0.9455426505653411,
+      "learning_rate": 2.1898504496595306e-05,
+      "loss": 1.4825,
+      "step": 12293
+    },
+    {
+      "epoch": 0.9456195677255596,
+      "learning_rate": 2.1836838426429086e-05,
+      "loss": 1.1331,
+      "step": 12294
+    },
+    {
+      "epoch": 0.9456964848857781,
+      "learning_rate": 2.1775258667750153e-05,
+      "loss": 1.0971,
+      "step": 12295
+    },
+    {
+      "epoch": 0.9457734020459965,
+      "learning_rate": 2.1713765224154635e-05,
+      "loss": 1.2781,
+      "step": 12296
+    },
+    {
+      "epoch": 0.9458503192062149,
+      "learning_rate": 2.165235809923283e-05,
+      "loss": 1.1165,
+      "step": 12297
+    },
+    {
+      "epoch": 0.9459272363664334,
+      "learning_rate": 2.1591037296570526e-05,
+      "loss": 0.9566,
+      "step": 12298
+    },
+    {
+      "epoch": 0.9460041535266518,
+      "learning_rate": 2.1529802819748533e-05,
+      "loss": 1.3316,
+      "step": 12299
+    },
+    {
+      "epoch": 0.9460810706868702,
+      "learning_rate": 2.146865467234199e-05,
+      "loss": 1.2123,
+      "step": 12300
+    },
+    {
+      "epoch": 0.9461579878470887,
+      "learning_rate": 2.1407592857921542e-05,
+      "loss": 1.2362,
+      "step": 12301
+    },
+    {
+      "epoch": 0.9462349050073071,
+      "learning_rate": 2.1346617380052834e-05,
+      "loss": 1.0055,
+      "step": 12302
+    },
+    {
+      "epoch": 0.9463118221675256,
+      "learning_rate": 2.128572824229602e-05,
+      "loss": 1.2387,
+      "step": 12303
+    },
+    {
+      "epoch": 0.946388739327744,
+      "learning_rate": 2.1224925448206755e-05,
+      "loss": 1.1984,
+      "step": 12304
+    },
+    {
+      "epoch": 0.9464656564879624,
+      "learning_rate": 2.11642090013352e-05,
+      "loss": 1.3221,
+      "step": 12305
+    },
+    {
+      "epoch": 0.946542573648181,
+      "learning_rate": 2.110357890522652e-05,
+      "loss": 0.9596,
+      "step": 12306
+    },
+    {
+      "epoch": 0.9466194908083994,
+      "learning_rate": 2.1043035163421374e-05,
+      "loss": 1.3926,
+      "step": 12307
+    },
+    {
+      "epoch": 0.9466964079686178,
+      "learning_rate": 2.098257777945478e-05,
+      "loss": 1.5158,
+      "step": 12308
+    },
+    {
+      "epoch": 0.9467733251288363,
+      "learning_rate": 2.0922206756856577e-05,
+      "loss": 1.2633,
+      "step": 12309
+    },
+    {
+      "epoch": 0.9468502422890547,
+      "learning_rate": 2.0861922099152275e-05,
+      "loss": 1.2649,
+      "step": 12310
+    },
+    {
+      "epoch": 0.9469271594492731,
+      "learning_rate": 2.0801723809862062e-05,
+      "loss": 1.0729,
+      "step": 12311
+    },
+    {
+      "epoch": 0.9470040766094916,
+      "learning_rate": 2.0741611892500624e-05,
+      "loss": 0.7765,
+      "step": 12312
+    },
+    {
+      "epoch": 0.94708099376971,
+      "learning_rate": 2.0681586350578152e-05,
+      "loss": 1.2071,
+      "step": 12313
+    },
+    {
+      "epoch": 0.9471579109299285,
+      "learning_rate": 2.0621647187599347e-05,
+      "loss": 0.839,
+      "step": 12314
+    },
+    {
+      "epoch": 0.9472348280901469,
+      "learning_rate": 2.056179440706457e-05,
+      "loss": 1.1766,
+      "step": 12315
+    },
+    {
+      "epoch": 0.9473117452503653,
+      "learning_rate": 2.0502028012468367e-05,
+      "loss": 1.2616,
+      "step": 12316
+    },
+    {
+      "epoch": 0.9473886624105838,
+      "learning_rate": 2.044234800730077e-05,
+      "loss": 1.3573,
+      "step": 12317
+    },
+    {
+      "epoch": 0.9474655795708022,
+      "learning_rate": 2.0382754395046332e-05,
+      "loss": 1.0533,
+      "step": 12318
+    },
+    {
+      "epoch": 0.9475424967310206,
+      "learning_rate": 2.0323247179184935e-05,
+      "loss": 1.2565,
+      "step": 12319
+    },
+    {
+      "epoch": 0.9476194138912392,
+      "learning_rate": 2.0263826363190962e-05,
+      "loss": 1.0266,
+      "step": 12320
+    },
+    {
+      "epoch": 0.9476963310514576,
+      "learning_rate": 2.0204491950534642e-05,
+      "loss": 0.9547,
+      "step": 12321
+    },
+    {
+      "epoch": 0.9477732482116761,
+      "learning_rate": 2.0145243944680036e-05,
+      "loss": 1.3065,
+      "step": 12322
+    },
+    {
+      "epoch": 0.9478501653718945,
+      "learning_rate": 2.0086082349087043e-05,
+      "loss": 1.349,
+      "step": 12323
+    },
+    {
+      "epoch": 0.9479270825321129,
+      "learning_rate": 2.0027007167210066e-05,
+      "loss": 1.2137,
+      "step": 12324
+    },
+    {
+      "epoch": 0.9480039996923314,
+      "learning_rate": 1.9968018402498346e-05,
+      "loss": 1.0298,
+      "step": 12325
+    },
+    {
+      "epoch": 0.9480809168525498,
+      "learning_rate": 1.9909116058396626e-05,
+      "loss": 1.2058,
+      "step": 12326
+    },
+    {
+      "epoch": 0.9481578340127682,
+      "learning_rate": 1.985030013834416e-05,
+      "loss": 1.0441,
+      "step": 12327
+    },
+    {
+      "epoch": 0.9482347511729867,
+      "learning_rate": 1.9791570645775025e-05,
+      "loss": 0.878,
+      "step": 12328
+    },
+    {
+      "epoch": 0.9483116683332051,
+      "learning_rate": 1.973292758411882e-05,
+      "loss": 1.1512,
+      "step": 12329
+    },
+    {
+      "epoch": 0.9483885854934235,
+      "learning_rate": 1.96743709567998e-05,
+      "loss": 1.0311,
+      "step": 12330
+    },
+    {
+      "epoch": 0.948465502653642,
+      "learning_rate": 1.9615900767236904e-05,
+      "loss": 1.0456,
+      "step": 12331
+    },
+    {
+      "epoch": 0.9485424198138604,
+      "learning_rate": 1.9557517018844396e-05,
+      "loss": 1.007,
+      "step": 12332
+    },
+    {
+      "epoch": 0.948619336974079,
+      "learning_rate": 1.949921971503138e-05,
+      "loss": 1.0526,
+      "step": 12333
+    },
+    {
+      "epoch": 0.9486962541342974,
+      "learning_rate": 1.944100885920197e-05,
+      "loss": 1.2026,
+      "step": 12334
+    },
+    {
+      "epoch": 0.9487731712945158,
+      "learning_rate": 1.9382884454754947e-05,
+      "loss": 1.0834,
+      "step": 12335
+    },
+    {
+      "epoch": 0.9488500884547343,
+      "learning_rate": 1.932484650508426e-05,
+      "loss": 1.0423,
+      "step": 12336
+    },
+    {
+      "epoch": 0.9489270056149527,
+      "learning_rate": 1.9266895013578866e-05,
+      "loss": 1.0567,
+      "step": 12337
+    },
+    {
+      "epoch": 0.9490039227751711,
+      "learning_rate": 1.9209029983622893e-05,
+      "loss": 0.9646,
+      "step": 12338
+    },
+    {
+      "epoch": 0.9490808399353896,
+      "learning_rate": 1.9151251418594806e-05,
+      "loss": 1.1774,
+      "step": 12339
+    },
+    {
+      "epoch": 0.949157757095608,
+      "learning_rate": 1.9093559321868404e-05,
+      "loss": 0.8241,
+      "step": 12340
+    },
+    {
+      "epoch": 0.9492346742558265,
+      "learning_rate": 1.903595369681249e-05,
+      "loss": 1.0441,
+      "step": 12341
+    },
+    {
+      "epoch": 0.9493115914160449,
+      "learning_rate": 1.8978434546790546e-05,
+      "loss": 1.4644,
+      "step": 12342
+    },
+    {
+      "epoch": 0.9493885085762633,
+      "learning_rate": 1.8921001875161546e-05,
+      "loss": 1.3055,
+      "step": 12343
+    },
+    {
+      "epoch": 0.9494654257364818,
+      "learning_rate": 1.886365568527848e-05,
+      "loss": 1.0192,
+      "step": 12344
+    },
+    {
+      "epoch": 0.9495423428967003,
+      "learning_rate": 1.88063959804905e-05,
+      "loss": 1.0575,
+      "step": 12345
+    },
+    {
+      "epoch": 0.9496192600569187,
+      "learning_rate": 1.8749222764140426e-05,
+      "loss": 0.9611,
+      "step": 12346
+    },
+    {
+      "epoch": 0.9496961772171372,
+      "learning_rate": 1.86921360395671e-05,
+      "loss": 1.447,
+      "step": 12347
+    },
+    {
+      "epoch": 0.9497730943773556,
+      "learning_rate": 1.8635135810103676e-05,
+      "loss": 1.3901,
+      "step": 12348
+    },
+    {
+      "epoch": 0.949850011537574,
+      "learning_rate": 1.8578222079078665e-05,
+      "loss": 0.9232,
+      "step": 12349
+    },
+    {
+      "epoch": 0.9499269286977925,
+      "learning_rate": 1.852139484981491e-05,
+      "loss": 1.4881,
+      "step": 12350
+    },
+    {
+      "epoch": 0.9500038458580109,
+      "learning_rate": 1.8464654125631087e-05,
+      "loss": 1.4199,
+      "step": 12351
+    },
+    {
+      "epoch": 0.9500807630182294,
+      "learning_rate": 1.8407999909840045e-05,
+      "loss": 1.2796,
+      "step": 12352
+    },
+    {
+      "epoch": 0.9501576801784478,
+      "learning_rate": 1.8351432205749974e-05,
+      "loss": 1.4685,
+      "step": 12353
+    },
+    {
+      "epoch": 0.9502345973386662,
+      "learning_rate": 1.8294951016663895e-05,
+      "loss": 0.8389,
+      "step": 12354
+    },
+    {
+      "epoch": 0.9503115144988847,
+      "learning_rate": 1.8238556345879675e-05,
+      "loss": 1.0521,
+      "step": 12355
+    },
+    {
+      "epoch": 0.9503884316591031,
+      "learning_rate": 1.8182248196690678e-05,
+      "loss": 1.272,
+      "step": 12356
+    },
+    {
+      "epoch": 0.9504653488193215,
+      "learning_rate": 1.8126026572384435e-05,
+      "loss": 1.0702,
+      "step": 12357
+    },
+    {
+      "epoch": 0.9505422659795401,
+      "learning_rate": 1.8069891476243827e-05,
+      "loss": 0.8724,
+      "step": 12358
+    },
+    {
+      "epoch": 0.9506191831397585,
+      "learning_rate": 1.801384291154673e-05,
+      "loss": 1.3949,
+      "step": 12359
+    },
+    {
+      "epoch": 0.950696100299977,
+      "learning_rate": 1.795788088156586e-05,
+      "loss": 1.252,
+      "step": 12360
+    },
+    {
+      "epoch": 0.9507730174601954,
+      "learning_rate": 1.7902005389568765e-05,
+      "loss": 1.1387,
+      "step": 12361
+    },
+    {
+      "epoch": 0.9508499346204138,
+      "learning_rate": 1.784621643881834e-05,
+      "loss": 1.2472,
+      "step": 12362
+    },
+    {
+      "epoch": 0.9509268517806323,
+      "learning_rate": 1.7790514032571815e-05,
+      "loss": 1.1041,
+      "step": 12363
+    },
+    {
+      "epoch": 0.9510037689408507,
+      "learning_rate": 1.7734898174081916e-05,
+      "loss": 0.9482,
+      "step": 12364
+    },
+    {
+      "epoch": 0.9510806861010691,
+      "learning_rate": 1.767936886659621e-05,
+      "loss": 1.2077,
+      "step": 12365
+    },
+    {
+      "epoch": 0.9511576032612876,
+      "learning_rate": 1.7623926113356946e-05,
+      "loss": 1.1982,
+      "step": 12366
+    },
+    {
+      "epoch": 0.951234520421506,
+      "learning_rate": 1.756856991760153e-05,
+      "loss": 1.4091,
+      "step": 12367
+    },
+    {
+      "epoch": 0.9513114375817244,
+      "learning_rate": 1.7513300282562206e-05,
+      "loss": 1.1945,
+      "step": 12368
+    },
+    {
+      "epoch": 0.9513883547419429,
+      "learning_rate": 1.7458117211466395e-05,
+      "loss": 1.4266,
+      "step": 12369
+    },
+    {
+      "epoch": 0.9514652719021613,
+      "learning_rate": 1.740302070753619e-05,
+      "loss": 1.3746,
+      "step": 12370
+    },
+    {
+      "epoch": 0.9515421890623799,
+      "learning_rate": 1.7348010773988676e-05,
+      "loss": 1.2154,
+      "step": 12371
+    },
+    {
+      "epoch": 0.9516191062225983,
+      "learning_rate": 1.7293087414035956e-05,
+      "loss": 1.1474,
+      "step": 12372
+    },
+    {
+      "epoch": 0.9516960233828167,
+      "learning_rate": 1.7238250630885132e-05,
+      "loss": 0.9347,
+      "step": 12373
+    },
+    {
+      "epoch": 0.9517729405430352,
+      "learning_rate": 1.7183500427738142e-05,
+      "loss": 0.8565,
+      "step": 12374
+    },
+    {
+      "epoch": 0.9518498577032536,
+      "learning_rate": 1.7128836807791926e-05,
+      "loss": 1.2474,
+      "step": 12375
+    },
+    {
+      "epoch": 0.951926774863472,
+      "learning_rate": 1.7074259774238432e-05,
+      "loss": 1.1588,
+      "step": 12376
+    },
+    {
+      "epoch": 0.9520036920236905,
+      "learning_rate": 1.701976933026428e-05,
+      "loss": 0.9822,
+      "step": 12377
+    },
+    {
+      "epoch": 0.9520806091839089,
+      "learning_rate": 1.696536547905142e-05,
+      "loss": 1.2391,
+      "step": 12378
+    },
+    {
+      "epoch": 0.9521575263441274,
+      "learning_rate": 1.6911048223776316e-05,
+      "loss": 0.9747,
+      "step": 12379
+    },
+    {
+      "epoch": 0.9522344435043458,
+      "learning_rate": 1.685681756761076e-05,
+      "loss": 1.4819,
+      "step": 12380
+    },
+    {
+      "epoch": 0.9523113606645642,
+      "learning_rate": 1.680267351372139e-05,
+      "loss": 1.3563,
+      "step": 12381
+    },
+    {
+      "epoch": 0.9523882778247827,
+      "learning_rate": 1.6748616065269673e-05,
+      "loss": 1.059,
+      "step": 12382
+    },
+    {
+      "epoch": 0.9524651949850012,
+      "learning_rate": 1.6694645225412087e-05,
+      "loss": 1.4287,
+      "step": 12383
+    },
+    {
+      "epoch": 0.9525421121452196,
+      "learning_rate": 1.6640760997299942e-05,
+      "loss": 1.0282,
+      "step": 12384
+    },
+    {
+      "epoch": 0.9526190293054381,
+      "learning_rate": 1.6586963384079722e-05,
+      "loss": 1.0989,
+      "step": 12385
+    },
+    {
+      "epoch": 0.9526959464656565,
+      "learning_rate": 1.6533252388892584e-05,
+      "loss": 1.3305,
+      "step": 12386
+    },
+    {
+      "epoch": 0.9527728636258749,
+      "learning_rate": 1.6479628014875014e-05,
+      "loss": 1.2509,
+      "step": 12387
+    },
+    {
+      "epoch": 0.9528497807860934,
+      "learning_rate": 1.642609026515801e-05,
+      "loss": 1.2583,
+      "step": 12388
+    },
+    {
+      "epoch": 0.9529266979463118,
+      "learning_rate": 1.637263914286774e-05,
+      "loss": 0.9596,
+      "step": 12389
+    },
+    {
+      "epoch": 0.9530036151065303,
+      "learning_rate": 1.6319274651125204e-05,
+      "loss": 0.7987,
+      "step": 12390
+    },
+    {
+      "epoch": 0.9530805322667487,
+      "learning_rate": 1.626599679304658e-05,
+      "loss": 1.459,
+      "step": 12391
+    },
+    {
+      "epoch": 0.9531574494269671,
+      "learning_rate": 1.621280557174287e-05,
+      "loss": 0.8505,
+      "step": 12392
+    },
+    {
+      "epoch": 0.9532343665871856,
+      "learning_rate": 1.6159700990319436e-05,
+      "loss": 1.2911,
+      "step": 12393
+    },
+    {
+      "epoch": 0.953311283747404,
+      "learning_rate": 1.6106683051877623e-05,
+      "loss": 0.6452,
+      "step": 12394
+    },
+    {
+      "epoch": 0.9533882009076224,
+      "learning_rate": 1.6053751759513124e-05,
+      "loss": 1.2833,
+      "step": 12395
+    },
+    {
+      "epoch": 0.953465118067841,
+      "learning_rate": 1.600090711631663e-05,
+      "loss": 1.3633,
+      "step": 12396
+    },
+    {
+      "epoch": 0.9535420352280594,
+      "learning_rate": 1.5948149125373678e-05,
+      "loss": 0.9192,
+      "step": 12397
+    },
+    {
+      "epoch": 0.9536189523882779,
+      "learning_rate": 1.5895477789764966e-05,
+      "loss": 1.0924,
+      "step": 12398
+    },
+    {
+      "epoch": 0.9536958695484963,
+      "learning_rate": 1.584289311256587e-05,
+      "loss": 1.0657,
+      "step": 12399
+    },
+    {
+      "epoch": 0.9537727867087147,
+      "learning_rate": 1.5790395096847264e-05,
+      "loss": 1.6161,
+      "step": 12400
+    },
+    {
+      "epoch": 0.9538497038689332,
+      "learning_rate": 1.5737983745674035e-05,
+      "loss": 1.2523,
+      "step": 12401
+    },
+    {
+      "epoch": 0.9539266210291516,
+      "learning_rate": 1.568565906210706e-05,
+      "loss": 1.4484,
+      "step": 12402
+    },
+    {
+      "epoch": 0.95400353818937,
+      "learning_rate": 1.5633421049201236e-05,
+      "loss": 1.0825,
+      "step": 12403
+    },
+    {
+      "epoch": 0.9540804553495885,
+      "learning_rate": 1.5581269710006785e-05,
+      "loss": 1.1632,
+      "step": 12404
+    },
+    {
+      "epoch": 0.9541573725098069,
+      "learning_rate": 1.5529205047569273e-05,
+      "loss": 0.9711,
+      "step": 12405
+    },
+    {
+      "epoch": 0.9542342896700254,
+      "learning_rate": 1.547722706492827e-05,
+      "loss": 0.9484,
+      "step": 12406
+    },
+    {
+      "epoch": 0.9543112068302438,
+      "learning_rate": 1.542533576511934e-05,
+      "loss": 1.3789,
+      "step": 12407
+    },
+    {
+      "epoch": 0.9543881239904622,
+      "learning_rate": 1.5373531151172236e-05,
+      "loss": 0.9206,
+      "step": 12408
+    },
+    {
+      "epoch": 0.9544650411506808,
+      "learning_rate": 1.5321813226111703e-05,
+      "loss": 1.1812,
+      "step": 12409
+    },
+    {
+      "epoch": 0.9545419583108992,
+      "learning_rate": 1.5270181992957987e-05,
+      "loss": 0.9874,
+      "step": 12410
+    },
+    {
+      "epoch": 0.9546188754711176,
+      "learning_rate": 1.5218637454725682e-05,
+      "loss": 1.3513,
+      "step": 12411
+    },
+    {
+      "epoch": 0.9546957926313361,
+      "learning_rate": 1.5167179614424376e-05,
+      "loss": 1.4242,
+      "step": 12412
+    },
+    {
+      "epoch": 0.9547727097915545,
+      "learning_rate": 1.5115808475059001e-05,
+      "loss": 1.3357,
+      "step": 12413
+    },
+    {
+      "epoch": 0.9548496269517729,
+      "learning_rate": 1.5064524039629157e-05,
+      "loss": 1.1638,
+      "step": 12414
+    },
+    {
+      "epoch": 0.9549265441119914,
+      "learning_rate": 1.5013326311129117e-05,
+      "loss": 1.4263,
+      "step": 12415
+    },
+    {
+      "epoch": 0.9550034612722098,
+      "learning_rate": 1.496221529254882e-05,
+      "loss": 1.3826,
+      "step": 12416
+    },
+    {
+      "epoch": 0.9550803784324283,
+      "learning_rate": 1.4911190986872213e-05,
+      "loss": 1.1397,
+      "step": 12417
+    },
+    {
+      "epoch": 0.9551572955926467,
+      "learning_rate": 1.4860253397079082e-05,
+      "loss": 1.0621,
+      "step": 12418
+    },
+    {
+      "epoch": 0.9552342127528651,
+      "learning_rate": 1.4809402526143378e-05,
+      "loss": 1.0134,
+      "step": 12419
+    },
+    {
+      "epoch": 0.9553111299130836,
+      "learning_rate": 1.4758638377034562e-05,
+      "loss": 1.1756,
+      "step": 12420
+    },
+    {
+      "epoch": 0.955388047073302,
+      "learning_rate": 1.470796095271676e-05,
+      "loss": 1.0331,
+      "step": 12421
+    },
+    {
+      "epoch": 0.9554649642335205,
+      "learning_rate": 1.4657370256148939e-05,
+      "loss": 0.7614,
+      "step": 12422
+    },
+    {
+      "epoch": 0.955541881393739,
+      "learning_rate": 1.4606866290285404e-05,
+      "loss": 1.2887,
+      "step": 12423
+    },
+    {
+      "epoch": 0.9556187985539574,
+      "learning_rate": 1.4556449058074794e-05,
+      "loss": 1.156,
+      "step": 12424
+    },
+    {
+      "epoch": 0.9556957157141759,
+      "learning_rate": 1.4506118562461257e-05,
+      "loss": 0.9547,
+      "step": 12425
+    },
+    {
+      "epoch": 0.9557726328743943,
+      "learning_rate": 1.4455874806383772e-05,
+      "loss": 0.8543,
+      "step": 12426
+    },
+    {
+      "epoch": 0.9558495500346127,
+      "learning_rate": 1.4405717792775996e-05,
+      "loss": 1.0461,
+      "step": 12427
+    },
+    {
+      "epoch": 0.9559264671948312,
+      "learning_rate": 1.4355647524566417e-05,
+      "loss": 0.9796,
+      "step": 12428
+    },
+    {
+      "epoch": 0.9560033843550496,
+      "learning_rate": 1.4305664004678865e-05,
+      "loss": 1.3172,
+      "step": 12429
+    },
+    {
+      "epoch": 0.956080301515268,
+      "learning_rate": 1.4255767236032002e-05,
+      "loss": 1.2666,
+      "step": 12430
+    },
+    {
+      "epoch": 0.9561572186754865,
+      "learning_rate": 1.4205957221539334e-05,
+      "loss": 1.1766,
+      "step": 12431
+    },
+    {
+      "epoch": 0.9562341358357049,
+      "learning_rate": 1.4156233964109366e-05,
+      "loss": 1.1884,
+      "step": 12432
+    },
+    {
+      "epoch": 0.9563110529959233,
+      "learning_rate": 1.4106597466645277e-05,
+      "loss": 0.9319,
+      "step": 12433
+    },
+    {
+      "epoch": 0.9563879701561419,
+      "learning_rate": 1.4057047732045414e-05,
+      "loss": 1.0771,
+      "step": 12434
+    },
+    {
+      "epoch": 0.9564648873163603,
+      "learning_rate": 1.4007584763203297e-05,
+      "loss": 1.1815,
+      "step": 12435
+    },
+    {
+      "epoch": 0.9565418044765788,
+      "learning_rate": 1.3958208563006947e-05,
+      "loss": 1.0333,
+      "step": 12436
+    },
+    {
+      "epoch": 0.9566187216367972,
+      "learning_rate": 1.390891913433956e-05,
+      "loss": 1.2589,
+      "step": 12437
+    },
+    {
+      "epoch": 0.9566956387970156,
+      "learning_rate": 1.3859716480079165e-05,
+      "loss": 1.2224,
+      "step": 12438
+    },
+    {
+      "epoch": 0.9567725559572341,
+      "learning_rate": 1.3810600603098466e-05,
+      "loss": 1.0726,
+      "step": 12439
+    },
+    {
+      "epoch": 0.9568494731174525,
+      "learning_rate": 1.3761571506266002e-05,
+      "loss": 0.8625,
+      "step": 12440
+    },
+    {
+      "epoch": 0.9569263902776709,
+      "learning_rate": 1.3712629192444314e-05,
+      "loss": 1.1425,
+      "step": 12441
+    },
+    {
+      "epoch": 0.9570033074378894,
+      "learning_rate": 1.3663773664490953e-05,
+      "loss": 1.7033,
+      "step": 12442
+    },
+    {
+      "epoch": 0.9570802245981078,
+      "learning_rate": 1.3615004925258968e-05,
+      "loss": 1.3728,
+      "step": 12443
+    },
+    {
+      "epoch": 0.9571571417583263,
+      "learning_rate": 1.3566322977595913e-05,
+      "loss": 1.5058,
+      "step": 12444
+    },
+    {
+      "epoch": 0.9572340589185447,
+      "learning_rate": 1.351772782434435e-05,
+      "loss": 1.179,
+      "step": 12445
+    },
+    {
+      "epoch": 0.9573109760787631,
+      "learning_rate": 1.3469219468341842e-05,
+      "loss": 1.303,
+      "step": 12446
+    },
+    {
+      "epoch": 0.9573878932389817,
+      "learning_rate": 1.342079791242079e-05,
+      "loss": 1.1784,
+      "step": 12447
+    },
+    {
+      "epoch": 0.9574648103992001,
+      "learning_rate": 1.3372463159408765e-05,
+      "loss": 1.2931,
+      "step": 12448
+    },
+    {
+      "epoch": 0.9575417275594185,
+      "learning_rate": 1.332421521212801e-05,
+      "loss": 1.0764,
+      "step": 12449
+    },
+    {
+      "epoch": 0.957618644719637,
+      "learning_rate": 1.3276054073395438e-05,
+      "loss": 0.9208,
+      "step": 12450
+    },
+    {
+      "epoch": 0.9576955618798554,
+      "learning_rate": 1.3227979746023633e-05,
+      "loss": 1.3162,
+      "step": 12451
+    },
+    {
+      "epoch": 0.9577724790400738,
+      "learning_rate": 1.3179992232819515e-05,
+      "loss": 1.0227,
+      "step": 12452
+    },
+    {
+      "epoch": 0.9578493962002923,
+      "learning_rate": 1.3132091536585344e-05,
+      "loss": 1.5541,
+      "step": 12453
+    },
+    {
+      "epoch": 0.9579263133605107,
+      "learning_rate": 1.3084277660117716e-05,
+      "loss": 1.3618,
+      "step": 12454
+    },
+    {
+      "epoch": 0.9580032305207292,
+      "learning_rate": 1.303655060620873e-05,
+      "loss": 1.1838,
+      "step": 12455
+    },
+    {
+      "epoch": 0.9580801476809476,
+      "learning_rate": 1.298891037764549e-05,
+      "loss": 1.0831,
+      "step": 12456
+    },
+    {
+      "epoch": 0.958157064841166,
+      "learning_rate": 1.2941356977209273e-05,
+      "loss": 1.1509,
+      "step": 12457
+    },
+    {
+      "epoch": 0.9582339820013845,
+      "learning_rate": 1.2893890407677023e-05,
+      "loss": 0.9196,
+      "step": 12458
+    },
+    {
+      "epoch": 0.958310899161603,
+      "learning_rate": 1.2846510671820356e-05,
+      "loss": 1.3795,
+      "step": 12459
+    },
+    {
+      "epoch": 0.9583878163218214,
+      "learning_rate": 1.2799217772405891e-05,
+      "loss": 1.2916,
+      "step": 12460
+    },
+    {
+      "epoch": 0.9584647334820399,
+      "learning_rate": 1.2752011712194922e-05,
+      "loss": 1.0964,
+      "step": 12461
+    },
+    {
+      "epoch": 0.9585416506422583,
+      "learning_rate": 1.2704892493944076e-05,
+      "loss": 1.1146,
+      "step": 12462
+    },
+    {
+      "epoch": 0.9586185678024768,
+      "learning_rate": 1.2657860120404485e-05,
+      "loss": 0.8127,
+      "step": 12463
+    },
+    {
+      "epoch": 0.9586954849626952,
+      "learning_rate": 1.2610914594322453e-05,
+      "loss": 1.0171,
+      "step": 12464
+    },
+    {
+      "epoch": 0.9587724021229136,
+      "learning_rate": 1.2564055918439287e-05,
+      "loss": 1.2993,
+      "step": 12465
+    },
+    {
+      "epoch": 0.9588493192831321,
+      "learning_rate": 1.2517284095491133e-05,
+      "loss": 0.9041,
+      "step": 12466
+    },
+    {
+      "epoch": 0.9589262364433505,
+      "learning_rate": 1.247059912820897e-05,
+      "loss": 1.2157,
+      "step": 12467
+    },
+    {
+      "epoch": 0.9590031536035689,
+      "learning_rate": 1.2424001019318786e-05,
+      "loss": 1.3155,
+      "step": 12468
+    },
+    {
+      "epoch": 0.9590800707637874,
+      "learning_rate": 1.2377489771541406e-05,
+      "loss": 0.9835,
+      "step": 12469
+    },
+    {
+      "epoch": 0.9591569879240058,
+      "learning_rate": 1.2331065387592821e-05,
+      "loss": 0.9406,
+      "step": 12470
+    },
+    {
+      "epoch": 0.9592339050842242,
+      "learning_rate": 1.2284727870183865e-05,
+      "loss": 0.9119,
+      "step": 12471
+    },
+    {
+      "epoch": 0.9593108222444428,
+      "learning_rate": 1.2238477222019872e-05,
+      "loss": 1.0957,
+      "step": 12472
+    },
+    {
+      "epoch": 0.9593877394046612,
+      "learning_rate": 1.219231344580185e-05,
+      "loss": 1.2068,
+      "step": 12473
+    },
+    {
+      "epoch": 0.9594646565648797,
+      "learning_rate": 1.2146236544224975e-05,
+      "loss": 1.003,
+      "step": 12474
+    },
+    {
+      "epoch": 0.9595415737250981,
+      "learning_rate": 1.2100246519980262e-05,
+      "loss": 1.226,
+      "step": 12475
+    },
+    {
+      "epoch": 0.9596184908853165,
+      "learning_rate": 1.205434337575273e-05,
+      "loss": 1.0051,
+      "step": 12476
+    },
+    {
+      "epoch": 0.959695408045535,
+      "learning_rate": 1.2008527114222734e-05,
+      "loss": 1.2554,
+      "step": 12477
+    },
+    {
+      "epoch": 0.9597723252057534,
+      "learning_rate": 1.1962797738065633e-05,
+      "loss": 0.9832,
+      "step": 12478
+    },
+    {
+      "epoch": 0.9598492423659718,
+      "learning_rate": 1.1917155249951461e-05,
+      "loss": 1.0128,
+      "step": 12479
+    },
+    {
+      "epoch": 0.9599261595261903,
+      "learning_rate": 1.1871599652545584e-05,
+      "loss": 1.3611,
+      "step": 12480
+    },
+    {
+      "epoch": 0.9600030766864087,
+      "learning_rate": 1.1826130948507707e-05,
+      "loss": 1.4286,
+      "step": 12481
+    },
+    {
+      "epoch": 0.9600799938466272,
+      "learning_rate": 1.1780749140493041e-05,
+      "loss": 1.0974,
+      "step": 12482
+    },
+    {
+      "epoch": 0.9601569110068456,
+      "learning_rate": 1.1735454231151465e-05,
+      "loss": 1.0853,
+      "step": 12483
+    },
+    {
+      "epoch": 0.960233828167064,
+      "learning_rate": 1.1690246223127865e-05,
+      "loss": 1.3005,
+      "step": 12484
+    },
+    {
+      "epoch": 0.9603107453272826,
+      "learning_rate": 1.1645125119061795e-05,
+      "loss": 0.9851,
+      "step": 12485
+    },
+    {
+      "epoch": 0.960387662487501,
+      "learning_rate": 1.1600090921587813e-05,
+      "loss": 1.1901,
+      "step": 12486
+    },
+    {
+      "epoch": 0.9604645796477194,
+      "learning_rate": 1.1555143633335986e-05,
+      "loss": 1.0553,
+      "step": 12487
+    },
+    {
+      "epoch": 0.9605414968079379,
+      "learning_rate": 1.1510283256930376e-05,
+      "loss": 0.9551,
+      "step": 12488
+    },
+    {
+      "epoch": 0.9606184139681563,
+      "learning_rate": 1.1465509794990559e-05,
+      "loss": 1.1452,
+      "step": 12489
+    },
+    {
+      "epoch": 0.9606953311283747,
+      "learning_rate": 1.1420823250131107e-05,
+      "loss": 1.1404,
+      "step": 12490
+    },
+    {
+      "epoch": 0.9607722482885932,
+      "learning_rate": 1.1376223624960935e-05,
+      "loss": 1.4506,
+      "step": 12491
+    },
+    {
+      "epoch": 0.9608491654488116,
+      "learning_rate": 1.133171092208446e-05,
+      "loss": 1.1455,
+      "step": 12492
+    },
+    {
+      "epoch": 0.9609260826090301,
+      "learning_rate": 1.1287285144100933e-05,
+      "loss": 1.0521,
+      "step": 12493
+    },
+    {
+      "epoch": 0.9610029997692485,
+      "learning_rate": 1.124294629360445e-05,
+      "loss": 1.0328,
+      "step": 12494
+    },
+    {
+      "epoch": 0.9610799169294669,
+      "learning_rate": 1.119869437318377e-05,
+      "loss": 1.0154,
+      "step": 12495
+    },
+    {
+      "epoch": 0.9611568340896854,
+      "learning_rate": 1.115452938542283e-05,
+      "loss": 0.9358,
+      "step": 12496
+    },
+    {
+      "epoch": 0.9612337512499038,
+      "learning_rate": 1.1110451332900729e-05,
+      "loss": 1.2245,
+      "step": 12497
+    },
+    {
+      "epoch": 0.9613106684101222,
+      "learning_rate": 1.106646021819091e-05,
+      "loss": 1.2063,
+      "step": 12498
+    },
+    {
+      "epoch": 0.9613875855703408,
+      "learning_rate": 1.102255604386232e-05,
+      "loss": 1.1398,
+      "step": 12499
+    },
+    {
+      "epoch": 0.9614645027305592,
+      "learning_rate": 1.0978738812478405e-05,
+      "loss": 1.2032,
+      "step": 12500
+    },
+    {
+      "epoch": 0.9615414198907777,
+      "learning_rate": 1.0935008526597623e-05,
+      "loss": 1.3763,
+      "step": 12501
+    },
+    {
+      "epoch": 0.9616183370509961,
+      "learning_rate": 1.089136518877376e-05,
+      "loss": 1.1248,
+      "step": 12502
+    },
+    {
+      "epoch": 0.9616952542112145,
+      "learning_rate": 1.0847808801554949e-05,
+      "loss": 1.2183,
+      "step": 12503
+    },
+    {
+      "epoch": 0.961772171371433,
+      "learning_rate": 1.0804339367484484e-05,
+      "loss": 1.0694,
+      "step": 12504
+    },
+    {
+      "epoch": 0.9618490885316514,
+      "learning_rate": 1.076095688910067e-05,
+      "loss": 1.0625,
+      "step": 12505
+    },
+    {
+      "epoch": 0.9619260056918698,
+      "learning_rate": 1.071766136893665e-05,
+      "loss": 1.0587,
+      "step": 12506
+    },
+    {
+      "epoch": 0.9620029228520883,
+      "learning_rate": 1.0674452809520397e-05,
+      "loss": 0.867,
+      "step": 12507
+    },
+    {
+      "epoch": 0.9620798400123067,
+      "learning_rate": 1.0631331213375061e-05,
+      "loss": 1.0942,
+      "step": 12508
+    },
+    {
+      "epoch": 0.9621567571725252,
+      "learning_rate": 1.0588296583018464e-05,
+      "loss": 1.0028,
+      "step": 12509
+    },
+    {
+      "epoch": 0.9622336743327436,
+      "learning_rate": 1.0545348920963426e-05,
+      "loss": 1.3383,
+      "step": 12510
+    },
+    {
+      "epoch": 0.962310591492962,
+      "learning_rate": 1.0502488229717777e-05,
+      "loss": 1.1583,
+      "step": 12511
+    },
+    {
+      "epoch": 0.9623875086531806,
+      "learning_rate": 1.045971451178418e-05,
+      "loss": 0.9778,
+      "step": 12512
+    },
+    {
+      "epoch": 0.962464425813399,
+      "learning_rate": 1.0417027769660137e-05,
+      "loss": 1.3648,
+      "step": 12513
+    },
+    {
+      "epoch": 0.9625413429736174,
+      "learning_rate": 1.037442800583832e-05,
+      "loss": 1.3772,
+      "step": 12514
+    },
+    {
+      "epoch": 0.9626182601338359,
+      "learning_rate": 1.0331915222806077e-05,
+      "loss": 1.37,
+      "step": 12515
+    },
+    {
+      "epoch": 0.9626951772940543,
+      "learning_rate": 1.028948942304575e-05,
+      "loss": 1.0969,
+      "step": 12516
+    },
+    {
+      "epoch": 0.9627720944542727,
+      "learning_rate": 1.024715060903486e-05,
+      "loss": 1.1798,
+      "step": 12517
+    },
+    {
+      "epoch": 0.9628490116144912,
+      "learning_rate": 1.0204898783245097e-05,
+      "loss": 0.8726,
+      "step": 12518
+    },
+    {
+      "epoch": 0.9629259287747096,
+      "learning_rate": 1.0162733948144321e-05,
+      "loss": 1.2034,
+      "step": 12519
+    },
+    {
+      "epoch": 0.9630028459349281,
+      "learning_rate": 1.0120656106193894e-05,
+      "loss": 1.0849,
+      "step": 12520
+    },
+    {
+      "epoch": 0.9630797630951465,
+      "learning_rate": 1.0078665259851183e-05,
+      "loss": 0.9967,
+      "step": 12521
+    },
+    {
+      "epoch": 0.9631566802553649,
+      "learning_rate": 1.0036761411567897e-05,
+      "loss": 0.8526,
+      "step": 12522
+    },
+    {
+      "epoch": 0.9632335974155835,
+      "learning_rate": 9.994944563790908e-06,
+      "loss": 1.4954,
+      "step": 12523
+    },
+    {
+      "epoch": 0.9633105145758019,
+      "learning_rate": 9.9532147189621e-06,
+      "loss": 1.2077,
+      "step": 12524
+    },
+    {
+      "epoch": 0.9633874317360203,
+      "learning_rate": 9.911571879517856e-06,
+      "loss": 1.0329,
+      "step": 12525
+    },
+    {
+      "epoch": 0.9634643488962388,
+      "learning_rate": 9.87001604788973e-06,
+      "loss": 1.0367,
+      "step": 12526
+    },
+    {
+      "epoch": 0.9635412660564572,
+      "learning_rate": 9.828547226504614e-06,
+      "loss": 0.7542,
+      "step": 12527
+    },
+    {
+      "epoch": 0.9636181832166757,
+      "learning_rate": 9.787165417783405e-06,
+      "loss": 1.1541,
+      "step": 12528
+    },
+    {
+      "epoch": 0.9636951003768941,
+      "learning_rate": 9.745870624142838e-06,
+      "loss": 1.4798,
+      "step": 12529
+    },
+    {
+      "epoch": 0.9637720175371125,
+      "learning_rate": 9.704662847993984e-06,
+      "loss": 1.0329,
+      "step": 12530
+    },
+    {
+      "epoch": 0.963848934697331,
+      "learning_rate": 9.663542091742916e-06,
+      "loss": 1.0714,
+      "step": 12531
+    },
+    {
+      "epoch": 0.9639258518575494,
+      "learning_rate": 9.622508357790883e-06,
+      "loss": 1.1419,
+      "step": 12532
+    },
+    {
+      "epoch": 0.9640027690177678,
+      "learning_rate": 9.581561648533798e-06,
+      "loss": 1.1956,
+      "step": 12533
+    },
+    {
+      "epoch": 0.9640796861779863,
+      "learning_rate": 9.540701966362586e-06,
+      "loss": 1.0352,
+      "step": 12534
+    },
+    {
+      "epoch": 0.9641566033382047,
+      "learning_rate": 9.49992931366317e-06,
+      "loss": 1.0903,
+      "step": 12535
+    },
+    {
+      "epoch": 0.9642335204984231,
+      "learning_rate": 9.45924369281631e-06,
+      "loss": 1.1682,
+      "step": 12536
+    },
+    {
+      "epoch": 0.9643104376586417,
+      "learning_rate": 9.418645106197443e-06,
+      "loss": 1.2397,
+      "step": 12537
+    },
+    {
+      "epoch": 0.9643873548188601,
+      "learning_rate": 9.378133556177504e-06,
+      "loss": 1.0185,
+      "step": 12538
+    },
+    {
+      "epoch": 0.9644642719790786,
+      "learning_rate": 9.337709045121601e-06,
+      "loss": 0.9796,
+      "step": 12539
+    },
+    {
+      "epoch": 0.964541189139297,
+      "learning_rate": 9.297371575390512e-06,
+      "loss": 1.3726,
+      "step": 12540
+    },
+    {
+      "epoch": 0.9646181062995154,
+      "learning_rate": 9.25712114933952e-06,
+      "loss": 0.9965,
+      "step": 12541
+    },
+    {
+      "epoch": 0.9646950234597339,
+      "learning_rate": 9.216957769318912e-06,
+      "loss": 1.0806,
+      "step": 12542
+    },
+    {
+      "epoch": 0.9647719406199523,
+      "learning_rate": 9.176881437673645e-06,
+      "loss": 1.4456,
+      "step": 12543
+    },
+    {
+      "epoch": 0.9648488577801707,
+      "learning_rate": 9.136892156744015e-06,
+      "loss": 1.9841,
+      "step": 12544
+    },
+    {
+      "epoch": 0.9649257749403892,
+      "learning_rate": 9.096989928864984e-06,
+      "loss": 1.0315,
+      "step": 12545
+    },
+    {
+      "epoch": 0.9650026921006076,
+      "learning_rate": 9.057174756366526e-06,
+      "loss": 1.4377,
+      "step": 12546
+    },
+    {
+      "epoch": 0.9650796092608261,
+      "learning_rate": 9.017446641573446e-06,
+      "loss": 1.2672,
+      "step": 12547
+    },
+    {
+      "epoch": 0.9651565264210445,
+      "learning_rate": 8.977805586805555e-06,
+      "loss": 1.2858,
+      "step": 12548
+    },
+    {
+      "epoch": 0.965233443581263,
+      "learning_rate": 8.938251594377334e-06,
+      "loss": 1.4411,
+      "step": 12549
+    },
+    {
+      "epoch": 0.9653103607414815,
+      "learning_rate": 8.89878466659877e-06,
+      "loss": 1.3483,
+      "step": 12550
+    },
+    {
+      "epoch": 0.9653872779016999,
+      "learning_rate": 8.859404805774018e-06,
+      "loss": 1.1452,
+      "step": 12551
+    },
+    {
+      "epoch": 0.9654641950619183,
+      "learning_rate": 8.820112014202742e-06,
+      "loss": 1.4143,
+      "step": 12552
+    },
+    {
+      "epoch": 0.9655411122221368,
+      "learning_rate": 8.780906294179104e-06,
+      "loss": 1.2448,
+      "step": 12553
+    },
+    {
+      "epoch": 0.9656180293823552,
+      "learning_rate": 8.741787647992439e-06,
+      "loss": 1.277,
+      "step": 12554
+    },
+    {
+      "epoch": 0.9656949465425736,
+      "learning_rate": 8.702756077927087e-06,
+      "loss": 1.1117,
+      "step": 12555
+    },
+    {
+      "epoch": 0.9657718637027921,
+      "learning_rate": 8.663811586261894e-06,
+      "loss": 1.2591,
+      "step": 12556
+    },
+    {
+      "epoch": 0.9658487808630105,
+      "learning_rate": 8.624954175271038e-06,
+      "loss": 1.0719,
+      "step": 12557
+    },
+    {
+      "epoch": 0.965925698023229,
+      "learning_rate": 8.586183847223371e-06,
+      "loss": 1.2018,
+      "step": 12558
+    },
+    {
+      "epoch": 0.9660026151834474,
+      "learning_rate": 8.547500604382752e-06,
+      "loss": 1.4612,
+      "step": 12559
+    },
+    {
+      "epoch": 0.9660795323436658,
+      "learning_rate": 8.508904449007871e-06,
+      "loss": 1.2367,
+      "step": 12560
+    },
+    {
+      "epoch": 0.9661564495038844,
+      "learning_rate": 8.470395383352425e-06,
+      "loss": 1.4056,
+      "step": 12561
+    },
+    {
+      "epoch": 0.9662333666641028,
+      "learning_rate": 8.431973409665116e-06,
+      "loss": 1.1051,
+      "step": 12562
+    },
+    {
+      "epoch": 0.9663102838243212,
+      "learning_rate": 8.393638530189319e-06,
+      "loss": 1.113,
+      "step": 12563
+    },
+    {
+      "epoch": 0.9663872009845397,
+      "learning_rate": 8.355390747163405e-06,
+      "loss": 1.101,
+      "step": 12564
+    },
+    {
+      "epoch": 0.9664641181447581,
+      "learning_rate": 8.317230062820925e-06,
+      "loss": 1.1721,
+      "step": 12565
+    },
+    {
+      "epoch": 0.9665410353049766,
+      "learning_rate": 8.279156479389926e-06,
+      "loss": 1.3101,
+      "step": 12566
+    },
+    {
+      "epoch": 0.966617952465195,
+      "learning_rate": 8.241169999093468e-06,
+      "loss": 1.8327,
+      "step": 12567
+    },
+    {
+      "epoch": 0.9666948696254134,
+      "learning_rate": 8.203270624150106e-06,
+      "loss": 1.2764,
+      "step": 12568
+    },
+    {
+      "epoch": 0.9667717867856319,
+      "learning_rate": 8.165458356772238e-06,
+      "loss": 0.9915,
+      "step": 12569
+    },
+    {
+      "epoch": 0.9668487039458503,
+      "learning_rate": 8.127733199167931e-06,
+      "loss": 1.4001,
+      "step": 12570
+    },
+    {
+      "epoch": 0.9669256211060687,
+      "learning_rate": 8.090095153540255e-06,
+      "loss": 1.1301,
+      "step": 12571
+    },
+    {
+      "epoch": 0.9670025382662872,
+      "learning_rate": 8.052544222086621e-06,
+      "loss": 1.5621,
+      "step": 12572
+    },
+    {
+      "epoch": 0.9670794554265056,
+      "learning_rate": 8.015080406999942e-06,
+      "loss": 1.0522,
+      "step": 12573
+    },
+    {
+      "epoch": 0.967156372586724,
+      "learning_rate": 7.977703710467632e-06,
+      "loss": 1.1197,
+      "step": 12574
+    },
+    {
+      "epoch": 0.9672332897469426,
+      "learning_rate": 7.940414134671948e-06,
+      "loss": 0.9458,
+      "step": 12575
+    },
+    {
+      "epoch": 0.967310206907161,
+      "learning_rate": 7.903211681790645e-06,
+      "loss": 1.316,
+      "step": 12576
+    },
+    {
+      "epoch": 0.9673871240673795,
+      "learning_rate": 7.866096353995823e-06,
+      "loss": 1.4618,
+      "step": 12577
+    },
+    {
+      "epoch": 0.9674640412275979,
+      "learning_rate": 7.829068153454577e-06,
+      "loss": 1.0523,
+      "step": 12578
+    },
+    {
+      "epoch": 0.9675409583878163,
+      "learning_rate": 7.792127082329348e-06,
+      "loss": 1.1657,
+      "step": 12579
+    },
+    {
+      "epoch": 0.9676178755480348,
+      "learning_rate": 7.755273142776742e-06,
+      "loss": 1.1556,
+      "step": 12580
+    },
+    {
+      "epoch": 0.9676947927082532,
+      "learning_rate": 7.718506336949038e-06,
+      "loss": 1.3058,
+      "step": 12581
+    },
+    {
+      "epoch": 0.9677717098684716,
+      "learning_rate": 7.681826666993019e-06,
+      "loss": 1.0612,
+      "step": 12582
+    },
+    {
+      "epoch": 0.9678486270286901,
+      "learning_rate": 7.645234135050305e-06,
+      "loss": 1.1075,
+      "step": 12583
+    },
+    {
+      "epoch": 0.9679255441889085,
+      "learning_rate": 7.608728743257687e-06,
+      "loss": 1.2045,
+      "step": 12584
+    },
+    {
+      "epoch": 0.968002461349127,
+      "learning_rate": 7.572310493746793e-06,
+      "loss": 1.2167,
+      "step": 12585
+    },
+    {
+      "epoch": 0.9680793785093454,
+      "learning_rate": 7.535979388643921e-06,
+      "loss": 1.2357,
+      "step": 12586
+    },
+    {
+      "epoch": 0.9681562956695638,
+      "learning_rate": 7.499735430070709e-06,
+      "loss": 1.1877,
+      "step": 12587
+    },
+    {
+      "epoch": 0.9682332128297824,
+      "learning_rate": 7.463578620143297e-06,
+      "loss": 0.8865,
+      "step": 12588
+    },
+    {
+      "epoch": 0.9683101299900008,
+      "learning_rate": 7.427508960973162e-06,
+      "loss": 0.8846,
+      "step": 12589
+    },
+    {
+      "epoch": 0.9683870471502192,
+      "learning_rate": 7.391526454666286e-06,
+      "loss": 1.0404,
+      "step": 12590
+    },
+    {
+      "epoch": 0.9684639643104377,
+      "learning_rate": 7.355631103323657e-06,
+      "loss": 0.8971,
+      "step": 12591
+    },
+    {
+      "epoch": 0.9685408814706561,
+      "learning_rate": 7.319822909041429e-06,
+      "loss": 1.1847,
+      "step": 12592
+    },
+    {
+      "epoch": 0.9686177986308745,
+      "learning_rate": 7.2841018739102646e-06,
+      "loss": 0.8368,
+      "step": 12593
+    },
+    {
+      "epoch": 0.968694715791093,
+      "learning_rate": 7.2484680000161615e-06,
+      "loss": 1.0476,
+      "step": 12594
+    },
+    {
+      "epoch": 0.9687716329513114,
+      "learning_rate": 7.2129212894396224e-06,
+      "loss": 1.1231,
+      "step": 12595
+    },
+    {
+      "epoch": 0.9688485501115299,
+      "learning_rate": 7.177461744256319e-06,
+      "loss": 1.1153,
+      "step": 12596
+    },
+    {
+      "epoch": 0.9689254672717483,
+      "learning_rate": 7.14208936653693e-06,
+      "loss": 0.981,
+      "step": 12597
+    },
+    {
+      "epoch": 0.9690023844319667,
+      "learning_rate": 7.106804158346636e-06,
+      "loss": 1.3899,
+      "step": 12598
+    },
+    {
+      "epoch": 0.9690793015921852,
+      "learning_rate": 7.071606121745955e-06,
+      "loss": 0.9187,
+      "step": 12599
+    },
+    {
+      "epoch": 0.9691562187524037,
+      "learning_rate": 7.0364952587902435e-06,
+      "loss": 1.1633,
+      "step": 12600
+    },
+    {
+      "epoch": 0.9692331359126221,
+      "learning_rate": 7.001471571529361e-06,
+      "loss": 1.2645,
+      "step": 12601
+    },
+    {
+      "epoch": 0.9693100530728406,
+      "learning_rate": 6.9665350620085055e-06,
+      "loss": 1.2485,
+      "step": 12602
+    },
+    {
+      "epoch": 0.969386970233059,
+      "learning_rate": 6.931685732267712e-06,
+      "loss": 1.1676,
+      "step": 12603
+    },
+    {
+      "epoch": 0.9694638873932775,
+      "learning_rate": 6.8969235843416855e-06,
+      "loss": 1.0614,
+      "step": 12604
+    },
+    {
+      "epoch": 0.9695408045534959,
+      "learning_rate": 6.862248620260636e-06,
+      "loss": 1.1671,
+      "step": 12605
+    },
+    {
+      "epoch": 0.9696177217137143,
+      "learning_rate": 6.827660842048777e-06,
+      "loss": 1.6335,
+      "step": 12606
+    },
+    {
+      "epoch": 0.9696946388739328,
+      "learning_rate": 6.793160251725994e-06,
+      "loss": 1.5102,
+      "step": 12607
+    },
+    {
+      "epoch": 0.9697715560341512,
+      "learning_rate": 6.75874685130684e-06,
+      "loss": 1.2357,
+      "step": 12608
+    },
+    {
+      "epoch": 0.9698484731943696,
+      "learning_rate": 6.72442064280071e-06,
+      "loss": 0.9505,
+      "step": 12609
+    },
+    {
+      "epoch": 0.9699253903545881,
+      "learning_rate": 6.6901816282116665e-06,
+      "loss": 0.8884,
+      "step": 12610
+    },
+    {
+      "epoch": 0.9700023075148065,
+      "learning_rate": 6.6560298095394434e-06,
+      "loss": 1.1603,
+      "step": 12611
+    },
+    {
+      "epoch": 0.9700792246750249,
+      "learning_rate": 6.621965188777945e-06,
+      "loss": 0.9623,
+      "step": 12612
+    },
+    {
+      "epoch": 0.9701561418352435,
+      "learning_rate": 6.587987767916248e-06,
+      "loss": 1.0489,
+      "step": 12613
+    },
+    {
+      "epoch": 0.9702330589954619,
+      "learning_rate": 6.554097548938265e-06,
+      "loss": 1.1253,
+      "step": 12614
+    },
+    {
+      "epoch": 0.9703099761556804,
+      "learning_rate": 6.5202945338229124e-06,
+      "loss": 1.2029,
+      "step": 12615
+    },
+    {
+      "epoch": 0.9703868933158988,
+      "learning_rate": 6.486578724544279e-06,
+      "loss": 1.2329,
+      "step": 12616
+    },
+    {
+      "epoch": 0.9704638104761172,
+      "learning_rate": 6.4529501230706225e-06,
+      "loss": 1.3183,
+      "step": 12617
+    },
+    {
+      "epoch": 0.9705407276363357,
+      "learning_rate": 6.419408731365706e-06,
+      "loss": 0.7932,
+      "step": 12618
+    },
+    {
+      "epoch": 0.9706176447965541,
+      "learning_rate": 6.385954551388129e-06,
+      "loss": 1.4808,
+      "step": 12619
+    },
+    {
+      "epoch": 0.9706945619567725,
+      "learning_rate": 6.35258758509133e-06,
+      "loss": 1.3753,
+      "step": 12620
+    },
+    {
+      "epoch": 0.970771479116991,
+      "learning_rate": 6.319307834423582e-06,
+      "loss": 1.1444,
+      "step": 12621
+    },
+    {
+      "epoch": 0.9708483962772094,
+      "learning_rate": 6.286115301328166e-06,
+      "loss": 1.3147,
+      "step": 12622
+    },
+    {
+      "epoch": 0.9709253134374279,
+      "learning_rate": 6.25300998774303e-06,
+      "loss": 1.4053,
+      "step": 12623
+    },
+    {
+      "epoch": 0.9710022305976463,
+      "learning_rate": 6.2199918956014625e-06,
+      "loss": 1.0932,
+      "step": 12624
+    },
+    {
+      "epoch": 0.9710791477578647,
+      "learning_rate": 6.18706102683142e-06,
+      "loss": 1.3351,
+      "step": 12625
+    },
+    {
+      "epoch": 0.9711560649180833,
+      "learning_rate": 6.154217383355698e-06,
+      "loss": 1.2115,
+      "step": 12626
+    },
+    {
+      "epoch": 0.9712329820783017,
+      "learning_rate": 6.1214609670919295e-06,
+      "loss": 1.1545,
+      "step": 12627
+    },
+    {
+      "epoch": 0.9713098992385201,
+      "learning_rate": 6.088791779953085e-06,
+      "loss": 1.0827,
+      "step": 12628
+    },
+    {
+      "epoch": 0.9713868163987386,
+      "learning_rate": 6.056209823846637e-06,
+      "loss": 1.1002,
+      "step": 12629
+    },
+    {
+      "epoch": 0.971463733558957,
+      "learning_rate": 6.023715100674898e-06,
+      "loss": 0.9341,
+      "step": 12630
+    },
+    {
+      "epoch": 0.9715406507191755,
+      "learning_rate": 5.991307612335517e-06,
+      "loss": 1.0185,
+      "step": 12631
+    },
+    {
+      "epoch": 0.9716175678793939,
+      "learning_rate": 5.958987360720647e-06,
+      "loss": 1.1625,
+      "step": 12632
+    },
+    {
+      "epoch": 0.9716944850396123,
+      "learning_rate": 5.926754347717611e-06,
+      "loss": 1.3745,
+      "step": 12633
+    },
+    {
+      "epoch": 0.9717714021998308,
+      "learning_rate": 5.8946085752084046e-06,
+      "loss": 1.1692,
+      "step": 12634
+    },
+    {
+      "epoch": 0.9718483193600492,
+      "learning_rate": 5.8625500450701914e-06,
+      "loss": 1.182,
+      "step": 12635
+    },
+    {
+      "epoch": 0.9719252365202676,
+      "learning_rate": 5.830578759174809e-06,
+      "loss": 1.4195,
+      "step": 12636
+    },
+    {
+      "epoch": 0.9720021536804861,
+      "learning_rate": 5.7986947193889305e-06,
+      "loss": 1.0184,
+      "step": 12637
+    },
+    {
+      "epoch": 0.9720790708407046,
+      "learning_rate": 5.766897927574733e-06,
+      "loss": 1.1037,
+      "step": 12638
+    },
+    {
+      "epoch": 0.972155988000923,
+      "learning_rate": 5.735188385588564e-06,
+      "loss": 1.1467,
+      "step": 12639
+    },
+    {
+      "epoch": 0.9722329051611415,
+      "learning_rate": 5.703566095282109e-06,
+      "loss": 1.176,
+      "step": 12640
+    },
+    {
+      "epoch": 0.9723098223213599,
+      "learning_rate": 5.6720310585015586e-06,
+      "loss": 1.0016,
+      "step": 12641
+    },
+    {
+      "epoch": 0.9723867394815784,
+      "learning_rate": 5.6405832770886065e-06,
+      "loss": 1.064,
+      "step": 12642
+    },
+    {
+      "epoch": 0.9724636566417968,
+      "learning_rate": 5.609222752879284e-06,
+      "loss": 1.0039,
+      "step": 12643
+    },
+    {
+      "epoch": 0.9725405738020152,
+      "learning_rate": 5.5779494877049585e-06,
+      "loss": 1.4048,
+      "step": 12644
+    },
+    {
+      "epoch": 0.9726174909622337,
+      "learning_rate": 5.5467634833915035e-06,
+      "loss": 1.0396,
+      "step": 12645
+    },
+    {
+      "epoch": 0.9726944081224521,
+      "learning_rate": 5.515664741760129e-06,
+      "loss": 1.2299,
+      "step": 12646
+    },
+    {
+      "epoch": 0.9727713252826705,
+      "learning_rate": 5.4846532646267155e-06,
+      "loss": 1.2927,
+      "step": 12647
+    },
+    {
+      "epoch": 0.972848242442889,
+      "learning_rate": 5.453729053801648e-06,
+      "loss": 1.4449,
+      "step": 12648
+    },
+    {
+      "epoch": 0.9729251596031074,
+      "learning_rate": 5.422892111091149e-06,
+      "loss": 1.2064,
+      "step": 12649
+    },
+    {
+      "epoch": 0.973002076763326,
+      "learning_rate": 5.392142438295445e-06,
+      "loss": 1.3436,
+      "step": 12650
+    },
+    {
+      "epoch": 0.9730789939235444,
+      "learning_rate": 5.361480037210265e-06,
+      "loss": 1.117,
+      "step": 12651
+    },
+    {
+      "epoch": 0.9731559110837628,
+      "learning_rate": 5.330904909625845e-06,
+      "loss": 1.1208,
+      "step": 12652
+    },
+    {
+      "epoch": 0.9732328282439813,
+      "learning_rate": 5.300417057327589e-06,
+      "loss": 1.3434,
+      "step": 12653
+    },
+    {
+      "epoch": 0.9733097454041997,
+      "learning_rate": 5.270016482095908e-06,
+      "loss": 1.2208,
+      "step": 12654
+    },
+    {
+      "epoch": 0.9733866625644181,
+      "learning_rate": 5.239703185705546e-06,
+      "loss": 1.2378,
+      "step": 12655
+    },
+    {
+      "epoch": 0.9734635797246366,
+      "learning_rate": 5.209477169926758e-06,
+      "loss": 1.2216,
+      "step": 12656
+    },
+    {
+      "epoch": 0.973540496884855,
+      "learning_rate": 5.1793384365246274e-06,
+      "loss": 1.2038,
+      "step": 12657
+    },
+    {
+      "epoch": 0.9736174140450734,
+      "learning_rate": 5.149286987258583e-06,
+      "loss": 1.2282,
+      "step": 12658
+    },
+    {
+      "epoch": 0.9736943312052919,
+      "learning_rate": 5.119322823883554e-06,
+      "loss": 1.1535,
+      "step": 12659
+    },
+    {
+      "epoch": 0.9737712483655103,
+      "learning_rate": 5.0894459481493075e-06,
+      "loss": 1.2682,
+      "step": 12660
+    },
+    {
+      "epoch": 0.9738481655257288,
+      "learning_rate": 5.059656361800114e-06,
+      "loss": 1.9436,
+      "step": 12661
+    },
+    {
+      "epoch": 0.9739250826859472,
+      "learning_rate": 5.029954066575748e-06,
+      "loss": 1.5157,
+      "step": 12662
+    },
+    {
+      "epoch": 0.9740019998461656,
+      "learning_rate": 5.000339064210324e-06,
+      "loss": 1.2199,
+      "step": 12663
+    },
+    {
+      "epoch": 0.9740789170063842,
+      "learning_rate": 4.970811356433124e-06,
+      "loss": 0.8043,
+      "step": 12664
+    },
+    {
+      "epoch": 0.9741558341666026,
+      "learning_rate": 4.941370944968271e-06,
+      "loss": 0.9901,
+      "step": 12665
+    },
+    {
+      "epoch": 0.974232751326821,
+      "learning_rate": 4.9120178315350535e-06,
+      "loss": 0.8299,
+      "step": 12666
+    },
+    {
+      "epoch": 0.9743096684870395,
+      "learning_rate": 4.882752017847103e-06,
+      "loss": 1.0728,
+      "step": 12667
+    },
+    {
+      "epoch": 0.9743865856472579,
+      "learning_rate": 4.853573505613551e-06,
+      "loss": 0.8153,
+      "step": 12668
+    },
+    {
+      "epoch": 0.9744635028074764,
+      "learning_rate": 4.824482296537869e-06,
+      "loss": 0.9062,
+      "step": 12669
+    },
+    {
+      "epoch": 0.9745404199676948,
+      "learning_rate": 4.79547839231903e-06,
+      "loss": 1.367,
+      "step": 12670
+    },
+    {
+      "epoch": 0.9746173371279132,
+      "learning_rate": 4.766561794650515e-06,
+      "loss": 0.9308,
+      "step": 12671
+    },
+    {
+      "epoch": 0.9746942542881317,
+      "learning_rate": 4.737732505220638e-06,
+      "loss": 0.8003,
+      "step": 12672
+    },
+    {
+      "epoch": 0.9747711714483501,
+      "learning_rate": 4.708990525713053e-06,
+      "loss": 1.1944,
+      "step": 12673
+    },
+    {
+      "epoch": 0.9748480886085685,
+      "learning_rate": 4.680335857805751e-06,
+      "loss": 1.4942,
+      "step": 12674
+    },
+    {
+      "epoch": 0.974925005768787,
+      "learning_rate": 4.65176850317206e-06,
+      "loss": 1.3383,
+      "step": 12675
+    },
+    {
+      "epoch": 0.9750019229290054,
+      "learning_rate": 4.62328846347998e-06,
+      "loss": 1.0785,
+      "step": 12676
+    },
+    {
+      "epoch": 0.9750788400892239,
+      "learning_rate": 4.5948957403925126e-06,
+      "loss": 1.28,
+      "step": 12677
+    },
+    {
+      "epoch": 0.9751557572494424,
+      "learning_rate": 4.566590335567666e-06,
+      "loss": 1.4324,
+      "step": 12678
+    },
+    {
+      "epoch": 0.9752326744096608,
+      "learning_rate": 4.538372250657952e-06,
+      "loss": 1.1537,
+      "step": 12679
+    },
+    {
+      "epoch": 0.9753095915698793,
+      "learning_rate": 4.5102414873112176e-06,
+      "loss": 1.1609,
+      "step": 12680
+    },
+    {
+      "epoch": 0.9753865087300977,
+      "learning_rate": 4.48219804717015e-06,
+      "loss": 1.3438,
+      "step": 12681
+    },
+    {
+      "epoch": 0.9754634258903161,
+      "learning_rate": 4.454241931872105e-06,
+      "loss": 1.1787,
+      "step": 12682
+    },
+    {
+      "epoch": 0.9755403430505346,
+      "learning_rate": 4.426373143049611e-06,
+      "loss": 1.0784,
+      "step": 12683
+    },
+    {
+      "epoch": 0.975617260210753,
+      "learning_rate": 4.3985916823297e-06,
+      "loss": 0.9945,
+      "step": 12684
+    },
+    {
+      "epoch": 0.9756941773709714,
+      "learning_rate": 4.370897551334741e-06,
+      "loss": 1.4363,
+      "step": 12685
+    },
+    {
+      "epoch": 0.9757710945311899,
+      "learning_rate": 4.343290751681772e-06,
+      "loss": 1.2473,
+      "step": 12686
+    },
+    {
+      "epoch": 0.9758480116914083,
+      "learning_rate": 4.315771284982672e-06,
+      "loss": 0.9725,
+      "step": 12687
+    },
+    {
+      "epoch": 0.9759249288516268,
+      "learning_rate": 4.288339152844656e-06,
+      "loss": 1.2346,
+      "step": 12688
+    },
+    {
+      "epoch": 0.9760018460118453,
+      "learning_rate": 4.260994356869108e-06,
+      "loss": 1.38,
+      "step": 12689
+    },
+    {
+      "epoch": 0.9760787631720637,
+      "learning_rate": 4.233736898653084e-06,
+      "loss": 1.3028,
+      "step": 12690
+    },
+    {
+      "epoch": 0.9761556803322822,
+      "learning_rate": 4.206566779787979e-06,
+      "loss": 1.2133,
+      "step": 12691
+    },
+    {
+      "epoch": 0.9762325974925006,
+      "learning_rate": 4.179484001860356e-06,
+      "loss": 1.0899,
+      "step": 12692
+    },
+    {
+      "epoch": 0.976309514652719,
+      "learning_rate": 4.1524885664514515e-06,
+      "loss": 0.9205,
+      "step": 12693
+    },
+    {
+      "epoch": 0.9763864318129375,
+      "learning_rate": 4.1255804751376695e-06,
+      "loss": 1.0226,
+      "step": 12694
+    },
+    {
+      "epoch": 0.9764633489731559,
+      "learning_rate": 4.098759729490254e-06,
+      "loss": 1.0614,
+      "step": 12695
+    },
+    {
+      "epoch": 0.9765402661333743,
+      "learning_rate": 4.072026331075284e-06,
+      "loss": 1.2945,
+      "step": 12696
+    },
+    {
+      "epoch": 0.9766171832935928,
+      "learning_rate": 4.045380281453681e-06,
+      "loss": 1.066,
+      "step": 12697
+    },
+    {
+      "epoch": 0.9766941004538112,
+      "learning_rate": 4.018821582181364e-06,
+      "loss": 1.1933,
+      "step": 12698
+    },
+    {
+      "epoch": 0.9767710176140297,
+      "learning_rate": 3.992350234809261e-06,
+      "loss": 0.955,
+      "step": 12699
+    },
+    {
+      "epoch": 0.9768479347742481,
+      "learning_rate": 3.965966240882802e-06,
+      "loss": 0.9804,
+      "step": 12700
+    },
+    {
+      "epoch": 0.9769248519344665,
+      "learning_rate": 3.939669601942753e-06,
+      "loss": 1.519,
+      "step": 12701
+    },
+    {
+      "epoch": 0.9770017690946851,
+      "learning_rate": 3.913460319524387e-06,
+      "loss": 0.7899,
+      "step": 12702
+    },
+    {
+      "epoch": 0.9770786862549035,
+      "learning_rate": 3.8873383951584795e-06,
+      "loss": 1.2498,
+      "step": 12703
+    },
+    {
+      "epoch": 0.9771556034151219,
+      "learning_rate": 3.8613038303699775e-06,
+      "loss": 1.149,
+      "step": 12704
+    },
+    {
+      "epoch": 0.9772325205753404,
+      "learning_rate": 3.835356626679165e-06,
+      "loss": 1.2062,
+      "step": 12705
+    },
+    {
+      "epoch": 0.9773094377355588,
+      "learning_rate": 3.8094967856013297e-06,
+      "loss": 1.079,
+      "step": 12706
+    },
+    {
+      "epoch": 0.9773863548957773,
+      "learning_rate": 3.7837243086459303e-06,
+      "loss": 1.4197,
+      "step": 12707
+    },
+    {
+      "epoch": 0.9774632720559957,
+      "learning_rate": 3.7580391973184302e-06,
+      "loss": 1.7759,
+      "step": 12708
+    },
+    {
+      "epoch": 0.9775401892162141,
+      "learning_rate": 3.7324414531184626e-06,
+      "loss": 1.1377,
+      "step": 12709
+    },
+    {
+      "epoch": 0.9776171063764326,
+      "learning_rate": 3.706931077540332e-06,
+      "loss": 1.1648,
+      "step": 12710
+    },
+    {
+      "epoch": 0.977694023536651,
+      "learning_rate": 3.6815080720740136e-06,
+      "loss": 1.1485,
+      "step": 12711
+    },
+    {
+      "epoch": 0.9777709406968694,
+      "learning_rate": 3.6561724382039863e-06,
+      "loss": 1.1104,
+      "step": 12712
+    },
+    {
+      "epoch": 0.9778478578570879,
+      "learning_rate": 3.6309241774094004e-06,
+      "loss": 1.0424,
+      "step": 12713
+    },
+    {
+      "epoch": 0.9779247750173063,
+      "learning_rate": 3.6057632911645766e-06,
+      "loss": 1.3053,
+      "step": 12714
+    },
+    {
+      "epoch": 0.9780016921775248,
+      "learning_rate": 3.5806897809388395e-06,
+      "loss": 1.2633,
+      "step": 12715
+    },
+    {
+      "epoch": 0.9780786093377433,
+      "learning_rate": 3.555703648196018e-06,
+      "loss": 1.0491,
+      "step": 12716
+    },
+    {
+      "epoch": 0.9781555264979617,
+      "learning_rate": 3.5308048943954453e-06,
+      "loss": 1.366,
+      "step": 12717
+    },
+    {
+      "epoch": 0.9782324436581802,
+      "learning_rate": 3.5059935209906247e-06,
+      "loss": 1.0456,
+      "step": 12718
+    },
+    {
+      "epoch": 0.9783093608183986,
+      "learning_rate": 3.4812695294303977e-06,
+      "loss": 1.1009,
+      "step": 12719
+    },
+    {
+      "epoch": 0.978386277978617,
+      "learning_rate": 3.456632921158609e-06,
+      "loss": 1.1536,
+      "step": 12720
+    },
+    {
+      "epoch": 0.9784631951388355,
+      "learning_rate": 3.4320836976134418e-06,
+      "loss": 1.8793,
+      "step": 12721
+    },
+    {
+      "epoch": 0.9785401122990539,
+      "learning_rate": 3.4076218602289156e-06,
+      "loss": 1.0634,
+      "step": 12722
+    },
+    {
+      "epoch": 0.9786170294592723,
+      "learning_rate": 3.383247410432888e-06,
+      "loss": 1.1016,
+      "step": 12723
+    },
+    {
+      "epoch": 0.9786939466194908,
+      "learning_rate": 3.358960349648721e-06,
+      "loss": 1.2406,
+      "step": 12724
+    },
+    {
+      "epoch": 0.9787708637797092,
+      "learning_rate": 3.3347606792947793e-06,
+      "loss": 1.2966,
+      "step": 12725
+    },
+    {
+      "epoch": 0.9788477809399277,
+      "learning_rate": 3.3106484007837667e-06,
+      "loss": 0.7612,
+      "step": 12726
+    },
+    {
+      "epoch": 0.9789246981001462,
+      "learning_rate": 3.2866235155238897e-06,
+      "loss": 1.1185,
+      "step": 12727
+    },
+    {
+      "epoch": 0.9790016152603646,
+      "learning_rate": 3.262686024918027e-06,
+      "loss": 1.002,
+      "step": 12728
+    },
+    {
+      "epoch": 0.9790785324205831,
+      "learning_rate": 3.2388359303635596e-06,
+      "loss": 1.3115,
+      "step": 12729
+    },
+    {
+      "epoch": 0.9791554495808015,
+      "learning_rate": 3.2150732332535404e-06,
+      "loss": 1.2793,
+      "step": 12730
+    },
+    {
+      "epoch": 0.9792323667410199,
+      "learning_rate": 3.1913979349751932e-06,
+      "loss": 1.0122,
+      "step": 12731
+    },
+    {
+      "epoch": 0.9793092839012384,
+      "learning_rate": 3.1678100369110783e-06,
+      "loss": 1.0464,
+      "step": 12732
+    },
+    {
+      "epoch": 0.9793862010614568,
+      "learning_rate": 3.1443095404385945e-06,
+      "loss": 0.8579,
+      "step": 12733
+    },
+    {
+      "epoch": 0.9794631182216753,
+      "learning_rate": 3.120896446929644e-06,
+      "loss": 1.3535,
+      "step": 12734
+    },
+    {
+      "epoch": 0.9795400353818937,
+      "learning_rate": 3.0975707577517997e-06,
+      "loss": 0.8382,
+      "step": 12735
+    },
+    {
+      "epoch": 0.9796169525421121,
+      "learning_rate": 3.0743324742668054e-06,
+      "loss": 1.0523,
+      "step": 12736
+    },
+    {
+      "epoch": 0.9796938697023306,
+      "learning_rate": 3.0511815978314096e-06,
+      "loss": 1.3411,
+      "step": 12737
+    },
+    {
+      "epoch": 0.979770786862549,
+      "learning_rate": 3.0281181297976965e-06,
+      "loss": 1.0219,
+      "step": 12738
+    },
+    {
+      "epoch": 0.9798477040227674,
+      "learning_rate": 3.005142071512257e-06,
+      "loss": 0.8128,
+      "step": 12739
+    },
+    {
+      "epoch": 0.979924621182986,
+      "learning_rate": 2.98225342431685e-06,
+      "loss": 1.1951,
+      "step": 12740
+    },
+    {
+      "epoch": 0.9800015383432044,
+      "learning_rate": 2.95945218954774e-06,
+      "loss": 0.8736,
+      "step": 12741
+    },
+    {
+      "epoch": 0.9800784555034228,
+      "learning_rate": 2.9367383685363625e-06,
+      "loss": 1.0182,
+      "step": 12742
+    },
+    {
+      "epoch": 0.9801553726636413,
+      "learning_rate": 2.9141119626089898e-06,
+      "loss": 1.1931,
+      "step": 12743
+    },
+    {
+      "epoch": 0.9802322898238597,
+      "learning_rate": 2.8915729730868976e-06,
+      "loss": 1.1738,
+      "step": 12744
+    },
+    {
+      "epoch": 0.9803092069840782,
+      "learning_rate": 2.8691214012860343e-06,
+      "loss": 1.3247,
+      "step": 12745
+    },
+    {
+      "epoch": 0.9803861241442966,
+      "learning_rate": 2.8467572485175176e-06,
+      "loss": 1.1912,
+      "step": 12746
+    },
+    {
+      "epoch": 0.980463041304515,
+      "learning_rate": 2.824480516087136e-06,
+      "loss": 1.1418,
+      "step": 12747
+    },
+    {
+      "epoch": 0.9805399584647335,
+      "learning_rate": 2.802291205295682e-06,
+      "loss": 1.3108,
+      "step": 12748
+    },
+    {
+      "epoch": 0.9806168756249519,
+      "learning_rate": 2.780189317438786e-06,
+      "loss": 1.0052,
+      "step": 12749
+    },
+    {
+      "epoch": 0.9806937927851703,
+      "learning_rate": 2.758174853806916e-06,
+      "loss": 1.3417,
+      "step": 12750
+    },
+    {
+      "epoch": 0.9807707099453888,
+      "learning_rate": 2.7362478156857105e-06,
+      "loss": 1.1238,
+      "step": 12751
+    },
+    {
+      "epoch": 0.9808476271056072,
+      "learning_rate": 2.714408204355312e-06,
+      "loss": 1.1047,
+      "step": 12752
+    },
+    {
+      "epoch": 0.9809245442658258,
+      "learning_rate": 2.6926560210911997e-06,
+      "loss": 1.1214,
+      "step": 12753
+    },
+    {
+      "epoch": 0.9810014614260442,
+      "learning_rate": 2.670991267163192e-06,
+      "loss": 1.2733,
+      "step": 12754
+    },
+    {
+      "epoch": 0.9810783785862626,
+      "learning_rate": 2.6494139438366092e-06,
+      "loss": 1.2716,
+      "step": 12755
+    },
+    {
+      "epoch": 0.9811552957464811,
+      "learning_rate": 2.6279240523711113e-06,
+      "loss": 1.1705,
+      "step": 12756
+    },
+    {
+      "epoch": 0.9812322129066995,
+      "learning_rate": 2.6065215940216935e-06,
+      "loss": 0.8963,
+      "step": 12757
+    },
+    {
+      "epoch": 0.9813091300669179,
+      "learning_rate": 2.58520657003819e-06,
+      "loss": 0.8876,
+      "step": 12758
+    },
+    {
+      "epoch": 0.9813860472271364,
+      "learning_rate": 2.563978981664772e-06,
+      "loss": 1.2798,
+      "step": 12759
+    },
+    {
+      "epoch": 0.9814629643873548,
+      "learning_rate": 2.5428388301414474e-06,
+      "loss": 0.9096,
+      "step": 12760
+    },
+    {
+      "epoch": 0.9815398815475732,
+      "learning_rate": 2.5217861167022295e-06,
+      "loss": 0.9567,
+      "step": 12761
+    },
+    {
+      "epoch": 0.9816167987077917,
+      "learning_rate": 2.5008208425766345e-06,
+      "loss": 0.9947,
+      "step": 12762
+    },
+    {
+      "epoch": 0.9816937158680101,
+      "learning_rate": 2.479943008988683e-06,
+      "loss": 1.5009,
+      "step": 12763
+    },
+    {
+      "epoch": 0.9817706330282286,
+      "learning_rate": 2.4591526171575673e-06,
+      "loss": 1.259,
+      "step": 12764
+    },
+    {
+      "epoch": 0.981847550188447,
+      "learning_rate": 2.4384496682973158e-06,
+      "loss": 1.125,
+      "step": 12765
+    },
+    {
+      "epoch": 0.9819244673486655,
+      "learning_rate": 2.417834163616628e-06,
+      "loss": 1.0267,
+      "step": 12766
+    },
+    {
+      "epoch": 0.982001384508884,
+      "learning_rate": 2.3973061043193746e-06,
+      "loss": 1.3648,
+      "step": 12767
+    },
+    {
+      "epoch": 0.9820783016691024,
+      "learning_rate": 2.376865491604263e-06,
+      "loss": 1.1793,
+      "step": 12768
+    },
+    {
+      "epoch": 0.9821552188293208,
+      "learning_rate": 2.3565123266646725e-06,
+      "loss": 1.2886,
+      "step": 12769
+    },
+    {
+      "epoch": 0.9822321359895393,
+      "learning_rate": 2.3362466106891524e-06,
+      "loss": 1.2724,
+      "step": 12770
+    },
+    {
+      "epoch": 0.9823090531497577,
+      "learning_rate": 2.3160683448609223e-06,
+      "loss": 1.0261,
+      "step": 12771
+    },
+    {
+      "epoch": 0.9823859703099762,
+      "learning_rate": 2.295977530358373e-06,
+      "loss": 0.873,
+      "step": 12772
+    },
+    {
+      "epoch": 0.9824628874701946,
+      "learning_rate": 2.2759741683545665e-06,
+      "loss": 1.3855,
+      "step": 12773
+    },
+    {
+      "epoch": 0.982539804630413,
+      "learning_rate": 2.2560582600175684e-06,
+      "loss": 1.3186,
+      "step": 12774
+    },
+    {
+      "epoch": 0.9826167217906315,
+      "learning_rate": 2.2362298065101148e-06,
+      "loss": 1.333,
+      "step": 12775
+    },
+    {
+      "epoch": 0.9826936389508499,
+      "learning_rate": 2.21648880899028e-06,
+      "loss": 1.3524,
+      "step": 12776
+    },
+    {
+      "epoch": 0.9827705561110683,
+      "learning_rate": 2.196835268610642e-06,
+      "loss": 1.0582,
+      "step": 12777
+    },
+    {
+      "epoch": 0.9828474732712869,
+      "learning_rate": 2.1772691865186155e-06,
+      "loss": 1.4895,
+      "step": 12778
+    },
+    {
+      "epoch": 0.9829243904315053,
+      "learning_rate": 2.157790563856954e-06,
+      "loss": 1.1967,
+      "step": 12779
+    },
+    {
+      "epoch": 0.9830013075917237,
+      "learning_rate": 2.1383994017629137e-06,
+      "loss": 1.2466,
+      "step": 12780
+    },
+    {
+      "epoch": 0.9830782247519422,
+      "learning_rate": 2.1190957013685895e-06,
+      "loss": 1.1307,
+      "step": 12781
+    },
+    {
+      "epoch": 0.9831551419121606,
+      "learning_rate": 2.0998794638014128e-06,
+      "loss": 0.887,
+      "step": 12782
+    },
+    {
+      "epoch": 0.9832320590723791,
+      "learning_rate": 2.080750690183486e-06,
+      "loss": 1.3147,
+      "step": 12783
+    },
+    {
+      "epoch": 0.9833089762325975,
+      "learning_rate": 2.061709381631582e-06,
+      "loss": 0.7593,
+      "step": 12784
+    },
+    {
+      "epoch": 0.9833858933928159,
+      "learning_rate": 2.0427555392574793e-06,
+      "loss": 1.0692,
+      "step": 12785
+    },
+    {
+      "epoch": 0.9834628105530344,
+      "learning_rate": 2.023889164167958e-06,
+      "loss": 0.9097,
+      "step": 12786
+    },
+    {
+      "epoch": 0.9835397277132528,
+      "learning_rate": 2.0051102574648037e-06,
+      "loss": 1.2492,
+      "step": 12787
+    },
+    {
+      "epoch": 0.9836166448734712,
+      "learning_rate": 1.9864188202444734e-06,
+      "loss": 1.042,
+      "step": 12788
+    },
+    {
+      "epoch": 0.9836935620336897,
+      "learning_rate": 1.96781485359826e-06,
+      "loss": 0.9379,
+      "step": 12789
+    },
+    {
+      "epoch": 0.9837704791939081,
+      "learning_rate": 1.949298358612628e-06,
+      "loss": 1.2095,
+      "step": 12790
+    },
+    {
+      "epoch": 0.9838473963541267,
+      "learning_rate": 1.930869336368546e-06,
+      "loss": 1.1779,
+      "step": 12791
+    },
+    {
+      "epoch": 0.9839243135143451,
+      "learning_rate": 1.9125277879424864e-06,
+      "loss": 0.7955,
+      "step": 12792
+    },
+    {
+      "epoch": 0.9840012306745635,
+      "learning_rate": 1.8942737144050926e-06,
+      "loss": 1.16,
+      "step": 12793
+    },
+    {
+      "epoch": 0.984078147834782,
+      "learning_rate": 1.8761071168223454e-06,
+      "loss": 0.9418,
+      "step": 12794
+    },
+    {
+      "epoch": 0.9841550649950004,
+      "learning_rate": 1.8580279962548963e-06,
+      "loss": 1.0033,
+      "step": 12795
+    },
+    {
+      "epoch": 0.9842319821552188,
+      "learning_rate": 1.8400363537585674e-06,
+      "loss": 1.109,
+      "step": 12796
+    },
+    {
+      "epoch": 0.9843088993154373,
+      "learning_rate": 1.8221321903840182e-06,
+      "loss": 1.2688,
+      "step": 12797
+    },
+    {
+      "epoch": 0.9843858164756557,
+      "learning_rate": 1.8043155071764129e-06,
+      "loss": 1.2503,
+      "step": 12798
+    },
+    {
+      "epoch": 0.9844627336358741,
+      "learning_rate": 1.786586305176252e-06,
+      "loss": 0.8924,
+      "step": 12799
+    },
+    {
+      "epoch": 0.9845396507960926,
+      "learning_rate": 1.7689445854185416e-06,
+      "loss": 1.4539,
+      "step": 12800
+    },
+    {
+      "epoch": 0.984616567956311,
+      "learning_rate": 1.75139034893379e-06,
+      "loss": 1.037,
+      "step": 12801
+    },
+    {
+      "epoch": 0.9846934851165295,
+      "learning_rate": 1.7339235967466783e-06,
+      "loss": 1.0452,
+      "step": 12802
+    },
+    {
+      "epoch": 0.984770402276748,
+      "learning_rate": 1.7165443298772233e-06,
+      "loss": 0.9585,
+      "step": 12803
+    },
+    {
+      "epoch": 0.9848473194369664,
+      "learning_rate": 1.69925254934028e-06,
+      "loss": 1.0599,
+      "step": 12804
+    },
+    {
+      "epoch": 0.9849242365971849,
+      "learning_rate": 1.6820482561455409e-06,
+      "loss": 0.885,
+      "step": 12805
+    },
+    {
+      "epoch": 0.9850011537574033,
+      "learning_rate": 1.664931451297369e-06,
+      "loss": 1.106,
+      "step": 12806
+    },
+    {
+      "epoch": 0.9850780709176217,
+      "learning_rate": 1.6479021357954649e-06,
+      "loss": 1.3251,
+      "step": 12807
+    },
+    {
+      "epoch": 0.9851549880778402,
+      "learning_rate": 1.6309603106340331e-06,
+      "loss": 1.2526,
+      "step": 12808
+    },
+    {
+      "epoch": 0.9852319052380586,
+      "learning_rate": 1.6141059768024492e-06,
+      "loss": 1.0131,
+      "step": 12809
+    },
+    {
+      "epoch": 0.9853088223982771,
+      "learning_rate": 1.5973391352847588e-06,
+      "loss": 1.4628,
+      "step": 12810
+    },
+    {
+      "epoch": 0.9853857395584955,
+      "learning_rate": 1.5806597870600126e-06,
+      "loss": 1.3573,
+      "step": 12811
+    },
+    {
+      "epoch": 0.9854626567187139,
+      "learning_rate": 1.5640679331022644e-06,
+      "loss": 1.0887,
+      "step": 12812
+    },
+    {
+      "epoch": 0.9855395738789324,
+      "learning_rate": 1.547563574380073e-06,
+      "loss": 1.1108,
+      "step": 12813
+    },
+    {
+      "epoch": 0.9856164910391508,
+      "learning_rate": 1.5311467118575006e-06,
+      "loss": 1.5457,
+      "step": 12814
+    },
+    {
+      "epoch": 0.9856934081993692,
+      "learning_rate": 1.5148173464927806e-06,
+      "loss": 1.2486,
+      "step": 12815
+    },
+    {
+      "epoch": 0.9857703253595878,
+      "learning_rate": 1.4985754792394833e-06,
+      "loss": 1.2806,
+      "step": 12816
+    },
+    {
+      "epoch": 0.9858472425198062,
+      "learning_rate": 1.4824211110461839e-06,
+      "loss": 1.2897,
+      "step": 12817
+    },
+    {
+      "epoch": 0.9859241596800246,
+      "learning_rate": 1.4663542428561272e-06,
+      "loss": 1.2803,
+      "step": 12818
+    },
+    {
+      "epoch": 0.9860010768402431,
+      "learning_rate": 1.45037487560723e-06,
+      "loss": 1.1977,
+      "step": 12819
+    },
+    {
+      "epoch": 0.9860779940004615,
+      "learning_rate": 1.4344830102327456e-06,
+      "loss": 1.0733,
+      "step": 12820
+    },
+    {
+      "epoch": 0.98615491116068,
+      "learning_rate": 1.4186786476604318e-06,
+      "loss": 0.967,
+      "step": 12821
+    },
+    {
+      "epoch": 0.9862318283208984,
+      "learning_rate": 1.402961788813384e-06,
+      "loss": 0.895,
+      "step": 12822
+    },
+    {
+      "epoch": 0.9863087454811168,
+      "learning_rate": 1.3873324346093673e-06,
+      "loss": 1.3207,
+      "step": 12823
+    },
+    {
+      "epoch": 0.9863856626413353,
+      "learning_rate": 1.3717905859604862e-06,
+      "loss": 1.4154,
+      "step": 12824
+    },
+    {
+      "epoch": 0.9864625798015537,
+      "learning_rate": 1.3563362437748472e-06,
+      "loss": 1.1526,
+      "step": 12825
+    },
+    {
+      "epoch": 0.9865394969617721,
+      "learning_rate": 1.3409694089543956e-06,
+      "loss": 1.3104,
+      "step": 12826
+    },
+    {
+      "epoch": 0.9866164141219906,
+      "learning_rate": 1.3256900823967465e-06,
+      "loss": 0.8908,
+      "step": 12827
+    },
+    {
+      "epoch": 0.986693331282209,
+      "learning_rate": 1.3104982649938534e-06,
+      "loss": 1.2367,
+      "step": 12828
+    },
+    {
+      "epoch": 0.9867702484424276,
+      "learning_rate": 1.2953939576330065e-06,
+      "loss": 1.1903,
+      "step": 12829
+    },
+    {
+      "epoch": 0.986847165602646,
+      "learning_rate": 1.2803771611958337e-06,
+      "loss": 1.1108,
+      "step": 12830
+    },
+    {
+      "epoch": 0.9869240827628644,
+      "learning_rate": 1.2654478765594668e-06,
+      "loss": 1.2312,
+      "step": 12831
+    },
+    {
+      "epoch": 0.9870009999230829,
+      "learning_rate": 1.250606104595542e-06,
+      "loss": 1.1009,
+      "step": 12832
+    },
+    {
+      "epoch": 0.9870779170833013,
+      "learning_rate": 1.235851846170699e-06,
+      "loss": 0.9126,
+      "step": 12833
+    },
+    {
+      "epoch": 0.9871548342435197,
+      "learning_rate": 1.221185102146416e-06,
+      "loss": 1.1545,
+      "step": 12834
+    },
+    {
+      "epoch": 0.9872317514037382,
+      "learning_rate": 1.2066058733790075e-06,
+      "loss": 1.1847,
+      "step": 12835
+    },
+    {
+      "epoch": 0.9873086685639566,
+      "learning_rate": 1.1921141607201257e-06,
+      "loss": 1.2383,
+      "step": 12836
+    },
+    {
+      "epoch": 0.9873855857241751,
+      "learning_rate": 1.1777099650154277e-06,
+      "loss": 1.4643,
+      "step": 12837
+    },
+    {
+      "epoch": 0.9874625028843935,
+      "learning_rate": 1.1633932871065733e-06,
+      "loss": 1.1681,
+      "step": 12838
+    },
+    {
+      "epoch": 0.9875394200446119,
+      "learning_rate": 1.149164127829061e-06,
+      "loss": 1.3279,
+      "step": 12839
+    },
+    {
+      "epoch": 0.9876163372048304,
+      "learning_rate": 1.1350224880138926e-06,
+      "loss": 1.1594,
+      "step": 12840
+    },
+    {
+      "epoch": 0.9876932543650488,
+      "learning_rate": 1.1209683684869075e-06,
+      "loss": 1.2254,
+      "step": 12841
+    },
+    {
+      "epoch": 0.9877701715252672,
+      "learning_rate": 1.107001770068783e-06,
+      "loss": 1.0006,
+      "step": 12842
+    },
+    {
+      "epoch": 0.9878470886854858,
+      "learning_rate": 1.0931226935746996e-06,
+      "loss": 1.1089,
+      "step": 12843
+    },
+    {
+      "epoch": 0.9879240058457042,
+      "learning_rate": 1.0793311398155093e-06,
+      "loss": 1.3209,
+      "step": 12844
+    },
+    {
+      "epoch": 0.9880009230059226,
+      "learning_rate": 1.0656271095962345e-06,
+      "loss": 0.9682,
+      "step": 12845
+    },
+    {
+      "epoch": 0.9880778401661411,
+      "learning_rate": 1.0520106037170685e-06,
+      "loss": 1.1248,
+      "step": 12846
+    },
+    {
+      "epoch": 0.9881547573263595,
+      "learning_rate": 1.0384816229732086e-06,
+      "loss": 1.2041,
+      "step": 12847
+    },
+    {
+      "epoch": 0.988231674486578,
+      "learning_rate": 1.0250401681546896e-06,
+      "loss": 1.0256,
+      "step": 12848
+    },
+    {
+      "epoch": 0.9883085916467964,
+      "learning_rate": 1.0116862400462167e-06,
+      "loss": 0.9725,
+      "step": 12849
+    },
+    {
+      "epoch": 0.9883855088070148,
+      "learning_rate": 9.984198394275002e-07,
+      "loss": 0.7537,
+      "step": 12850
+    },
+    {
+      "epoch": 0.9884624259672333,
+      "learning_rate": 9.852409670734197e-07,
+      "loss": 1.165,
+      "step": 12851
+    },
+    {
+      "epoch": 0.9885393431274517,
+      "learning_rate": 9.721496237531934e-07,
+      "loss": 1.3722,
+      "step": 12852
+    },
+    {
+      "epoch": 0.9886162602876701,
+      "learning_rate": 9.59145810231543e-07,
+      "loss": 0.8697,
+      "step": 12853
+    },
+    {
+      "epoch": 0.9886931774478886,
+      "learning_rate": 9.46229527267528e-07,
+      "loss": 1.3825,
+      "step": 12854
+    },
+    {
+      "epoch": 0.988770094608107,
+      "learning_rate": 9.33400775615545e-07,
+      "loss": 1.2608,
+      "step": 12855
+    },
+    {
+      "epoch": 0.9888470117683256,
+      "learning_rate": 9.206595560244946e-07,
+      "loss": 0.8974,
+      "step": 12856
+    },
+    {
+      "epoch": 0.988923928928544,
+      "learning_rate": 9.080058692384485e-07,
+      "loss": 0.872,
+      "step": 12857
+    },
+    {
+      "epoch": 0.9890008460887624,
+      "learning_rate": 8.954397159963157e-07,
+      "loss": 1.4197,
+      "step": 12858
+    },
+    {
+      "epoch": 0.9890777632489809,
+      "learning_rate": 8.829610970318425e-07,
+      "loss": 1.2919,
+      "step": 12859
+    },
+    {
+      "epoch": 0.9891546804091993,
+      "learning_rate": 8.705700130734462e-07,
+      "loss": 1.3476,
+      "step": 12860
+    },
+    {
+      "epoch": 0.9892315975694177,
+      "learning_rate": 8.582664648448813e-07,
+      "loss": 1.3725,
+      "step": 12861
+    },
+    {
+      "epoch": 0.9893085147296362,
+      "learning_rate": 8.460504530645729e-07,
+      "loss": 0.9024,
+      "step": 12862
+    },
+    {
+      "epoch": 0.9893854318898546,
+      "learning_rate": 8.339219784457841e-07,
+      "loss": 0.93,
+      "step": 12863
+    },
+    {
+      "epoch": 0.989462349050073,
+      "learning_rate": 8.218810416966149e-07,
+      "loss": 1.0988,
+      "step": 12864
+    },
+    {
+      "epoch": 0.9895392662102915,
+      "learning_rate": 8.099276435201697e-07,
+      "loss": 1.0911,
+      "step": 12865
+    },
+    {
+      "epoch": 0.9896161833705099,
+      "learning_rate": 7.980617846145566e-07,
+      "loss": 0.839,
+      "step": 12866
+    },
+    {
+      "epoch": 0.9896931005307285,
+      "learning_rate": 7.862834656723882e-07,
+      "loss": 1.2614,
+      "step": 12867
+    },
+    {
+      "epoch": 0.9897700176909469,
+      "learning_rate": 7.74592687381781e-07,
+      "loss": 1.2021,
+      "step": 12868
+    },
+    {
+      "epoch": 0.9898469348511653,
+      "learning_rate": 7.629894504250223e-07,
+      "loss": 0.8893,
+      "step": 12869
+    },
+    {
+      "epoch": 0.9899238520113838,
+      "learning_rate": 7.514737554797368e-07,
+      "loss": 1.1556,
+      "step": 12870
+    },
+    {
+      "epoch": 0.9900007691716022,
+      "learning_rate": 7.400456032185532e-07,
+      "loss": 0.9451,
+      "step": 12871
+    },
+    {
+      "epoch": 0.9900776863318206,
+      "learning_rate": 7.287049943084379e-07,
+      "loss": 1.2126,
+      "step": 12872
+    },
+    {
+      "epoch": 0.9901546034920391,
+      "learning_rate": 7.174519294118609e-07,
+      "loss": 0.9153,
+      "step": 12873
+    },
+    {
+      "epoch": 0.9902315206522575,
+      "learning_rate": 7.062864091857967e-07,
+      "loss": 1.0964,
+      "step": 12874
+    },
+    {
+      "epoch": 0.990308437812476,
+      "learning_rate": 6.952084342820574e-07,
+      "loss": 1.3543,
+      "step": 12875
+    },
+    {
+      "epoch": 0.9903853549726944,
+      "learning_rate": 6.84218005347792e-07,
+      "loss": 1.212,
+      "step": 12876
+    },
+    {
+      "epoch": 0.9904622721329128,
+      "learning_rate": 6.733151230244872e-07,
+      "loss": 0.9412,
+      "step": 12877
+    },
+    {
+      "epoch": 0.9905391892931313,
+      "learning_rate": 6.624997879488004e-07,
+      "loss": 1.377,
+      "step": 12878
+    },
+    {
+      "epoch": 0.9906161064533497,
+      "learning_rate": 6.51772000752393e-07,
+      "loss": 0.9764,
+      "step": 12879
+    },
+    {
+      "epoch": 0.9906930236135681,
+      "learning_rate": 6.411317620615975e-07,
+      "loss": 1.0797,
+      "step": 12880
+    },
+    {
+      "epoch": 0.9907699407737867,
+      "learning_rate": 6.305790724977501e-07,
+      "loss": 0.9857,
+      "step": 12881
+    },
+    {
+      "epoch": 0.9908468579340051,
+      "learning_rate": 6.201139326770245e-07,
+      "loss": 0.8654,
+      "step": 12882
+    },
+    {
+      "epoch": 0.9909237750942235,
+      "learning_rate": 6.09736343210432e-07,
+      "loss": 1.3121,
+      "step": 12883
+    },
+    {
+      "epoch": 0.991000692254442,
+      "learning_rate": 5.994463047039878e-07,
+      "loss": 1.084,
+      "step": 12884
+    },
+    {
+      "epoch": 0.9910776094146604,
+      "learning_rate": 5.892438177583781e-07,
+      "loss": 1.2455,
+      "step": 12885
+    },
+    {
+      "epoch": 0.9911545265748789,
+      "learning_rate": 5.791288829696262e-07,
+      "loss": 1.2246,
+      "step": 12886
+    },
+    {
+      "epoch": 0.9912314437350973,
+      "learning_rate": 5.691015009280931e-07,
+      "loss": 1.2855,
+      "step": 12887
+    },
+    {
+      "epoch": 0.9913083608953157,
+      "learning_rate": 5.59161672219477e-07,
+      "loss": 1.3215,
+      "step": 12888
+    },
+    {
+      "epoch": 0.9913852780555342,
+      "learning_rate": 5.493093974239804e-07,
+      "loss": 1.4003,
+      "step": 12889
+    },
+    {
+      "epoch": 0.9914621952157526,
+      "learning_rate": 5.39544677117143e-07,
+      "loss": 1.0938,
+      "step": 12890
+    },
+    {
+      "epoch": 0.991539112375971,
+      "learning_rate": 5.298675118688423e-07,
+      "loss": 1.211,
+      "step": 12891
+    },
+    {
+      "epoch": 0.9916160295361895,
+      "learning_rate": 5.202779022444592e-07,
+      "loss": 1.3325,
+      "step": 12892
+    },
+    {
+      "epoch": 0.991692946696408,
+      "learning_rate": 5.107758488035464e-07,
+      "loss": 1.0643,
+      "step": 12893
+    },
+    {
+      "epoch": 0.9917698638566265,
+      "learning_rate": 5.013613521013261e-07,
+      "loss": 1.3499,
+      "step": 12894
+    },
+    {
+      "epoch": 0.9918467810168449,
+      "learning_rate": 4.920344126873588e-07,
+      "loss": 1.4176,
+      "step": 12895
+    },
+    {
+      "epoch": 0.9919236981770633,
+      "learning_rate": 4.827950311062091e-07,
+      "loss": 0.9883,
+      "step": 12896
+    },
+    {
+      "epoch": 0.9920006153372818,
+      "learning_rate": 4.736432078974451e-07,
+      "loss": 1.3637,
+      "step": 12897
+    },
+    {
+      "epoch": 0.9920775324975002,
+      "learning_rate": 4.6457894359530627e-07,
+      "loss": 1.0593,
+      "step": 12898
+    },
+    {
+      "epoch": 0.9921544496577186,
+      "learning_rate": 4.55602238729369e-07,
+      "loss": 1.0567,
+      "step": 12899
+    },
+    {
+      "epoch": 0.9922313668179371,
+      "learning_rate": 4.46713093823381e-07,
+      "loss": 1.1161,
+      "step": 12900
+    },
+    {
+      "epoch": 0.9923082839781555,
+      "learning_rate": 4.3791150939676003e-07,
+      "loss": 1.0579,
+      "step": 12901
+    },
+    {
+      "epoch": 0.9923852011383739,
+      "learning_rate": 4.291974859632619e-07,
+      "loss": 1.2888,
+      "step": 12902
+    },
+    {
+      "epoch": 0.9924621182985924,
+      "learning_rate": 4.205710240318128e-07,
+      "loss": 1.2811,
+      "step": 12903
+    },
+    {
+      "epoch": 0.9925390354588108,
+      "learning_rate": 4.1203212410600987e-07,
+      "loss": 1.1893,
+      "step": 12904
+    },
+    {
+      "epoch": 0.9926159526190294,
+      "learning_rate": 4.035807866844543e-07,
+      "loss": 1.189,
+      "step": 12905
+    },
+    {
+      "epoch": 0.9926928697792478,
+      "learning_rate": 3.952170122605847e-07,
+      "loss": 1.0412,
+      "step": 12906
+    },
+    {
+      "epoch": 0.9927697869394662,
+      "learning_rate": 3.8694080132301023e-07,
+      "loss": 0.9806,
+      "step": 12907
+    },
+    {
+      "epoch": 0.9928467040996847,
+      "learning_rate": 3.787521543546779e-07,
+      "loss": 0.9748,
+      "step": 12908
+    },
+    {
+      "epoch": 0.9929236212599031,
+      "learning_rate": 3.706510718338718e-07,
+      "loss": 1.4697,
+      "step": 12909
+    },
+    {
+      "epoch": 0.9930005384201215,
+      "learning_rate": 3.626375542337135e-07,
+      "loss": 1.1323,
+      "step": 12910
+    },
+    {
+      "epoch": 0.99307745558034,
+      "learning_rate": 3.5471160202199535e-07,
+      "loss": 1.2568,
+      "step": 12911
+    },
+    {
+      "epoch": 0.9931543727405584,
+      "learning_rate": 3.46873215661514e-07,
+      "loss": 1.2582,
+      "step": 12912
+    },
+    {
+      "epoch": 0.9932312899007769,
+      "learning_rate": 3.3912239561006975e-07,
+      "loss": 1.1581,
+      "step": 12913
+    },
+    {
+      "epoch": 0.9933082070609953,
+      "learning_rate": 3.314591423201341e-07,
+      "loss": 1.2994,
+      "step": 12914
+    },
+    {
+      "epoch": 0.9933851242212137,
+      "learning_rate": 3.238834562393489e-07,
+      "loss": 1.1417,
+      "step": 12915
+    },
+    {
+      "epoch": 0.9934620413814322,
+      "learning_rate": 3.163953378098605e-07,
+      "loss": 1.2175,
+      "step": 12916
+    },
+    {
+      "epoch": 0.9935389585416506,
+      "learning_rate": 3.0899478746881925e-07,
+      "loss": 1.0339,
+      "step": 12917
+    },
+    {
+      "epoch": 0.993615875701869,
+      "learning_rate": 3.016818056487125e-07,
+      "loss": 1.2117,
+      "step": 12918
+    },
+    {
+      "epoch": 0.9936927928620876,
+      "learning_rate": 2.944563927763655e-07,
+      "loss": 1.1594,
+      "step": 12919
+    },
+    {
+      "epoch": 0.993769710022306,
+      "learning_rate": 2.873185492734409e-07,
+      "loss": 1.1495,
+      "step": 12920
+    },
+    {
+      "epoch": 0.9938466271825244,
+      "learning_rate": 2.802682755571051e-07,
+      "loss": 1.4044,
+      "step": 12921
+    },
+    {
+      "epoch": 0.9939235443427429,
+      "learning_rate": 2.7330557203886217e-07,
+      "loss": 1.4963,
+      "step": 12922
+    },
+    {
+      "epoch": 0.9940004615029613,
+      "learning_rate": 2.664304391253869e-07,
+      "loss": 0.805,
+      "step": 12923
+    },
+    {
+      "epoch": 0.9940773786631798,
+      "learning_rate": 2.5964287721785827e-07,
+      "loss": 1.0027,
+      "step": 12924
+    },
+    {
+      "epoch": 0.9941542958233982,
+      "learning_rate": 2.5294288671279255e-07,
+      "loss": 0.9452,
+      "step": 12925
+    },
+    {
+      "epoch": 0.9942312129836166,
+      "learning_rate": 2.4633046800154325e-07,
+      "loss": 0.9951,
+      "step": 12926
+    },
+    {
+      "epoch": 0.9943081301438351,
+      "learning_rate": 2.398056214699684e-07,
+      "loss": 1.0305,
+      "step": 12927
+    },
+    {
+      "epoch": 0.9943850473040535,
+      "learning_rate": 2.3336834749909663e-07,
+      "loss": 1.3752,
+      "step": 12928
+    },
+    {
+      "epoch": 0.9944619644642719,
+      "learning_rate": 2.270186464649604e-07,
+      "loss": 1.1421,
+      "step": 12929
+    },
+    {
+      "epoch": 0.9945388816244904,
+      "learning_rate": 2.2075651873809667e-07,
+      "loss": 1.1417,
+      "step": 12930
+    },
+    {
+      "epoch": 0.9946157987847088,
+      "learning_rate": 2.1458196468437942e-07,
+      "loss": 1.2761,
+      "step": 12931
+    },
+    {
+      "epoch": 0.9946927159449274,
+      "learning_rate": 2.0849498466435357e-07,
+      "loss": 1.4439,
+      "step": 12932
+    },
+    {
+      "epoch": 0.9947696331051458,
+      "learning_rate": 2.02495579033235e-07,
+      "loss": 1.2078,
+      "step": 12933
+    },
+    {
+      "epoch": 0.9948465502653642,
+      "learning_rate": 1.965837481414101e-07,
+      "loss": 1.1953,
+      "step": 12934
+    },
+    {
+      "epoch": 0.9949234674255827,
+      "learning_rate": 1.9075949233426925e-07,
+      "loss": 1.4175,
+      "step": 12935
+    },
+    {
+      "epoch": 0.9950003845858011,
+      "learning_rate": 1.850228119515407e-07,
+      "loss": 1.3363,
+      "step": 12936
+    },
+    {
+      "epoch": 0.9950773017460195,
+      "learning_rate": 1.793737073286228e-07,
+      "loss": 0.9397,
+      "step": 12937
+    },
+    {
+      "epoch": 0.995154218906238,
+      "learning_rate": 1.7381217879508527e-07,
+      "loss": 1.2953,
+      "step": 12938
+    },
+    {
+      "epoch": 0.9952311360664564,
+      "learning_rate": 1.6833822667566834e-07,
+      "loss": 1.5886,
+      "step": 12939
+    },
+    {
+      "epoch": 0.9953080532266748,
+      "learning_rate": 1.6295185129011625e-07,
+      "loss": 1.1365,
+      "step": 12940
+    },
+    {
+      "epoch": 0.9953849703868933,
+      "learning_rate": 1.5765305295284416e-07,
+      "loss": 1.2621,
+      "step": 12941
+    },
+    {
+      "epoch": 0.9954618875471117,
+      "learning_rate": 1.5244183197327122e-07,
+      "loss": 1.1917,
+      "step": 12942
+    },
+    {
+      "epoch": 0.9955388047073302,
+      "learning_rate": 1.473181886558206e-07,
+      "loss": 1.0094,
+      "step": 12943
+    },
+    {
+      "epoch": 0.9956157218675487,
+      "learning_rate": 1.4228212329958634e-07,
+      "loss": 1.0105,
+      "step": 12944
+    },
+    {
+      "epoch": 0.9956926390277671,
+      "learning_rate": 1.3733363619850003e-07,
+      "loss": 1.1703,
+      "step": 12945
+    },
+    {
+      "epoch": 0.9957695561879856,
+      "learning_rate": 1.3247272764166373e-07,
+      "loss": 0.9972,
+      "step": 12946
+    },
+    {
+      "epoch": 0.995846473348204,
+      "learning_rate": 1.27699397913017e-07,
+      "loss": 1.2288,
+      "step": 12947
+    },
+    {
+      "epoch": 0.9959233905084224,
+      "learning_rate": 1.2301364729100374e-07,
+      "loss": 1.133,
+      "step": 12948
+    },
+    {
+      "epoch": 0.9960003076686409,
+      "learning_rate": 1.1841547604940494e-07,
+      "loss": 1.0169,
+      "step": 12949
+    },
+    {
+      "epoch": 0.9960772248288593,
+      "learning_rate": 1.1390488445650604e-07,
+      "loss": 1.1699,
+      "step": 12950
+    },
+    {
+      "epoch": 0.9961541419890778,
+      "learning_rate": 1.09481872776096e-07,
+      "loss": 1.3666,
+      "step": 12951
+    },
+    {
+      "epoch": 0.9962310591492962,
+      "learning_rate": 1.0514644126596862e-07,
+      "loss": 1.343,
+      "step": 12952
+    },
+    {
+      "epoch": 0.9963079763095146,
+      "learning_rate": 1.008985901795878e-07,
+      "loss": 1.0168,
+      "step": 12953
+    },
+    {
+      "epoch": 0.9963848934697331,
+      "learning_rate": 9.673831976492187e-08,
+      "loss": 0.9452,
+      "step": 12954
+    },
+    {
+      "epoch": 0.9964618106299515,
+      "learning_rate": 9.26656302647766e-08,
+      "loss": 1.0083,
+      "step": 12955
+    },
+    {
+      "epoch": 0.9965387277901699,
+      "learning_rate": 8.868052191712828e-08,
+      "loss": 1.3389,
+      "step": 12956
+    },
+    {
+      "epoch": 0.9966156449503885,
+      "learning_rate": 8.478299495462416e-08,
+      "loss": 1.0279,
+      "step": 12957
+    },
+    {
+      "epoch": 0.9966925621106069,
+      "learning_rate": 8.097304960474894e-08,
+      "loss": 0.9121,
+      "step": 12958
+    },
+    {
+      "epoch": 0.9967694792708254,
+      "learning_rate": 7.72506860899913e-08,
+      "loss": 1.2067,
+      "step": 12959
+    },
+    {
+      "epoch": 0.9968463964310438,
+      "learning_rate": 7.361590462784395e-08,
+      "loss": 1.0315,
+      "step": 12960
+    },
+    {
+      "epoch": 0.9969233135912622,
+      "learning_rate": 7.00687054304705e-08,
+      "loss": 1.1543,
+      "step": 12961
+    },
+    {
+      "epoch": 0.9970002307514807,
+      "learning_rate": 6.660908870487203e-08,
+      "loss": 1.5518,
+      "step": 12962
+    },
+    {
+      "epoch": 0.9970771479116991,
+      "learning_rate": 6.323705465322016e-08,
+      "loss": 1.4173,
+      "step": 12963
+    },
+    {
+      "epoch": 0.9971540650719175,
+      "learning_rate": 5.995260347235742e-08,
+      "loss": 1.4398,
+      "step": 12964
+    },
+    {
+      "epoch": 0.997230982232136,
+      "learning_rate": 5.675573535413037e-08,
+      "loss": 1.5524,
+      "step": 12965
+    },
+    {
+      "epoch": 0.9973078993923544,
+      "learning_rate": 5.3646450485056455e-08,
+      "loss": 1.7572,
+      "step": 12966
+    },
+    {
+      "epoch": 0.9973848165525728,
+      "learning_rate": 5.062474904665715e-08,
+      "loss": 1.2099,
+      "step": 12967
+    },
+    {
+      "epoch": 0.9974617337127913,
+      "learning_rate": 4.7690631215624444e-08,
+      "loss": 1.2903,
+      "step": 12968
+    },
+    {
+      "epoch": 0.9975386508730097,
+      "learning_rate": 4.484409716315474e-08,
+      "loss": 0.8052,
+      "step": 12969
+    },
+    {
+      "epoch": 0.9976155680332283,
+      "learning_rate": 4.208514705544841e-08,
+      "loss": 1.1351,
+      "step": 12970
+    },
+    {
+      "epoch": 0.9976924851934467,
+      "learning_rate": 3.941378105354332e-08,
+      "loss": 1.1921,
+      "step": 12971
+    },
+    {
+      "epoch": 0.9977694023536651,
+      "learning_rate": 3.682999931364783e-08,
+      "loss": 1.4235,
+      "step": 12972
+    },
+    {
+      "epoch": 0.9978463195138836,
+      "learning_rate": 3.43338019863082e-08,
+      "loss": 1.1581,
+      "step": 12973
+    },
+    {
+      "epoch": 0.997923236674102,
+      "learning_rate": 3.192518921740772e-08,
+      "loss": 0.9665,
+      "step": 12974
+    },
+    {
+      "epoch": 0.9980001538343204,
+      "learning_rate": 2.960416114783371e-08,
+      "loss": 1.4974,
+      "step": 12975
+    },
+    {
+      "epoch": 0.9980770709945389,
+      "learning_rate": 2.7370717912811316e-08,
+      "loss": 1.2694,
+      "step": 12976
+    },
+    {
+      "epoch": 0.9981539881547573,
+      "learning_rate": 2.522485964273624e-08,
+      "loss": 1.3921,
+      "step": 12977
+    },
+    {
+      "epoch": 0.9982309053149758,
+      "learning_rate": 2.316658646317471e-08,
+      "loss": 1.1122,
+      "step": 12978
+    },
+    {
+      "epoch": 0.9983078224751942,
+      "learning_rate": 2.119589849403081e-08,
+      "loss": 1.2248,
+      "step": 12979
+    },
+    {
+      "epoch": 0.9983847396354126,
+      "learning_rate": 1.9312795850545682e-08,
+      "loss": 1.2476,
+      "step": 12980
+    },
+    {
+      "epoch": 0.9984616567956311,
+      "learning_rate": 1.7517278642631416e-08,
+      "loss": 0.8254,
+      "step": 12981
+    },
+    {
+      "epoch": 0.9985385739558496,
+      "learning_rate": 1.5809346975204087e-08,
+      "loss": 0.9032,
+      "step": 12982
+    },
+    {
+      "epoch": 0.998615491116068,
+      "learning_rate": 1.418900094768416e-08,
+      "loss": 1.2986,
+      "step": 12983
+    },
+    {
+      "epoch": 0.9986924082762865,
+      "learning_rate": 1.2656240655162243e-08,
+      "loss": 1.372,
+      "step": 12984
+    },
+    {
+      "epoch": 0.9987693254365049,
+      "learning_rate": 1.1211066186733732e-08,
+      "loss": 0.941,
+      "step": 12985
+    },
+    {
+      "epoch": 0.9988462425967233,
+      "learning_rate": 9.85347762699762e-09,
+      "loss": 1.1172,
+      "step": 12986
+    },
+    {
+      "epoch": 0.9989231597569418,
+      "learning_rate": 8.583475055223832e-09,
+      "loss": 1.1312,
+      "step": 12987
+    },
+    {
+      "epoch": 0.9990000769171602,
+      "learning_rate": 7.4010585455197566e-09,
+      "loss": 1.2059,
+      "step": 12988
+    },
+    {
+      "epoch": 0.9990769940773787,
+      "learning_rate": 6.306228166830241e-09,
+      "loss": 1.466,
+      "step": 12989
+    },
+    {
+      "epoch": 0.9991539112375971,
+      "learning_rate": 5.2989839832706665e-09,
+      "loss": 1.1152,
+      "step": 12990
+    },
+    {
+      "epoch": 0.9992308283978155,
+      "learning_rate": 4.379326053460808e-09,
+      "loss": 1.209,
+      "step": 12991
+    },
+    {
+      "epoch": 0.999307745558034,
+      "learning_rate": 3.547254431357505e-09,
+      "loss": 1.1204,
+      "step": 12992
+    },
+    {
+      "epoch": 0.9993846627182524,
+      "learning_rate": 2.8027691652554587e-09,
+      "loss": 1.1816,
+      "step": 12993
+    },
+    {
+      "epoch": 0.9994615798784708,
+      "learning_rate": 2.145870298786434e-09,
+      "loss": 1.1467,
+      "step": 12994
+    },
+    {
+      "epoch": 0.9995384970386894,
+      "learning_rate": 1.5765578702531258e-09,
+      "loss": 1.1449,
+      "step": 12995
+    },
+    {
+      "epoch": 0.9996154141989078,
+      "learning_rate": 1.0948319129622242e-09,
+      "loss": 1.2414,
+      "step": 12996
+    },
+    {
+      "epoch": 0.9996923313591263,
+      "learning_rate": 7.006924548913496e-10,
+      "loss": 0.8868,
+      "step": 12997
+    },
+    {
+      "epoch": 0.9997692485193447,
+      "learning_rate": 3.9413951935518555e-10,
+      "loss": 1.1303,
+      "step": 12998
+    },
+    {
+      "epoch": 0.9998461656795631,
+      "learning_rate": 1.7517312400627817e-10,
+      "loss": 1.3999,
+      "step": 12999
+    },
+    {
+      "epoch": 0.9999230828397816,
+      "learning_rate": 4.3793281667703356e-11,
+      "loss": 1.2557,
+      "step": 13000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 13001,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.419903693543424e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}