diff --git "a/checkpoint-1033400/trainer_state.json" "b/checkpoint-1033400/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1033400/trainer_state.json"
@@ -0,0 +1,12417 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 28.999985968653974,
+  "eval_steps": 500,
+  "global_step": 1033400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5e-05,
+      "loss": 44.5033,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00015,
+      "loss": 6.0308,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000225,
+      "loss": 5.869,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0003,
+      "loss": 5.7816,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029985942156660604,
+      "loss": 5.7394,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002997188431332121,
+      "loss": 5.4199,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029957826469981816,
+      "loss": 4.4523,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002994376862664242,
+      "loss": 3.9274,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002992971078330303,
+      "loss": 3.6766,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029915652939963635,
+      "loss": 3.4911,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002990159509662424,
+      "loss": 3.3588,
+      "step": 5500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00029887537253284847,
+      "loss": 3.288,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00029873479409945453,
+      "loss": 3.2493,
+      "step": 6500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002985942156660606,
+      "loss": 3.2086,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029845363723266665,
+      "loss": 3.0969,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0002983130587992727,
+      "loss": 3.0552,
+      "step": 8000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002981724803658788,
+      "loss": 3.0411,
+      "step": 8500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029803190193248484,
+      "loss": 2.9908,
+      "step": 9000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0002978913234990909,
+      "loss": 2.9927,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029775074506569696,
+      "loss": 2.9911,
+      "step": 10000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000297610166632303,
+      "loss": 2.9854,
+      "step": 10500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002974695881989091,
+      "loss": 2.9732,
+      "step": 11000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029732900976551515,
+      "loss": 2.9407,
+      "step": 11500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002971884313321212,
+      "loss": 2.9384,
+      "step": 12000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002970478528987273,
+      "loss": 2.9073,
+      "step": 12500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00029690727446533334,
+      "loss": 2.9253,
+      "step": 13000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002967666960319394,
+      "loss": 2.8914,
+      "step": 13500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00029662611759854546,
+      "loss": 2.8701,
+      "step": 14000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002964855391651515,
+      "loss": 2.8428,
+      "step": 14500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0002963449607317576,
+      "loss": 2.8333,
+      "step": 15000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029620438229836365,
+      "loss": 2.8261,
+      "step": 15500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002960638038649697,
+      "loss": 2.8314,
+      "step": 16000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029592322543157577,
+      "loss": 2.8064,
+      "step": 16500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029578264699818183,
+      "loss": 2.8008,
+      "step": 17000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0002956420685647879,
+      "loss": 2.7798,
+      "step": 17500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00029550149013139396,
+      "loss": 2.7898,
+      "step": 18000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000295360911698,
+      "loss": 2.7996,
+      "step": 18500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002952203332646061,
+      "loss": 2.7596,
+      "step": 19000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00029507975483121214,
+      "loss": 2.7718,
+      "step": 19500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002949391763978182,
+      "loss": 2.774,
+      "step": 20000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00029479859796442427,
+      "loss": 2.751,
+      "step": 20500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0002946580195310303,
+      "loss": 2.7666,
+      "step": 21000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0002945174410976364,
+      "loss": 2.7769,
+      "step": 21500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029437686266424245,
+      "loss": 2.7581,
+      "step": 22000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0002942362842308485,
+      "loss": 2.7098,
+      "step": 22500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002940957057974546,
+      "loss": 2.7282,
+      "step": 23000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029395512736406064,
+      "loss": 2.7095,
+      "step": 23500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002938145489306667,
+      "loss": 2.7139,
+      "step": 24000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00029367397049727276,
+      "loss": 2.6969,
+      "step": 24500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002935333920638788,
+      "loss": 2.705,
+      "step": 25000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002933928136304849,
+      "loss": 2.7324,
+      "step": 25500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00029325223519709095,
+      "loss": 2.6955,
+      "step": 26000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000293111656763697,
+      "loss": 2.6963,
+      "step": 26500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00029297107833030307,
+      "loss": 2.678,
+      "step": 27000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029283049989690913,
+      "loss": 2.7222,
+      "step": 27500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0002926899214635152,
+      "loss": 2.7119,
+      "step": 28000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029254934303012126,
+      "loss": 2.7088,
+      "step": 28500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0002924087645967273,
+      "loss": 2.6917,
+      "step": 29000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0002922681861633334,
+      "loss": 2.6737,
+      "step": 29500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00029212760772993944,
+      "loss": 2.6373,
+      "step": 30000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002919870292965455,
+      "loss": 2.6861,
+      "step": 30500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00029184645086315157,
+      "loss": 2.658,
+      "step": 31000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00029170587242975763,
+      "loss": 2.6517,
+      "step": 31500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002915652939963637,
+      "loss": 2.6492,
+      "step": 32000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00029142471556296975,
+      "loss": 2.6422,
+      "step": 32500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002912841371295758,
+      "loss": 2.6518,
+      "step": 33000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.0002911435586961819,
+      "loss": 2.6342,
+      "step": 33500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00029100298026278794,
+      "loss": 2.6301,
+      "step": 34000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000290862401829394,
+      "loss": 2.6546,
+      "step": 34500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029072182339600006,
+      "loss": 2.6329,
+      "step": 35000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0002905812449626061,
+      "loss": 2.6626,
+      "step": 35500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0002904406665292122,
+      "loss": 2.5699,
+      "step": 36000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029030008809581825,
+      "loss": 2.5145,
+      "step": 36500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0002901595096624243,
+      "loss": 2.5691,
+      "step": 37000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00029001893122903037,
+      "loss": 2.5582,
+      "step": 37500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00028987835279563643,
+      "loss": 2.6044,
+      "step": 38000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002897377743622425,
+      "loss": 2.5418,
+      "step": 38500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00028959719592884856,
+      "loss": 2.5534,
+      "step": 39000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0002894566174954546,
+      "loss": 2.5921,
+      "step": 39500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002893160390620607,
+      "loss": 2.5474,
+      "step": 40000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028917546062866674,
+      "loss": 2.5525,
+      "step": 40500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028903488219527275,
+      "loss": 2.5519,
+      "step": 41000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00028889430376187887,
+      "loss": 2.5802,
+      "step": 41500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028875372532848493,
+      "loss": 2.4925,
+      "step": 42000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000288613146895091,
+      "loss": 2.5546,
+      "step": 42500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028847256846169705,
+      "loss": 2.513,
+      "step": 43000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0002883319900283031,
+      "loss": 2.5301,
+      "step": 43500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002881914115949092,
+      "loss": 2.514,
+      "step": 44000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00028805083316151524,
+      "loss": 2.5389,
+      "step": 44500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00028791025472812125,
+      "loss": 2.5552,
+      "step": 45000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00028776967629472736,
+      "loss": 2.5773,
+      "step": 45500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002876290978613334,
+      "loss": 2.5215,
+      "step": 46000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002874885194279395,
+      "loss": 2.5633,
+      "step": 46500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00028734794099454555,
+      "loss": 2.5442,
+      "step": 47000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0002872073625611516,
+      "loss": 2.5673,
+      "step": 47500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028706678412775767,
+      "loss": 2.5251,
+      "step": 48000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028692620569436373,
+      "loss": 2.4993,
+      "step": 48500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00028678562726096974,
+      "loss": 2.5582,
+      "step": 49000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0002866450488275758,
+      "loss": 2.5427,
+      "step": 49500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0002865044703941819,
+      "loss": 2.5727,
+      "step": 50000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.000286363891960788,
+      "loss": 2.5369,
+      "step": 50500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00028622331352739404,
+      "loss": 2.5114,
+      "step": 51000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0002860827350940001,
+      "loss": 2.5438,
+      "step": 51500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00028594215666060617,
+      "loss": 2.5043,
+      "step": 52000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00028580157822721223,
+      "loss": 2.5161,
+      "step": 52500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00028566099979381824,
+      "loss": 2.5474,
+      "step": 53000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002855204213604243,
+      "loss": 2.5154,
+      "step": 53500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0002853798429270304,
+      "loss": 2.4809,
+      "step": 54000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.0002852392644936365,
+      "loss": 2.5301,
+      "step": 54500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028509868606024254,
+      "loss": 2.504,
+      "step": 55000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0002849581076268486,
+      "loss": 2.482,
+      "step": 55500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00028481752919345466,
+      "loss": 2.5116,
+      "step": 56000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0002846769507600607,
+      "loss": 2.4682,
+      "step": 56500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00028453637232666673,
+      "loss": 2.5098,
+      "step": 57000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002843957938932728,
+      "loss": 2.4829,
+      "step": 57500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00028425521545987886,
+      "loss": 2.498,
+      "step": 58000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00028411463702648497,
+      "loss": 2.4492,
+      "step": 58500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00028397405859309103,
+      "loss": 2.4253,
+      "step": 59000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.0002838334801596971,
+      "loss": 2.4648,
+      "step": 59500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00028369290172630316,
+      "loss": 2.4785,
+      "step": 60000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.0002835523232929092,
+      "loss": 2.5128,
+      "step": 60500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0002834117448595153,
+      "loss": 2.4789,
+      "step": 61000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0002832711664261213,
+      "loss": 2.4908,
+      "step": 61500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00028313058799272735,
+      "loss": 2.4679,
+      "step": 62000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00028299000955933347,
+      "loss": 2.4972,
+      "step": 62500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00028284943112593953,
+      "loss": 2.4933,
+      "step": 63000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0002827088526925456,
+      "loss": 2.427,
+      "step": 63500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00028256827425915165,
+      "loss": 2.4344,
+      "step": 64000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002824276958257577,
+      "loss": 2.4604,
+      "step": 64500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002822871173923638,
+      "loss": 2.4777,
+      "step": 65000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002821465389589698,
+      "loss": 2.4681,
+      "step": 65500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00028200596052557585,
+      "loss": 2.4243,
+      "step": 66000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0002818653820921819,
+      "loss": 2.4423,
+      "step": 66500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.000281724803658788,
+      "loss": 2.4525,
+      "step": 67000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002815842252253941,
+      "loss": 2.4513,
+      "step": 67500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00028144364679200015,
+      "loss": 2.4303,
+      "step": 68000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.0002813030683586062,
+      "loss": 2.4732,
+      "step": 68500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00028116248992521227,
+      "loss": 2.4074,
+      "step": 69000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.0002810219114918183,
+      "loss": 2.4309,
+      "step": 69500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00028088133305842434,
+      "loss": 2.4698,
+      "step": 70000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002807407546250304,
+      "loss": 2.4569,
+      "step": 70500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002806001761916365,
+      "loss": 2.4364,
+      "step": 71000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002804595977582426,
+      "loss": 2.4081,
+      "step": 71500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00028031901932484864,
+      "loss": 2.3468,
+      "step": 72000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002801784408914547,
+      "loss": 2.3954,
+      "step": 72500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00028003786245806077,
+      "loss": 2.3745,
+      "step": 73000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002798972840246668,
+      "loss": 2.399,
+      "step": 73500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00027975670559127284,
+      "loss": 2.3572,
+      "step": 74000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002796161271578789,
+      "loss": 2.3931,
+      "step": 74500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00027947554872448496,
+      "loss": 2.3707,
+      "step": 75000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0002793349702910911,
+      "loss": 2.425,
+      "step": 75500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00027919439185769714,
+      "loss": 2.3634,
+      "step": 76000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.0002790538134243032,
+      "loss": 2.3933,
+      "step": 76500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00027891323499090926,
+      "loss": 2.356,
+      "step": 77000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00027877265655751527,
+      "loss": 2.3819,
+      "step": 77500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00027863207812412133,
+      "loss": 2.3687,
+      "step": 78000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002784914996907274,
+      "loss": 2.3515,
+      "step": 78500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00027835092125733346,
+      "loss": 2.3663,
+      "step": 79000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002782103428239396,
+      "loss": 2.3916,
+      "step": 79500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00027806976439054563,
+      "loss": 2.3404,
+      "step": 80000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002779291859571517,
+      "loss": 2.3482,
+      "step": 80500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027778860752375776,
+      "loss": 2.3524,
+      "step": 81000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00027764802909036377,
+      "loss": 2.3759,
+      "step": 81500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00027750745065696983,
+      "loss": 2.3498,
+      "step": 82000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002773668722235759,
+      "loss": 2.3667,
+      "step": 82500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00027722629379018195,
+      "loss": 2.3777,
+      "step": 83000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.000277085715356788,
+      "loss": 2.3955,
+      "step": 83500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00027694513692339413,
+      "loss": 2.3583,
+      "step": 84000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002768045584900002,
+      "loss": 2.376,
+      "step": 84500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00027666398005660625,
+      "loss": 2.3855,
+      "step": 85000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00027652340162321226,
+      "loss": 2.3843,
+      "step": 85500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002763828231898183,
+      "loss": 2.3652,
+      "step": 86000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002762422447564244,
+      "loss": 2.3697,
+      "step": 86500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00027610166632303045,
+      "loss": 2.3478,
+      "step": 87000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0002759610878896365,
+      "loss": 2.3332,
+      "step": 87500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0002758205094562426,
+      "loss": 2.381,
+      "step": 88000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002756799310228487,
+      "loss": 2.3139,
+      "step": 88500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00027553935258945475,
+      "loss": 2.3374,
+      "step": 89000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00027539877415606076,
+      "loss": 2.3663,
+      "step": 89500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002752581957226668,
+      "loss": 2.3629,
+      "step": 90000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002751176172892729,
+      "loss": 2.3424,
+      "step": 90500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00027497703885587894,
+      "loss": 2.3227,
+      "step": 91000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.000274836460422485,
+      "loss": 2.331,
+      "step": 91500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00027469588198909107,
+      "loss": 2.3293,
+      "step": 92000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002745553035556972,
+      "loss": 2.343,
+      "step": 92500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00027441472512230325,
+      "loss": 2.3556,
+      "step": 93000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0002742741466889093,
+      "loss": 2.3014,
+      "step": 93500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002741335682555153,
+      "loss": 2.3458,
+      "step": 94000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0002739929898221214,
+      "loss": 2.3415,
+      "step": 94500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00027385241138872744,
+      "loss": 2.3327,
+      "step": 95000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.0002737118329553335,
+      "loss": 2.3377,
+      "step": 95500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00027357125452193956,
+      "loss": 2.3452,
+      "step": 96000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002734306760885457,
+      "loss": 2.3191,
+      "step": 96500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00027329009765515174,
+      "loss": 2.3126,
+      "step": 97000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002731495192217578,
+      "loss": 2.3646,
+      "step": 97500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002730089407883638,
+      "loss": 2.3326,
+      "step": 98000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00027286836235496987,
+      "loss": 2.3596,
+      "step": 98500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00027272778392157593,
+      "loss": 2.3274,
+      "step": 99000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.000272587205488182,
+      "loss": 2.2924,
+      "step": 99500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00027244662705478806,
+      "loss": 2.3603,
+      "step": 100000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002723060486213941,
+      "loss": 2.354,
+      "step": 100500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00027216547018800024,
+      "loss": 2.3229,
+      "step": 101000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0002720248917546063,
+      "loss": 2.3265,
+      "step": 101500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002718843133212123,
+      "loss": 2.3141,
+      "step": 102000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00027174373488781837,
+      "loss": 2.3071,
+      "step": 102500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00027160315645442443,
+      "loss": 2.3051,
+      "step": 103000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002714625780210305,
+      "loss": 2.3332,
+      "step": 103500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00027132199958763655,
+      "loss": 2.3254,
+      "step": 104000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002711814211542426,
+      "loss": 2.2945,
+      "step": 104500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00027104084272084873,
+      "loss": 2.3055,
+      "step": 105000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002709002642874548,
+      "loss": 2.3189,
+      "step": 105500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002707596858540608,
+      "loss": 2.2937,
+      "step": 106000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00027061910742066686,
+      "loss": 2.283,
+      "step": 106500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0002704785289872729,
+      "loss": 2.3291,
+      "step": 107000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.000270337950553879,
+      "loss": 2.2292,
+      "step": 107500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00027019737212048505,
+      "loss": 2.2265,
+      "step": 108000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0002700567936870911,
+      "loss": 2.2228,
+      "step": 108500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00026991621525369717,
+      "loss": 2.2762,
+      "step": 109000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0002697756368203033,
+      "loss": 2.2504,
+      "step": 109500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002696350583869093,
+      "loss": 2.2539,
+      "step": 110000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00026949447995351536,
+      "loss": 2.2537,
+      "step": 110500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0002693539015201214,
+      "loss": 2.2348,
+      "step": 111000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002692133230867275,
+      "loss": 2.2565,
+      "step": 111500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00026907274465333354,
+      "loss": 2.2512,
+      "step": 112000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002689321662199396,
+      "loss": 2.2837,
+      "step": 112500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00026879158778654567,
+      "loss": 2.2672,
+      "step": 113000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0002686510093531518,
+      "loss": 2.27,
+      "step": 113500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.0002685104309197578,
+      "loss": 2.2441,
+      "step": 114000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00026836985248636385,
+      "loss": 2.2211,
+      "step": 114500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002682292740529699,
+      "loss": 2.2713,
+      "step": 115000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000268088695619576,
+      "loss": 2.2502,
+      "step": 115500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00026794811718618204,
+      "loss": 2.26,
+      "step": 116000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0002678075387527881,
+      "loss": 2.2297,
+      "step": 116500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00026766696031939416,
+      "loss": 2.2647,
+      "step": 117000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002675263818860002,
+      "loss": 2.2684,
+      "step": 117500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0002673858034526063,
+      "loss": 2.2555,
+      "step": 118000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00026724522501921235,
+      "loss": 2.2513,
+      "step": 118500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0002671046465858184,
+      "loss": 2.2653,
+      "step": 119000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002669640681524245,
+      "loss": 2.2498,
+      "step": 119500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00026682348971903054,
+      "loss": 2.2688,
+      "step": 120000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0002666829112856366,
+      "loss": 2.2662,
+      "step": 120500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00026654233285224266,
+      "loss": 2.2642,
+      "step": 121000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002664017544188487,
+      "loss": 2.2346,
+      "step": 121500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00026626117598545484,
+      "loss": 2.2195,
+      "step": 122000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00026612059755206084,
+      "loss": 2.2449,
+      "step": 122500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0002659800191186669,
+      "loss": 2.2538,
+      "step": 123000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00026583944068527297,
+      "loss": 2.2572,
+      "step": 123500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00026569886225187903,
+      "loss": 2.2885,
+      "step": 124000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002655582838184851,
+      "loss": 2.2428,
+      "step": 124500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00026541770538509115,
+      "loss": 2.2527,
+      "step": 125000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0002652771269516972,
+      "loss": 2.2884,
+      "step": 125500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002651365485183033,
+      "loss": 2.2668,
+      "step": 126000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00026499597008490934,
+      "loss": 2.2687,
+      "step": 126500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0002648553916515154,
+      "loss": 2.2512,
+      "step": 127000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00026471481321812146,
+      "loss": 2.2464,
+      "step": 127500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0002645742347847275,
+      "loss": 2.2876,
+      "step": 128000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0002644336563513336,
+      "loss": 2.232,
+      "step": 128500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00026429307791793965,
+      "loss": 2.2532,
+      "step": 129000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002641524994845457,
+      "loss": 2.2571,
+      "step": 129500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002640119210511518,
+      "loss": 2.272,
+      "step": 130000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00026387134261775784,
+      "loss": 2.2685,
+      "step": 130500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.0002637307641843639,
+      "loss": 2.2876,
+      "step": 131000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00026359018575096996,
+      "loss": 2.2326,
+      "step": 131500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.000263449607317576,
+      "loss": 2.2249,
+      "step": 132000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0002633090288841821,
+      "loss": 2.2476,
+      "step": 132500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00026316845045078815,
+      "loss": 2.2363,
+      "step": 133000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002630278720173942,
+      "loss": 2.2405,
+      "step": 133500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00026288729358400027,
+      "loss": 2.2412,
+      "step": 134000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00026274671515060633,
+      "loss": 2.2385,
+      "step": 134500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0002626061367172124,
+      "loss": 2.2333,
+      "step": 135000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00026246555828381846,
+      "loss": 2.2556,
+      "step": 135500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002623249798504245,
+      "loss": 2.2539,
+      "step": 136000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002621844014170306,
+      "loss": 2.2738,
+      "step": 136500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026204382298363664,
+      "loss": 2.2561,
+      "step": 137000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002619032445502427,
+      "loss": 2.22,
+      "step": 137500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00026176266611684876,
+      "loss": 2.2397,
+      "step": 138000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002616220876834548,
+      "loss": 2.2599,
+      "step": 138500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0002614815092500609,
+      "loss": 2.2284,
+      "step": 139000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00026134093081666695,
+      "loss": 2.248,
+      "step": 139500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.000261200352383273,
+      "loss": 2.2485,
+      "step": 140000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0002610597739498791,
+      "loss": 2.28,
+      "step": 140500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00026091919551648514,
+      "loss": 2.209,
+      "step": 141000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0002607786170830912,
+      "loss": 2.2213,
+      "step": 141500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00026063803864969726,
+      "loss": 2.2504,
+      "step": 142000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0002604974602163033,
+      "loss": 2.2513,
+      "step": 142500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0002603568817829094,
+      "loss": 2.1913,
+      "step": 143000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00026021630334951545,
+      "loss": 2.1501,
+      "step": 143500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0002600757249161215,
+      "loss": 2.1712,
+      "step": 144000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00025993514648272757,
+      "loss": 2.1857,
+      "step": 144500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00025979456804933363,
+      "loss": 2.173,
+      "step": 145000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0002596539896159397,
+      "loss": 2.192,
+      "step": 145500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00025951341118254576,
+      "loss": 2.1625,
+      "step": 146000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0002593728327491518,
+      "loss": 2.1566,
+      "step": 146500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0002592322543157579,
+      "loss": 2.2201,
+      "step": 147000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00025909167588236394,
+      "loss": 2.1807,
+      "step": 147500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00025895109744897,
+      "loss": 2.1862,
+      "step": 148000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00025881051901557607,
+      "loss": 2.1699,
+      "step": 148500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00025866994058218213,
+      "loss": 2.1755,
+      "step": 149000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0002585293621487882,
+      "loss": 2.1475,
+      "step": 149500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00025838878371539425,
+      "loss": 2.1974,
+      "step": 150000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002582482052820003,
+      "loss": 2.1115,
+      "step": 150500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0002581076268486064,
+      "loss": 2.1923,
+      "step": 151000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00025796704841521244,
+      "loss": 2.1547,
+      "step": 151500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002578264699818185,
+      "loss": 2.1803,
+      "step": 152000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00025768589154842456,
+      "loss": 2.1682,
+      "step": 152500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0002575453131150306,
+      "loss": 2.1428,
+      "step": 153000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0002574047346816367,
+      "loss": 2.1664,
+      "step": 153500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00025726415624824275,
+      "loss": 2.1683,
+      "step": 154000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0002571235778148488,
+      "loss": 2.1786,
+      "step": 154500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00025698299938145487,
+      "loss": 2.1604,
+      "step": 155000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00025684242094806093,
+      "loss": 2.1749,
+      "step": 155500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.000256701842514667,
+      "loss": 2.162,
+      "step": 156000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00025656126408127306,
+      "loss": 2.1696,
+      "step": 156500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002564206856478791,
+      "loss": 2.1784,
+      "step": 157000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0002562801072144852,
+      "loss": 2.1277,
+      "step": 157500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00025613952878109124,
+      "loss": 2.1739,
+      "step": 158000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0002559989503476973,
+      "loss": 2.1694,
+      "step": 158500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00025585837191430337,
+      "loss": 2.2062,
+      "step": 159000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00025571779348090943,
+      "loss": 2.1801,
+      "step": 159500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0002555772150475155,
+      "loss": 2.1527,
+      "step": 160000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00025543663661412155,
+      "loss": 2.1805,
+      "step": 160500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0002552960581807276,
+      "loss": 2.1429,
+      "step": 161000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0002551554797473337,
+      "loss": 2.181,
+      "step": 161500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00025501490131393974,
+      "loss": 2.2047,
+      "step": 162000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002548743228805458,
+      "loss": 2.1467,
+      "step": 162500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00025473374444715186,
+      "loss": 2.1619,
+      "step": 163000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0002545931660137579,
+      "loss": 2.2002,
+      "step": 163500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.000254452587580364,
+      "loss": 2.1779,
+      "step": 164000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00025431200914697005,
+      "loss": 2.175,
+      "step": 164500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0002541714307135761,
+      "loss": 2.1816,
+      "step": 165000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00025403085228018217,
+      "loss": 2.1696,
+      "step": 165500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00025389027384678823,
+      "loss": 2.2045,
+      "step": 166000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002537496954133943,
+      "loss": 2.1508,
+      "step": 166500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00025360911698000036,
+      "loss": 2.1918,
+      "step": 167000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0002534685385466064,
+      "loss": 2.1761,
+      "step": 167500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.0002533279601132125,
+      "loss": 2.155,
+      "step": 168000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00025318738167981854,
+      "loss": 2.1483,
+      "step": 168500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002530468032464246,
+      "loss": 2.1618,
+      "step": 169000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00025290622481303067,
+      "loss": 2.147,
+      "step": 169500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00025276564637963673,
+      "loss": 2.1718,
+      "step": 170000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002526250679462428,
+      "loss": 2.1329,
+      "step": 170500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00025248448951284885,
+      "loss": 2.1677,
+      "step": 171000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0002523439110794549,
+      "loss": 2.1127,
+      "step": 171500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.000252203332646061,
+      "loss": 2.1832,
+      "step": 172000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00025206275421266704,
+      "loss": 2.1843,
+      "step": 172500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0002519221757792731,
+      "loss": 2.2132,
+      "step": 173000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00025178159734587916,
+      "loss": 2.1789,
+      "step": 173500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0002516410189124852,
+      "loss": 2.1333,
+      "step": 174000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0002515004404790913,
+      "loss": 2.1702,
+      "step": 174500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00025135986204569735,
+      "loss": 2.1817,
+      "step": 175000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002512192836123034,
+      "loss": 2.1101,
+      "step": 175500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00025107870517890947,
+      "loss": 2.1573,
+      "step": 176000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00025093812674551553,
+      "loss": 2.1691,
+      "step": 176500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0002507975483121216,
+      "loss": 2.1822,
+      "step": 177000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00025065696987872766,
+      "loss": 2.1869,
+      "step": 177500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002505163914453337,
+      "loss": 2.1731,
+      "step": 178000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002503758130119398,
+      "loss": 2.1219,
+      "step": 178500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00025023523457854584,
+      "loss": 2.0762,
+      "step": 179000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0002500946561451519,
+      "loss": 2.1061,
+      "step": 179500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00024995407771175797,
+      "loss": 2.0847,
+      "step": 180000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.00024981349927836403,
+      "loss": 2.1121,
+      "step": 180500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002496729208449701,
+      "loss": 2.099,
+      "step": 181000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024953234241157615,
+      "loss": 2.1179,
+      "step": 181500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0002493917639781822,
+      "loss": 2.0985,
+      "step": 182000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002492511855447883,
+      "loss": 2.1106,
+      "step": 182500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00024911060711139434,
+      "loss": 2.0744,
+      "step": 183000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002489700286780004,
+      "loss": 2.0858,
+      "step": 183500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00024882945024460646,
+      "loss": 2.0899,
+      "step": 184000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0002486888718112125,
+      "loss": 2.0637,
+      "step": 184500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002485482933778186,
+      "loss": 2.1075,
+      "step": 185000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.00024840771494442465,
+      "loss": 2.1119,
+      "step": 185500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0002482671365110307,
+      "loss": 2.0968,
+      "step": 186000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024812655807763677,
+      "loss": 2.1092,
+      "step": 186500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00024798597964424283,
+      "loss": 2.0675,
+      "step": 187000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002478454012108489,
+      "loss": 2.0575,
+      "step": 187500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00024770482277745496,
+      "loss": 2.0931,
+      "step": 188000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.000247564244344061,
+      "loss": 2.0929,
+      "step": 188500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0002474236659106671,
+      "loss": 2.0896,
+      "step": 189000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00024728308747727314,
+      "loss": 2.1169,
+      "step": 189500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002471425090438792,
+      "loss": 2.1009,
+      "step": 190000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00024700193061048527,
+      "loss": 2.1346,
+      "step": 190500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00024686135217709133,
+      "loss": 2.0952,
+      "step": 191000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00024672077374369734,
+      "loss": 2.1042,
+      "step": 191500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.00024658019531030345,
+      "loss": 2.1056,
+      "step": 192000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002464396168769095,
+      "loss": 2.0961,
+      "step": 192500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0002462990384435156,
+      "loss": 2.1284,
+      "step": 193000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00024615846001012164,
+      "loss": 2.0954,
+      "step": 193500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0002460178815767277,
+      "loss": 2.1349,
+      "step": 194000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00024587730314333376,
+      "loss": 2.0941,
+      "step": 194500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002457367247099398,
+      "loss": 2.1228,
+      "step": 195000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00024559614627654583,
+      "loss": 2.0905,
+      "step": 195500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00024545556784315195,
+      "loss": 2.0905,
+      "step": 196000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.000245314989409758,
+      "loss": 2.1323,
+      "step": 196500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00024517441097636407,
+      "loss": 2.1154,
+      "step": 197000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00024503383254297013,
+      "loss": 2.1118,
+      "step": 197500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0002448932541095762,
+      "loss": 2.1114,
+      "step": 198000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024475267567618226,
+      "loss": 2.1129,
+      "step": 198500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002446120972427883,
+      "loss": 2.1318,
+      "step": 199000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.00024447151880939433,
+      "loss": 2.081,
+      "step": 199500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0002443309403760004,
+      "loss": 2.087,
+      "step": 200000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0002441903619426065,
+      "loss": 2.1043,
+      "step": 200500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00024404978350921254,
+      "loss": 2.1188,
+      "step": 201000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002439092050758186,
+      "loss": 2.1295,
+      "step": 201500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002437686266424247,
+      "loss": 2.0991,
+      "step": 202000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.00024362804820903075,
+      "loss": 2.0925,
+      "step": 202500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.00024348746977563682,
+      "loss": 2.0488,
+      "step": 203000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00024334689134224285,
+      "loss": 2.0945,
+      "step": 203500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002432063129088489,
+      "loss": 2.1083,
+      "step": 204000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00024306573447545497,
+      "loss": 2.1211,
+      "step": 204500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00024292515604206104,
+      "loss": 2.1377,
+      "step": 205000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0002427845776086671,
+      "loss": 2.1321,
+      "step": 205500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00024264399917527316,
+      "loss": 2.1156,
+      "step": 206000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00024250342074187925,
+      "loss": 2.0931,
+      "step": 206500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0002423628423084853,
+      "loss": 2.1073,
+      "step": 207000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.00024222226387509137,
+      "loss": 2.1215,
+      "step": 207500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0002420816854416974,
+      "loss": 2.1354,
+      "step": 208000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00024194110700830347,
+      "loss": 2.0763,
+      "step": 208500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.00024180052857490953,
+      "loss": 2.1315,
+      "step": 209000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0002416599501415156,
+      "loss": 2.1101,
+      "step": 209500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.00024151937170812166,
+      "loss": 2.0826,
+      "step": 210000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00024137879327472774,
+      "loss": 2.1161,
+      "step": 210500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0002412382148413338,
+      "loss": 2.1167,
+      "step": 211000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00024109763640793987,
+      "loss": 2.1309,
+      "step": 211500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0002409570579745459,
+      "loss": 2.0976,
+      "step": 212000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00024081647954115197,
+      "loss": 2.1108,
+      "step": 212500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00024067590110775803,
+      "loss": 2.1444,
+      "step": 213000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0002405353226743641,
+      "loss": 2.0976,
+      "step": 213500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00024039474424097015,
+      "loss": 2.0973,
+      "step": 214000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0002402541658075762,
+      "loss": 2.0388,
+      "step": 214500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0002401135873741823,
+      "loss": 2.0271,
+      "step": 215000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00023997300894078836,
+      "loss": 2.0473,
+      "step": 215500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0002398324305073944,
+      "loss": 2.0443,
+      "step": 216000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00023969185207400046,
+      "loss": 2.0398,
+      "step": 216500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.00023955127364060652,
+      "loss": 2.049,
+      "step": 217000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00023941069520721258,
+      "loss": 2.0455,
+      "step": 217500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00023927011677381865,
+      "loss": 2.0773,
+      "step": 218000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0002391295383404247,
+      "loss": 2.025,
+      "step": 218500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0002389889599070308,
+      "loss": 2.0532,
+      "step": 219000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00023884838147363686,
+      "loss": 2.0117,
+      "step": 219500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0002387078030402429,
+      "loss": 2.0746,
+      "step": 220000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00023856722460684896,
+      "loss": 2.1025,
+      "step": 220500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00023842664617345502,
+      "loss": 2.0361,
+      "step": 221000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00023828606774006108,
+      "loss": 2.0572,
+      "step": 221500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00023814548930666714,
+      "loss": 2.0498,
+      "step": 222000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0002380049108732732,
+      "loss": 2.0354,
+      "step": 222500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.00023786433243987927,
+      "loss": 2.0455,
+      "step": 223000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00023772375400648536,
+      "loss": 2.0328,
+      "step": 223500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0002375831755730914,
+      "loss": 2.0704,
+      "step": 224000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00023744259713969745,
+      "loss": 2.065,
+      "step": 224500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00023730201870630351,
+      "loss": 2.05,
+      "step": 225000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00023716144027290958,
+      "loss": 2.0312,
+      "step": 225500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00023702086183951564,
+      "loss": 2.0403,
+      "step": 226000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0002368802834061217,
+      "loss": 2.0701,
+      "step": 226500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00023673970497272776,
+      "loss": 2.0486,
+      "step": 227000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00023659912653933385,
+      "loss": 2.049,
+      "step": 227500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.00023645854810593986,
+      "loss": 2.0493,
+      "step": 228000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00023631796967254595,
+      "loss": 2.0892,
+      "step": 228500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.000236177391239152,
+      "loss": 2.0607,
+      "step": 229000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00023603681280575807,
+      "loss": 2.0407,
+      "step": 229500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00023589623437236413,
+      "loss": 2.0438,
+      "step": 230000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0002357556559389702,
+      "loss": 2.0424,
+      "step": 230500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00023561507750557626,
+      "loss": 2.0456,
+      "step": 231000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00023547449907218232,
+      "loss": 2.0296,
+      "step": 231500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00023533392063878835,
+      "loss": 2.0823,
+      "step": 232000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00023519334220539444,
+      "loss": 2.0428,
+      "step": 232500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0002350527637720005,
+      "loss": 2.0489,
+      "step": 233000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00023491218533860657,
+      "loss": 2.0478,
+      "step": 233500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00023477160690521263,
+      "loss": 2.0622,
+      "step": 234000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.0002346310284718187,
+      "loss": 2.0566,
+      "step": 234500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00023449045003842475,
+      "loss": 2.0431,
+      "step": 235000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.00023434987160503081,
+      "loss": 2.0194,
+      "step": 235500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.00023420929317163685,
+      "loss": 1.9937,
+      "step": 236000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0002340687147382429,
+      "loss": 2.0064,
+      "step": 236500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.000233928136304849,
+      "loss": 2.0274,
+      "step": 237000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00023378755787145506,
+      "loss": 2.0512,
+      "step": 237500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00023364697943806112,
+      "loss": 2.0267,
+      "step": 238000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00023350640100466719,
+      "loss": 2.0159,
+      "step": 238500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00023336582257127325,
+      "loss": 2.0458,
+      "step": 239000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0002332252441378793,
+      "loss": 2.037,
+      "step": 239500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00023308466570448537,
+      "loss": 2.0356,
+      "step": 240000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0002329440872710914,
+      "loss": 2.0204,
+      "step": 240500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.0002328035088376975,
+      "loss": 2.0731,
+      "step": 241000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00023266293040430356,
+      "loss": 2.0473,
+      "step": 241500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00023252235197090962,
+      "loss": 2.0236,
+      "step": 242000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00023238177353751568,
+      "loss": 2.0754,
+      "step": 242500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00023224119510412174,
+      "loss": 2.0216,
+      "step": 243000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0002321006166707278,
+      "loss": 2.0726,
+      "step": 243500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00023196003823733387,
+      "loss": 2.037,
+      "step": 244000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0002318194598039399,
+      "loss": 2.0778,
+      "step": 244500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00023167888137054596,
+      "loss": 2.0366,
+      "step": 245000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00023153830293715205,
+      "loss": 2.0554,
+      "step": 245500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00023139772450375811,
+      "loss": 2.036,
+      "step": 246000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00023125714607036418,
+      "loss": 2.0575,
+      "step": 246500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00023111656763697024,
+      "loss": 2.0008,
+      "step": 247000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0002309759892035763,
+      "loss": 2.0589,
+      "step": 247500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00023083541077018236,
+      "loss": 2.0505,
+      "step": 248000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0002306948323367884,
+      "loss": 2.0034,
+      "step": 248500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00023055425390339446,
+      "loss": 2.0229,
+      "step": 249000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00023041367547000055,
+      "loss": 2.0329,
+      "step": 249500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0002302730970366066,
+      "loss": 1.9613,
+      "step": 250000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00023013251860321267,
+      "loss": 1.9882,
+      "step": 250500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00022999194016981873,
+      "loss": 1.9655,
+      "step": 251000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.0002298513617364248,
+      "loss": 1.9578,
+      "step": 251500
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00022971078330303086,
+      "loss": 2.033,
+      "step": 252000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0002295702048696369,
+      "loss": 1.9843,
+      "step": 252500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00022942962643624295,
+      "loss": 1.9682,
+      "step": 253000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00022928904800284902,
+      "loss": 1.9537,
+      "step": 253500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0002291484695694551,
+      "loss": 2.0058,
+      "step": 254000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.00022900789113606117,
+      "loss": 1.95,
+      "step": 254500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.00022886731270266723,
+      "loss": 1.9726,
+      "step": 255000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0002287267342692733,
+      "loss": 1.9936,
+      "step": 255500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00022858615583587935,
+      "loss": 1.9598,
+      "step": 256000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0002284455774024854,
+      "loss": 1.9492,
+      "step": 256500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00022830499896909145,
+      "loss": 1.9778,
+      "step": 257000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0002281644205356975,
+      "loss": 1.9714,
+      "step": 257500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.0002280238421023036,
+      "loss": 1.9862,
+      "step": 258000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00022788326366890966,
+      "loss": 1.997,
+      "step": 258500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00022774268523551573,
+      "loss": 1.9797,
+      "step": 259000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0002276021068021218,
+      "loss": 2.0386,
+      "step": 259500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.00022746152836872785,
+      "loss": 1.9809,
+      "step": 260000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00022732094993533388,
+      "loss": 1.9879,
+      "step": 260500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00022718037150193995,
+      "loss": 2.0213,
+      "step": 261000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.000227039793068546,
+      "loss": 1.9486,
+      "step": 261500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00022689921463515207,
+      "loss": 1.977,
+      "step": 262000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00022675863620175816,
+      "loss": 1.9849,
+      "step": 262500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00022661805776836422,
+      "loss": 1.9433,
+      "step": 263000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00022647747933497028,
+      "loss": 2.0007,
+      "step": 263500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00022633690090157634,
+      "loss": 1.9975,
+      "step": 264000
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00022619632246818238,
+      "loss": 1.9288,
+      "step": 264500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00022605574403478844,
+      "loss": 1.996,
+      "step": 265000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0002259151656013945,
+      "loss": 1.9691,
+      "step": 265500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00022577458716800056,
+      "loss": 1.9859,
+      "step": 266000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00022563400873460665,
+      "loss": 1.9902,
+      "step": 266500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00022549343030121272,
+      "loss": 1.9626,
+      "step": 267000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.00022535285186781878,
+      "loss": 1.9749,
+      "step": 267500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.00022521227343442484,
+      "loss": 1.9832,
+      "step": 268000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0002250716950010309,
+      "loss": 1.9943,
+      "step": 268500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.00022493111656763694,
+      "loss": 1.9564,
+      "step": 269000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.000224790538134243,
+      "loss": 1.9827,
+      "step": 269500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00022464995970084906,
+      "loss": 1.9721,
+      "step": 270000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.00022450938126745512,
+      "loss": 1.9993,
+      "step": 270500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0002243688028340612,
+      "loss": 1.9978,
+      "step": 271000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00022422822440066727,
+      "loss": 1.9862,
+      "step": 271500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00022408764596727334,
+      "loss": 1.9931,
+      "step": 272000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0002239470675338794,
+      "loss": 1.9609,
+      "step": 272500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00022380648910048543,
+      "loss": 1.9802,
+      "step": 273000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0002236659106670915,
+      "loss": 1.9919,
+      "step": 273500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00022352533223369756,
+      "loss": 1.9749,
+      "step": 274000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.00022338475380030362,
+      "loss": 2.0121,
+      "step": 274500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0002232441753669097,
+      "loss": 1.9941,
+      "step": 275000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00022310359693351577,
+      "loss": 1.9829,
+      "step": 275500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.00022296301850012183,
+      "loss": 2.0042,
+      "step": 276000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0002228224400667279,
+      "loss": 1.9994,
+      "step": 276500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00022268186163333393,
+      "loss": 2.0085,
+      "step": 277000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00022254128319994,
+      "loss": 1.9972,
+      "step": 277500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00022240070476654605,
+      "loss": 2.0229,
+      "step": 278000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0002222601263331521,
+      "loss": 1.9405,
+      "step": 278500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00022211954789975818,
+      "loss": 1.9859,
+      "step": 279000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00022197896946636426,
+      "loss": 1.9755,
+      "step": 279500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.00022183839103297033,
+      "loss": 1.9853,
+      "step": 280000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0002216978125995764,
+      "loss": 2.0005,
+      "step": 280500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.00022155723416618242,
+      "loss": 1.9459,
+      "step": 281000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.00022141665573278848,
+      "loss": 1.9943,
+      "step": 281500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.00022127607729939455,
+      "loss": 1.9501,
+      "step": 282000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0002211354988660006,
+      "loss": 1.9885,
+      "step": 282500
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.00022099492043260667,
+      "loss": 2.0035,
+      "step": 283000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00022085434199921276,
+      "loss": 2.002,
+      "step": 283500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00022071376356581882,
+      "loss": 1.992,
+      "step": 284000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00022057318513242488,
+      "loss": 2.0223,
+      "step": 284500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00022043260669903092,
+      "loss": 2.0528,
+      "step": 285000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00022029202826563698,
+      "loss": 1.9177,
+      "step": 285500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00022015144983224304,
+      "loss": 1.9261,
+      "step": 286000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0002200108713988491,
+      "loss": 1.9438,
+      "step": 286500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.00021987029296545517,
+      "loss": 1.9014,
+      "step": 287000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.00021972971453206123,
+      "loss": 1.9587,
+      "step": 287500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.00021958913609866732,
+      "loss": 1.9076,
+      "step": 288000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00021944855766527338,
+      "loss": 1.9177,
+      "step": 288500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.00021930797923187941,
+      "loss": 1.9116,
+      "step": 289000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00021916740079848548,
+      "loss": 1.9303,
+      "step": 289500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00021902682236509154,
+      "loss": 1.9255,
+      "step": 290000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0002188862439316976,
+      "loss": 1.9152,
+      "step": 290500
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00021874566549830366,
+      "loss": 1.9149,
+      "step": 291000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00021860508706490972,
+      "loss": 1.9298,
+      "step": 291500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0002184645086315158,
+      "loss": 1.9279,
+      "step": 292000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00021832393019812187,
+      "loss": 1.9413,
+      "step": 292500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0002181833517647279,
+      "loss": 1.9344,
+      "step": 293000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.00021804277333133397,
+      "loss": 1.9529,
+      "step": 293500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00021790219489794003,
+      "loss": 1.9493,
+      "step": 294000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0002177616164645461,
+      "loss": 1.9435,
+      "step": 294500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.00021762103803115216,
+      "loss": 1.9373,
+      "step": 295000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.00021748045959775822,
+      "loss": 1.9651,
+      "step": 295500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.00021733988116436428,
+      "loss": 1.9283,
+      "step": 296000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00021719930273097037,
+      "loss": 1.9202,
+      "step": 296500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00021705872429757643,
+      "loss": 1.9429,
+      "step": 297000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00021691814586418247,
+      "loss": 1.9262,
+      "step": 297500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00021677756743078853,
+      "loss": 1.9686,
+      "step": 298000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0002166369889973946,
+      "loss": 1.9541,
+      "step": 298500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00021649641056400065,
+      "loss": 1.9397,
+      "step": 299000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00021635583213060671,
+      "loss": 1.9521,
+      "step": 299500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.00021621525369721278,
+      "loss": 1.9354,
+      "step": 300000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.00021607467526381887,
+      "loss": 1.933,
+      "step": 300500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.00021593409683042493,
+      "loss": 1.9874,
+      "step": 301000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00021579351839703096,
+      "loss": 1.9077,
+      "step": 301500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.00021565293996363702,
+      "loss": 1.9458,
+      "step": 302000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.00021551236153024309,
+      "loss": 1.9521,
+      "step": 302500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00021537178309684915,
+      "loss": 1.9498,
+      "step": 303000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0002152312046634552,
+      "loss": 1.918,
+      "step": 303500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.00021509062623006127,
+      "loss": 1.9309,
+      "step": 304000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00021495004779666733,
+      "loss": 1.9431,
+      "step": 304500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.00021480946936327342,
+      "loss": 1.9449,
+      "step": 305000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.00021466889092987946,
+      "loss": 1.9312,
+      "step": 305500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00021452831249648552,
+      "loss": 1.9315,
+      "step": 306000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.00021438773406309158,
+      "loss": 1.9769,
+      "step": 306500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.00021424715562969764,
+      "loss": 1.9534,
+      "step": 307000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0002141065771963037,
+      "loss": 1.9379,
+      "step": 307500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.00021396599876290977,
+      "loss": 1.9297,
+      "step": 308000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.00021382542032951583,
+      "loss": 1.9221,
+      "step": 308500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.00021368484189612192,
+      "loss": 1.9232,
+      "step": 309000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.00021354426346272793,
+      "loss": 1.9292,
+      "step": 309500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.00021340368502933401,
+      "loss": 1.9457,
+      "step": 310000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.00021326310659594008,
+      "loss": 1.9498,
+      "step": 310500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.00021312252816254614,
+      "loss": 1.9103,
+      "step": 311000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0002129819497291522,
+      "loss": 1.9209,
+      "step": 311500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.00021284137129575826,
+      "loss": 1.9357,
+      "step": 312000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00021270079286236432,
+      "loss": 1.9045,
+      "step": 312500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0002125602144289704,
+      "loss": 1.9402,
+      "step": 313000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.00021241963599557642,
+      "loss": 1.8733,
+      "step": 313500
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0002122790575621825,
+      "loss": 1.9188,
+      "step": 314000
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.00021213847912878857,
+      "loss": 1.94,
+      "step": 314500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.00021199790069539463,
+      "loss": 1.969,
+      "step": 315000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0002118573222620007,
+      "loss": 1.9897,
+      "step": 315500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.00021171674382860676,
+      "loss": 1.9422,
+      "step": 316000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.00021157616539521282,
+      "loss": 1.9329,
+      "step": 316500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.00021143558696181888,
+      "loss": 1.9291,
+      "step": 317000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00021129500852842492,
+      "loss": 1.9177,
+      "step": 317500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00021115443009503098,
+      "loss": 1.9471,
+      "step": 318000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.00021101385166163707,
+      "loss": 1.8992,
+      "step": 318500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00021087327322824313,
+      "loss": 1.9003,
+      "step": 319000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.0002107326947948492,
+      "loss": 1.9663,
+      "step": 319500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.00021059211636145525,
+      "loss": 1.963,
+      "step": 320000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.00021045153792806132,
+      "loss": 1.9454,
+      "step": 320500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.00021031095949466738,
+      "loss": 1.8856,
+      "step": 321000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0002101703810612734,
+      "loss": 1.8773,
+      "step": 321500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.00021002980262787947,
+      "loss": 1.8712,
+      "step": 322000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.00020988922419448556,
+      "loss": 1.8868,
+      "step": 322500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.00020974864576109163,
+      "loss": 1.8753,
+      "step": 323000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0002096080673276977,
+      "loss": 1.9066,
+      "step": 323500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.00020946748889430375,
+      "loss": 1.8513,
+      "step": 324000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.0002093269104609098,
+      "loss": 1.8249,
+      "step": 324500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.00020918633202751587,
+      "loss": 1.8887,
+      "step": 325000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.00020904575359412193,
+      "loss": 1.8882,
+      "step": 325500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.00020890517516072797,
+      "loss": 1.8697,
+      "step": 326000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.00020876459672733403,
+      "loss": 1.8819,
+      "step": 326500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.00020862401829394012,
+      "loss": 1.8538,
+      "step": 327000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.00020848343986054618,
+      "loss": 1.882,
+      "step": 327500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.00020834286142715224,
+      "loss": 1.8733,
+      "step": 328000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0002082022829937583,
+      "loss": 1.8481,
+      "step": 328500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.00020806170456036437,
+      "loss": 1.886,
+      "step": 329000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.00020792112612697043,
+      "loss": 1.9102,
+      "step": 329500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.00020778054769357647,
+      "loss": 1.8617,
+      "step": 330000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.00020763996926018253,
+      "loss": 1.8809,
+      "step": 330500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.00020749939082678862,
+      "loss": 1.8844,
+      "step": 331000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.00020735881239339468,
+      "loss": 1.8758,
+      "step": 331500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.00020721823396000074,
+      "loss": 1.9132,
+      "step": 332000
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0002070776555266068,
+      "loss": 1.8819,
+      "step": 332500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.00020693707709321286,
+      "loss": 1.8851,
+      "step": 333000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.00020679649865981893,
+      "loss": 1.8757,
+      "step": 333500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00020665592022642496,
+      "loss": 1.8982,
+      "step": 334000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.00020651534179303102,
+      "loss": 1.8726,
+      "step": 334500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.00020637476335963708,
+      "loss": 1.871,
+      "step": 335000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.00020623418492624317,
+      "loss": 1.8679,
+      "step": 335500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.00020609360649284924,
+      "loss": 1.846,
+      "step": 336000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.0002059530280594553,
+      "loss": 1.8835,
+      "step": 336500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.00020581244962606136,
+      "loss": 1.8742,
+      "step": 337000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.00020567187119266742,
+      "loss": 1.9074,
+      "step": 337500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.00020553129275927346,
+      "loss": 1.8882,
+      "step": 338000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00020539071432587952,
+      "loss": 1.8665,
+      "step": 338500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.00020525013589248558,
+      "loss": 1.8989,
+      "step": 339000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.00020510955745909167,
+      "loss": 1.892,
+      "step": 339500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.00020496897902569773,
+      "loss": 1.8602,
+      "step": 340000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0002048284005923038,
+      "loss": 1.8711,
+      "step": 340500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.00020468782215890985,
+      "loss": 1.8755,
+      "step": 341000
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.00020454724372551592,
+      "loss": 1.9111,
+      "step": 341500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.00020440666529212195,
+      "loss": 1.8911,
+      "step": 342000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.000204266086858728,
+      "loss": 1.8861,
+      "step": 342500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.00020412550842533408,
+      "loss": 1.9273,
+      "step": 343000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.00020398492999194014,
+      "loss": 1.8982,
+      "step": 343500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.00020384435155854623,
+      "loss": 1.9151,
+      "step": 344000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.0002037037731251523,
+      "loss": 1.9015,
+      "step": 344500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.00020356319469175835,
+      "loss": 1.888,
+      "step": 345000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.0002034226162583644,
+      "loss": 1.8713,
+      "step": 345500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00020328203782497045,
+      "loss": 1.8942,
+      "step": 346000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0002031414593915765,
+      "loss": 1.8574,
+      "step": 346500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.00020300088095818257,
+      "loss": 1.9104,
+      "step": 347000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.00020286030252478863,
+      "loss": 1.8821,
+      "step": 347500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.00020271972409139472,
+      "loss": 1.8594,
+      "step": 348000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.00020257914565800078,
+      "loss": 1.854,
+      "step": 348500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.00020243856722460685,
+      "loss": 1.901,
+      "step": 349000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0002022979887912129,
+      "loss": 1.9013,
+      "step": 349500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.00020215741035781894,
+      "loss": 1.8955,
+      "step": 350000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.000202016831924425,
+      "loss": 1.8922,
+      "step": 350500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.00020187625349103107,
+      "loss": 1.8906,
+      "step": 351000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.00020173567505763713,
+      "loss": 1.8719,
+      "step": 351500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.0002015950966242432,
+      "loss": 1.9044,
+      "step": 352000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.00020145451819084928,
+      "loss": 1.8842,
+      "step": 352500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.00020131393975745534,
+      "loss": 1.8419,
+      "step": 353000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0002011733613240614,
+      "loss": 1.9004,
+      "step": 353500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.00020103278289066746,
+      "loss": 1.8882,
+      "step": 354000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.0002008922044572735,
+      "loss": 1.9269,
+      "step": 354500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.00020075162602387956,
+      "loss": 1.8875,
+      "step": 355000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.00020061104759048562,
+      "loss": 1.8902,
+      "step": 355500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00020047046915709169,
+      "loss": 1.8823,
+      "step": 356000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.00020032989072369777,
+      "loss": 1.8574,
+      "step": 356500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00020018931229030384,
+      "loss": 1.7962,
+      "step": 357000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.0002000487338569099,
+      "loss": 1.8085,
+      "step": 357500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.00019990815542351596,
+      "loss": 1.8401,
+      "step": 358000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.000199767576990122,
+      "loss": 1.8218,
+      "step": 358500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.00019962699855672806,
+      "loss": 1.8065,
+      "step": 359000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.00019948642012333412,
+      "loss": 1.8245,
+      "step": 359500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.00019934584168994018,
+      "loss": 1.8177,
+      "step": 360000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.00019920526325654624,
+      "loss": 1.8626,
+      "step": 360500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.00019906468482315233,
+      "loss": 1.8341,
+      "step": 361000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0001989241063897584,
+      "loss": 1.8424,
+      "step": 361500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.00019878352795636446,
+      "loss": 1.8221,
+      "step": 362000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0001986429495229705,
+      "loss": 1.8554,
+      "step": 362500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.00019850237108957655,
+      "loss": 1.8426,
+      "step": 363000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.00019836179265618261,
+      "loss": 1.8355,
+      "step": 363500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.00019822121422278868,
+      "loss": 1.8147,
+      "step": 364000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.00019808063578939474,
+      "loss": 1.8212,
+      "step": 364500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.00019794005735600083,
+      "loss": 1.8188,
+      "step": 365000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.0001977994789226069,
+      "loss": 1.8313,
+      "step": 365500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.00019765890048921295,
+      "loss": 1.8165,
+      "step": 366000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.00019751832205581899,
+      "loss": 1.8183,
+      "step": 366500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.00019737774362242505,
+      "loss": 1.8196,
+      "step": 367000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0001972371651890311,
+      "loss": 1.8276,
+      "step": 367500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.00019709658675563717,
+      "loss": 1.8811,
+      "step": 368000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00019695600832224323,
+      "loss": 1.8258,
+      "step": 368500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.0001968154298888493,
+      "loss": 1.8061,
+      "step": 369000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.00019667485145545538,
+      "loss": 1.8099,
+      "step": 369500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.00019653427302206145,
+      "loss": 1.8491,
+      "step": 370000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.00019639369458866748,
+      "loss": 1.8424,
+      "step": 370500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.00019625311615527354,
+      "loss": 1.8419,
+      "step": 371000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.0001961125377218796,
+      "loss": 1.828,
+      "step": 371500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.00019597195928848567,
+      "loss": 1.8339,
+      "step": 372000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.00019583138085509173,
+      "loss": 1.8416,
+      "step": 372500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.0001956908024216978,
+      "loss": 1.8231,
+      "step": 373000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.00019555022398830388,
+      "loss": 1.8288,
+      "step": 373500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.00019540964555490994,
+      "loss": 1.8484,
+      "step": 374000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.00019526906712151598,
+      "loss": 1.8453,
+      "step": 374500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.00019512848868812204,
+      "loss": 1.8539,
+      "step": 375000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.0001949879102547281,
+      "loss": 1.869,
+      "step": 375500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00019484733182133416,
+      "loss": 1.8507,
+      "step": 376000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.00019470675338794022,
+      "loss": 1.8167,
+      "step": 376500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.0001945661749545463,
+      "loss": 1.823,
+      "step": 377000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.00019442559652115235,
+      "loss": 1.8547,
+      "step": 377500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.00019428501808775844,
+      "loss": 1.8602,
+      "step": 378000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.00019414443965436445,
+      "loss": 1.8406,
+      "step": 378500
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.00019400386122097053,
+      "loss": 1.8683,
+      "step": 379000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0001938632827875766,
+      "loss": 1.8577,
+      "step": 379500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.00019372270435418266,
+      "loss": 1.8518,
+      "step": 380000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.00019358212592078872,
+      "loss": 1.8134,
+      "step": 380500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.00019344154748739478,
+      "loss": 1.8355,
+      "step": 381000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.00019330096905400084,
+      "loss": 1.8328,
+      "step": 381500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.00019316039062060693,
+      "loss": 1.8329,
+      "step": 382000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.00019301981218721294,
+      "loss": 1.8755,
+      "step": 382500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.00019287923375381903,
+      "loss": 1.84,
+      "step": 383000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0001927386553204251,
+      "loss": 1.8026,
+      "step": 383500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.00019259807688703115,
+      "loss": 1.8435,
+      "step": 384000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.00019245749845363722,
+      "loss": 1.8323,
+      "step": 384500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.00019231692002024328,
+      "loss": 1.8357,
+      "step": 385000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.00019217634158684934,
+      "loss": 1.8223,
+      "step": 385500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0001920357631534554,
+      "loss": 1.8558,
+      "step": 386000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0001918951847200615,
+      "loss": 1.8361,
+      "step": 386500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.0001917546062866675,
+      "loss": 1.8391,
+      "step": 387000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.0001916140278532736,
+      "loss": 1.9066,
+      "step": 387500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.00019147344941987965,
+      "loss": 1.8619,
+      "step": 388000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.0001913328709864857,
+      "loss": 1.8718,
+      "step": 388500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.00019119229255309177,
+      "loss": 1.8475,
+      "step": 389000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.00019105171411969783,
+      "loss": 1.8525,
+      "step": 389500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.0001909111356863039,
+      "loss": 1.8086,
+      "step": 390000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.00019077055725290999,
+      "loss": 1.8516,
+      "step": 390500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.000190629978819516,
+      "loss": 1.8374,
+      "step": 391000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.00019048940038612208,
+      "loss": 1.8353,
+      "step": 391500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00019034882195272814,
+      "loss": 1.8845,
+      "step": 392000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.0001902082435193342,
+      "loss": 1.8034,
+      "step": 392500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00019006766508594027,
+      "loss": 1.7912,
+      "step": 393000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.00018992708665254633,
+      "loss": 1.7747,
+      "step": 393500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0001897865082191524,
+      "loss": 1.7816,
+      "step": 394000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.00018964592978575845,
+      "loss": 1.7557,
+      "step": 394500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.0001895053513523645,
+      "loss": 1.7835,
+      "step": 395000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.00018936477291897055,
+      "loss": 1.7824,
+      "step": 395500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.00018922419448557664,
+      "loss": 1.7959,
+      "step": 396000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.0001890836160521827,
+      "loss": 1.7989,
+      "step": 396500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.00018894303761878876,
+      "loss": 1.761,
+      "step": 397000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.00018880245918539483,
+      "loss": 1.7773,
+      "step": 397500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.0001886618807520009,
+      "loss": 1.7956,
+      "step": 398000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00018852130231860695,
+      "loss": 1.7832,
+      "step": 398500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.00018838072388521298,
+      "loss": 1.82,
+      "step": 399000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.00018824014545181905,
+      "loss": 1.7774,
+      "step": 399500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.00018809956701842514,
+      "loss": 1.8108,
+      "step": 400000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.0001879589885850312,
+      "loss": 1.8072,
+      "step": 400500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00018781841015163726,
+      "loss": 1.7976,
+      "step": 401000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00018767783171824332,
+      "loss": 1.7671,
+      "step": 401500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.00018753725328484938,
+      "loss": 1.8202,
+      "step": 402000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.00018739667485145545,
+      "loss": 1.7835,
+      "step": 402500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.00018725609641806148,
+      "loss": 1.79,
+      "step": 403000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.00018711551798466754,
+      "loss": 1.8187,
+      "step": 403500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.0001869749395512736,
+      "loss": 1.8261,
+      "step": 404000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.0001868343611178797,
+      "loss": 1.7671,
+      "step": 404500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00018669378268448575,
+      "loss": 1.797,
+      "step": 405000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.00018655320425109182,
+      "loss": 1.7898,
+      "step": 405500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00018641262581769788,
+      "loss": 1.7916,
+      "step": 406000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.00018627204738430394,
+      "loss": 1.852,
+      "step": 406500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.00018613146895090998,
+      "loss": 1.7775,
+      "step": 407000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.00018599089051751604,
+      "loss": 1.8036,
+      "step": 407500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.0001858503120841221,
+      "loss": 1.8111,
+      "step": 408000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0001857097336507282,
+      "loss": 1.7921,
+      "step": 408500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.00018556915521733425,
+      "loss": 1.8084,
+      "step": 409000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0001854285767839403,
+      "loss": 1.7578,
+      "step": 409500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.00018528799835054637,
+      "loss": 1.7714,
+      "step": 410000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.00018514741991715244,
+      "loss": 1.8062,
+      "step": 410500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.00018500684148375847,
+      "loss": 1.8261,
+      "step": 411000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.00018486626305036453,
+      "loss": 1.7864,
+      "step": 411500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.0001847256846169706,
+      "loss": 1.8151,
+      "step": 412000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.00018458510618357666,
+      "loss": 1.8115,
+      "step": 412500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.00018444452775018275,
+      "loss": 1.8313,
+      "step": 413000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.0001843039493167888,
+      "loss": 1.8024,
+      "step": 413500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.00018416337088339487,
+      "loss": 1.793,
+      "step": 414000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.00018402279245000093,
+      "loss": 1.7891,
+      "step": 414500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.000183882214016607,
+      "loss": 1.7916,
+      "step": 415000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.00018374163558321303,
+      "loss": 1.7597,
+      "step": 415500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0001836010571498191,
+      "loss": 1.809,
+      "step": 416000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.00018346047871642515,
+      "loss": 1.7811,
+      "step": 416500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00018331990028303124,
+      "loss": 1.8246,
+      "step": 417000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0001831793218496373,
+      "loss": 1.8383,
+      "step": 417500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00018303874341624336,
+      "loss": 1.7673,
+      "step": 418000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.00018289816498284943,
+      "loss": 1.7972,
+      "step": 418500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0001827575865494555,
+      "loss": 1.8094,
+      "step": 419000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.00018261700811606152,
+      "loss": 1.7746,
+      "step": 419500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00018247642968266759,
+      "loss": 1.8126,
+      "step": 420000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.00018233585124927365,
+      "loss": 1.8074,
+      "step": 420500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0001821952728158797,
+      "loss": 1.8232,
+      "step": 421000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0001820546943824858,
+      "loss": 1.8191,
+      "step": 421500
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.00018191411594909186,
+      "loss": 1.7828,
+      "step": 422000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00018177353751569792,
+      "loss": 1.7953,
+      "step": 422500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.00018163295908230398,
+      "loss": 1.8094,
+      "step": 423000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00018149238064891002,
+      "loss": 1.8087,
+      "step": 423500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.00018135180221551608,
+      "loss": 1.819,
+      "step": 424000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.00018121122378212214,
+      "loss": 1.7811,
+      "step": 424500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.0001810706453487282,
+      "loss": 1.7685,
+      "step": 425000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0001809300669153343,
+      "loss": 1.7985,
+      "step": 425500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00018078948848194036,
+      "loss": 1.7791,
+      "step": 426000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.00018064891004854642,
+      "loss": 1.8049,
+      "step": 426500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.00018050833161515248,
+      "loss": 1.8249,
+      "step": 427000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.00018036775318175851,
+      "loss": 1.7993,
+      "step": 427500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00018022717474836458,
+      "loss": 1.7924,
+      "step": 428000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.00018008659631497064,
+      "loss": 1.7594,
+      "step": 428500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.0001799460178815767,
+      "loss": 1.7235,
+      "step": 429000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.00017980543944818276,
+      "loss": 1.7484,
+      "step": 429500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.00017966486101478885,
+      "loss": 1.7384,
+      "step": 430000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.0001795242825813949,
+      "loss": 1.7548,
+      "step": 430500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00017938370414800098,
+      "loss": 1.7523,
+      "step": 431000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.000179243125714607,
+      "loss": 1.7387,
+      "step": 431500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00017910254728121307,
+      "loss": 1.7238,
+      "step": 432000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.00017896196884781913,
+      "loss": 1.7568,
+      "step": 432500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0001788213904144252,
+      "loss": 1.728,
+      "step": 433000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.00017868081198103126,
+      "loss": 1.7263,
+      "step": 433500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00017854023354763735,
+      "loss": 1.7219,
+      "step": 434000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0001783996551142434,
+      "loss": 1.7415,
+      "step": 434500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.00017825907668084947,
+      "loss": 1.7571,
+      "step": 435000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0001781184982474555,
+      "loss": 1.7429,
+      "step": 435500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.00017797791981406157,
+      "loss": 1.763,
+      "step": 436000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.00017783734138066763,
+      "loss": 1.7558,
+      "step": 436500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0001776967629472737,
+      "loss": 1.7665,
+      "step": 437000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.00017755618451387975,
+      "loss": 1.7411,
+      "step": 437500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.00017741560608048582,
+      "loss": 1.7478,
+      "step": 438000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0001772750276470919,
+      "loss": 1.7673,
+      "step": 438500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.00017713444921369797,
+      "loss": 1.752,
+      "step": 439000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.000176993870780304,
+      "loss": 1.7297,
+      "step": 439500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.00017685329234691006,
+      "loss": 1.7654,
+      "step": 440000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.00017671271391351612,
+      "loss": 1.7312,
+      "step": 440500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0001765721354801222,
+      "loss": 1.773,
+      "step": 441000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.00017643155704672825,
+      "loss": 1.7771,
+      "step": 441500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0001762909786133343,
+      "loss": 1.7174,
+      "step": 442000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0001761504001799404,
+      "loss": 1.7056,
+      "step": 442500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.00017600982174654646,
+      "loss": 1.7896,
+      "step": 443000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00017586924331315252,
+      "loss": 1.7391,
+      "step": 443500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.00017572866487975856,
+      "loss": 1.7606,
+      "step": 444000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.00017558808644636462,
+      "loss": 1.7515,
+      "step": 444500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.00017544750801297068,
+      "loss": 1.7781,
+      "step": 445000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.00017530692957957674,
+      "loss": 1.7423,
+      "step": 445500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0001751663511461828,
+      "loss": 1.7643,
+      "step": 446000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.00017502577271278887,
+      "loss": 1.7443,
+      "step": 446500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00017488519427939496,
+      "loss": 1.7836,
+      "step": 447000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.00017474461584600102,
+      "loss": 1.7667,
+      "step": 447500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.00017460403741260705,
+      "loss": 1.7431,
+      "step": 448000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.00017446345897921312,
+      "loss": 1.7748,
+      "step": 448500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.00017432288054581918,
+      "loss": 1.7667,
+      "step": 449000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.00017418230211242524,
+      "loss": 1.7635,
+      "step": 449500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0001740417236790313,
+      "loss": 1.77,
+      "step": 450000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.00017390114524563736,
+      "loss": 1.7338,
+      "step": 450500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.00017376056681224345,
+      "loss": 1.761,
+      "step": 451000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.00017361998837884951,
+      "loss": 1.7557,
+      "step": 451500
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.00017347940994545555,
+      "loss": 1.7942,
+      "step": 452000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0001733388315120616,
+      "loss": 1.7452,
+      "step": 452500
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.00017319825307866767,
+      "loss": 1.7466,
+      "step": 453000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.00017305767464527373,
+      "loss": 1.7551,
+      "step": 453500
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0001729170962118798,
+      "loss": 1.7765,
+      "step": 454000
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.00017277651777848586,
+      "loss": 1.7686,
+      "step": 454500
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.00017263593934509192,
+      "loss": 1.7711,
+      "step": 455000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.000172495360911698,
+      "loss": 1.7608,
+      "step": 455500
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.00017235478247830404,
+      "loss": 1.7719,
+      "step": 456000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0001722142040449101,
+      "loss": 1.7815,
+      "step": 456500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.00017207362561151617,
+      "loss": 1.7713,
+      "step": 457000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.00017193304717812223,
+      "loss": 1.7486,
+      "step": 457500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0001717924687447283,
+      "loss": 1.7744,
+      "step": 458000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.00017165189031133435,
+      "loss": 1.7748,
+      "step": 458500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.00017151131187794042,
+      "loss": 1.7558,
+      "step": 459000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0001713707334445465,
+      "loss": 1.765,
+      "step": 459500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0001712301550111525,
+      "loss": 1.7724,
+      "step": 460000
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0001710895765777586,
+      "loss": 1.7484,
+      "step": 460500
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.00017094899814436466,
+      "loss": 1.7719,
+      "step": 461000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.00017080841971097073,
+      "loss": 1.7918,
+      "step": 461500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.0001706678412775768,
+      "loss": 1.7792,
+      "step": 462000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.00017052726284418285,
+      "loss": 1.7368,
+      "step": 462500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0001703866844107889,
+      "loss": 1.7598,
+      "step": 463000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.00017024610597739497,
+      "loss": 1.7528,
+      "step": 463500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.000170105527544001,
+      "loss": 1.6807,
+      "step": 464000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0001699649491106071,
+      "loss": 1.7284,
+      "step": 464500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.00016982437067721316,
+      "loss": 1.6896,
+      "step": 465000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00016968379224381922,
+      "loss": 1.7024,
+      "step": 465500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.00016954321381042528,
+      "loss": 1.7025,
+      "step": 466000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.00016940263537703135,
+      "loss": 1.6848,
+      "step": 466500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.0001692620569436374,
+      "loss": 1.707,
+      "step": 467000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.00016912147851024347,
+      "loss": 1.7138,
+      "step": 467500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.0001689809000768495,
+      "loss": 1.6969,
+      "step": 468000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.00016884032164345557,
+      "loss": 1.6782,
+      "step": 468500
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.00016869974321006165,
+      "loss": 1.6959,
+      "step": 469000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.00016855916477666772,
+      "loss": 1.7567,
+      "step": 469500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.00016841858634327378,
+      "loss": 1.7163,
+      "step": 470000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.00016827800790987984,
+      "loss": 1.7081,
+      "step": 470500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.0001681374294764859,
+      "loss": 1.703,
+      "step": 471000
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.00016799685104309196,
+      "loss": 1.7047,
+      "step": 471500
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.00016785627260969803,
+      "loss": 1.7161,
+      "step": 472000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.00016771569417630406,
+      "loss": 1.7038,
+      "step": 472500
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.00016757511574291015,
+      "loss": 1.7404,
+      "step": 473000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.0001674345373095162,
+      "loss": 1.7217,
+      "step": 473500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00016729395887612227,
+      "loss": 1.702,
+      "step": 474000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.00016715338044272834,
+      "loss": 1.7199,
+      "step": 474500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.0001670128020093344,
+      "loss": 1.7062,
+      "step": 475000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.00016687222357594046,
+      "loss": 1.6744,
+      "step": 475500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.00016673164514254652,
+      "loss": 1.673,
+      "step": 476000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.00016659106670915256,
+      "loss": 1.7313,
+      "step": 476500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.00016645048827575862,
+      "loss": 1.7032,
+      "step": 477000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0001663099098423647,
+      "loss": 1.7284,
+      "step": 477500
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.00016616933140897077,
+      "loss": 1.7071,
+      "step": 478000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.00016602875297557683,
+      "loss": 1.7076,
+      "step": 478500
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.0001658881745421829,
+      "loss": 1.7219,
+      "step": 479000
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00016574759610878896,
+      "loss": 1.7143,
+      "step": 479500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.00016560701767539502,
+      "loss": 1.7173,
+      "step": 480000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.00016546643924200105,
+      "loss": 1.7236,
+      "step": 480500
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.00016532586080860711,
+      "loss": 1.7243,
+      "step": 481000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.0001651852823752132,
+      "loss": 1.7027,
+      "step": 481500
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.00016504470394181927,
+      "loss": 1.7147,
+      "step": 482000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.00016490412550842533,
+      "loss": 1.705,
+      "step": 482500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.0001647635470750314,
+      "loss": 1.6933,
+      "step": 483000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00016462296864163745,
+      "loss": 1.7353,
+      "step": 483500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0001644823902082435,
+      "loss": 1.6976,
+      "step": 484000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.00016434181177484955,
+      "loss": 1.7118,
+      "step": 484500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0001642012333414556,
+      "loss": 1.7086,
+      "step": 485000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.00016406065490806167,
+      "loss": 1.7019,
+      "step": 485500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00016392007647466776,
+      "loss": 1.7094,
+      "step": 486000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.00016377949804127382,
+      "loss": 1.7013,
+      "step": 486500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.00016363891960787988,
+      "loss": 1.7514,
+      "step": 487000
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.00016349834117448595,
+      "loss": 1.6936,
+      "step": 487500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.000163357762741092,
+      "loss": 1.7091,
+      "step": 488000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.00016321718430769804,
+      "loss": 1.7251,
+      "step": 488500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0001630766058743041,
+      "loss": 1.6919,
+      "step": 489000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.00016293602744091017,
+      "loss": 1.7196,
+      "step": 489500
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.00016279544900751626,
+      "loss": 1.7089,
+      "step": 490000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.00016265487057412232,
+      "loss": 1.727,
+      "step": 490500
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.00016251429214072838,
+      "loss": 1.7141,
+      "step": 491000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00016237371370733444,
+      "loss": 1.7047,
+      "step": 491500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.0001622331352739405,
+      "loss": 1.6898,
+      "step": 492000
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.00016209255684054654,
+      "loss": 1.6999,
+      "step": 492500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0001619519784071526,
+      "loss": 1.7309,
+      "step": 493000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.00016181139997375866,
+      "loss": 1.6896,
+      "step": 493500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.00016167082154036472,
+      "loss": 1.7286,
+      "step": 494000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.0001615302431069708,
+      "loss": 1.7278,
+      "step": 494500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.00016138966467357688,
+      "loss": 1.725,
+      "step": 495000
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00016124908624018294,
+      "loss": 1.7102,
+      "step": 495500
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.000161108507806789,
+      "loss": 1.6975,
+      "step": 496000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.00016096792937339503,
+      "loss": 1.72,
+      "step": 496500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0001608273509400011,
+      "loss": 1.7225,
+      "step": 497000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.00016068677250660716,
+      "loss": 1.6941,
+      "step": 497500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.00016054619407321322,
+      "loss": 1.7246,
+      "step": 498000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0001604056156398193,
+      "loss": 1.7029,
+      "step": 498500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.00016026503720642537,
+      "loss": 1.7228,
+      "step": 499000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.00016012445877303143,
+      "loss": 1.657,
+      "step": 499500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0001599838803396375,
+      "loss": 1.6552,
+      "step": 500000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.00015984330190624356,
+      "loss": 1.6641,
+      "step": 500500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0001597027234728496,
+      "loss": 1.6517,
+      "step": 501000
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.00015956214503945565,
+      "loss": 1.6249,
+      "step": 501500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00015942156660606172,
+      "loss": 1.6939,
+      "step": 502000
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.00015928098817266778,
+      "loss": 1.6659,
+      "step": 502500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.00015914040973927387,
+      "loss": 1.6344,
+      "step": 503000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.00015899983130587993,
+      "loss": 1.6471,
+      "step": 503500
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.000158859252872486,
+      "loss": 1.6678,
+      "step": 504000
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.00015871867443909205,
+      "loss": 1.645,
+      "step": 504500
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0001585780960056981,
+      "loss": 1.6122,
+      "step": 505000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.00015843751757230415,
+      "loss": 1.6648,
+      "step": 505500
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.0001582969391389102,
+      "loss": 1.634,
+      "step": 506000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.00015815636070551627,
+      "loss": 1.6539,
+      "step": 506500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.00015801578227212236,
+      "loss": 1.6509,
+      "step": 507000
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.00015787520383872842,
+      "loss": 1.6445,
+      "step": 507500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.00015773462540533449,
+      "loss": 1.6699,
+      "step": 508000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00015759404697194055,
+      "loss": 1.6545,
+      "step": 508500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.00015745346853854658,
+      "loss": 1.6798,
+      "step": 509000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00015731289010515264,
+      "loss": 1.6795,
+      "step": 509500
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.0001571723116717587,
+      "loss": 1.6626,
+      "step": 510000
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.00015703173323836477,
+      "loss": 1.675,
+      "step": 510500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.00015689115480497083,
+      "loss": 1.6708,
+      "step": 511000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.00015675057637157692,
+      "loss": 1.6486,
+      "step": 511500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.00015660999793818298,
+      "loss": 1.6444,
+      "step": 512000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.00015646941950478904,
+      "loss": 1.6837,
+      "step": 512500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.00015632884107139508,
+      "loss": 1.6924,
+      "step": 513000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.00015618826263800114,
+      "loss": 1.6624,
+      "step": 513500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0001560476842046072,
+      "loss": 1.6541,
+      "step": 514000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.00015590710577121326,
+      "loss": 1.6492,
+      "step": 514500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.00015576652733781933,
+      "loss": 1.6661,
+      "step": 515000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.00015562594890442541,
+      "loss": 1.6671,
+      "step": 515500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.00015548537047103148,
+      "loss": 1.6869,
+      "step": 516000
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.00015534479203763754,
+      "loss": 1.6793,
+      "step": 516500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.00015520421360424357,
+      "loss": 1.635,
+      "step": 517000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.00015506363517084964,
+      "loss": 1.6433,
+      "step": 517500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.0001549230567374557,
+      "loss": 1.6893,
+      "step": 518000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.00015478247830406176,
+      "loss": 1.6515,
+      "step": 518500
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.00015464189987066782,
+      "loss": 1.6721,
+      "step": 519000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.00015450132143727388,
+      "loss": 1.6938,
+      "step": 519500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.00015436074300387997,
+      "loss": 1.6508,
+      "step": 520000
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.00015422016457048603,
+      "loss": 1.6625,
+      "step": 520500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00015407958613709207,
+      "loss": 1.6956,
+      "step": 521000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.00015393900770369813,
+      "loss": 1.6799,
+      "step": 521500
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0001537984292703042,
+      "loss": 1.6834,
+      "step": 522000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.00015365785083691025,
+      "loss": 1.6736,
+      "step": 522500
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.00015351727240351632,
+      "loss": 1.6651,
+      "step": 523000
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.00015337669397012238,
+      "loss": 1.6767,
+      "step": 523500
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.00015323611553672847,
+      "loss": 1.6678,
+      "step": 524000
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.00015309553710333453,
+      "loss": 1.6838,
+      "step": 524500
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.00015295495866994056,
+      "loss": 1.6628,
+      "step": 525000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.00015281438023654663,
+      "loss": 1.6682,
+      "step": 525500
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.0001526738018031527,
+      "loss": 1.6808,
+      "step": 526000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.00015253322336975875,
+      "loss": 1.6699,
+      "step": 526500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.0001523926449363648,
+      "loss": 1.6419,
+      "step": 527000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00015225206650297087,
+      "loss": 1.6573,
+      "step": 527500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.00015211148806957694,
+      "loss": 1.6837,
+      "step": 528000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.00015197090963618302,
+      "loss": 1.6669,
+      "step": 528500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.00015183033120278906,
+      "loss": 1.6609,
+      "step": 529000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.00015168975276939512,
+      "loss": 1.7295,
+      "step": 529500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.00015154917433600118,
+      "loss": 1.6872,
+      "step": 530000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.00015140859590260725,
+      "loss": 1.6727,
+      "step": 530500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0001512680174692133,
+      "loss": 1.6716,
+      "step": 531000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.00015112743903581937,
+      "loss": 1.6782,
+      "step": 531500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.00015098686060242543,
+      "loss": 1.6883,
+      "step": 532000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.00015084628216903152,
+      "loss": 1.6422,
+      "step": 532500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.00015070570373563758,
+      "loss": 1.6991,
+      "step": 533000
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.00015056512530224362,
+      "loss": 1.6951,
+      "step": 533500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.00015042454686884968,
+      "loss": 1.6697,
+      "step": 534000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.00015028396843545574,
+      "loss": 1.7002,
+      "step": 534500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.0001501433900020618,
+      "loss": 1.6082,
+      "step": 535000
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.00015000281156866786,
+      "loss": 1.6386,
+      "step": 535500
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.00014986223313527393,
+      "loss": 1.6434,
+      "step": 536000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.00014972165470188,
+      "loss": 1.6171,
+      "step": 536500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.00014958107626848605,
+      "loss": 1.5879,
+      "step": 537000
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.0001494404978350921,
+      "loss": 1.6285,
+      "step": 537500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00014929991940169817,
+      "loss": 1.6323,
+      "step": 538000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.00014915934096830424,
+      "loss": 1.6375,
+      "step": 538500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0001490187625349103,
+      "loss": 1.6198,
+      "step": 539000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.00014887818410151636,
+      "loss": 1.6497,
+      "step": 539500
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.00014873760566812242,
+      "loss": 1.6025,
+      "step": 540000
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.00014859702723472848,
+      "loss": 1.642,
+      "step": 540500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.00014845644880133455,
+      "loss": 1.6113,
+      "step": 541000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.0001483158703679406,
+      "loss": 1.6269,
+      "step": 541500
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.00014817529193454667,
+      "loss": 1.6071,
+      "step": 542000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.00014803471350115273,
+      "loss": 1.6044,
+      "step": 542500
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0001478941350677588,
+      "loss": 1.6428,
+      "step": 543000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.00014775355663436486,
+      "loss": 1.5916,
+      "step": 543500
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.00014761297820097092,
+      "loss": 1.6241,
+      "step": 544000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00014747239976757698,
+      "loss": 1.6256,
+      "step": 544500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.00014733182133418304,
+      "loss": 1.6019,
+      "step": 545000
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.0001471912429007891,
+      "loss": 1.6239,
+      "step": 545500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.00014705066446739517,
+      "loss": 1.6393,
+      "step": 546000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.00014691008603400123,
+      "loss": 1.6319,
+      "step": 546500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0001467695076006073,
+      "loss": 1.6056,
+      "step": 547000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.00014662892916721335,
+      "loss": 1.6102,
+      "step": 547500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0001464883507338194,
+      "loss": 1.6085,
+      "step": 548000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00014634777230042547,
+      "loss": 1.6607,
+      "step": 548500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.00014620719386703154,
+      "loss": 1.6512,
+      "step": 549000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0001460666154336376,
+      "loss": 1.6552,
+      "step": 549500
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.00014592603700024366,
+      "loss": 1.618,
+      "step": 550000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.00014578545856684972,
+      "loss": 1.6183,
+      "step": 550500
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.00014564488013345578,
+      "loss": 1.6328,
+      "step": 551000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00014550430170006185,
+      "loss": 1.6187,
+      "step": 551500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.0001453637232666679,
+      "loss": 1.6064,
+      "step": 552000
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.00014522314483327397,
+      "loss": 1.6181,
+      "step": 552500
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.00014508256639988003,
+      "loss": 1.6118,
+      "step": 553000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0001449419879664861,
+      "loss": 1.6387,
+      "step": 553500
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.00014480140953309216,
+      "loss": 1.6194,
+      "step": 554000
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.00014466083109969822,
+      "loss": 1.6419,
+      "step": 554500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.00014452025266630428,
+      "loss": 1.6443,
+      "step": 555000
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00014437967423291034,
+      "loss": 1.6379,
+      "step": 555500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0001442390957995164,
+      "loss": 1.617,
+      "step": 556000
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.00014409851736612244,
+      "loss": 1.6325,
+      "step": 556500
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.00014395793893272853,
+      "loss": 1.6373,
+      "step": 557000
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0001438173604993346,
+      "loss": 1.6221,
+      "step": 557500
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.00014367678206594065,
+      "loss": 1.612,
+      "step": 558000
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.0001435362036325467,
+      "loss": 1.6379,
+      "step": 558500
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.00014339562519915278,
+      "loss": 1.6326,
+      "step": 559000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.00014325504676575884,
+      "loss": 1.6009,
+      "step": 559500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0001431144683323649,
+      "loss": 1.6199,
+      "step": 560000
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.00014297388989897096,
+      "loss": 1.6541,
+      "step": 560500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.00014283331146557702,
+      "loss": 1.6647,
+      "step": 561000
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.00014269273303218308,
+      "loss": 1.6486,
+      "step": 561500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.00014255215459878915,
+      "loss": 1.6654,
+      "step": 562000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0001424115761653952,
+      "loss": 1.6393,
+      "step": 562500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.00014227099773200127,
+      "loss": 1.6223,
+      "step": 563000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.00014213041929860733,
+      "loss": 1.677,
+      "step": 563500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0001419898408652134,
+      "loss": 1.5877,
+      "step": 564000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.00014184926243181946,
+      "loss": 1.6096,
+      "step": 564500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0001417086839984255,
+      "loss": 1.6199,
+      "step": 565000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.00014156810556503158,
+      "loss": 1.609,
+      "step": 565500
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.00014142752713163764,
+      "loss": 1.6239,
+      "step": 566000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0001412869486982437,
+      "loss": 1.6299,
+      "step": 566500
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.00014114637026484974,
+      "loss": 1.6673,
+      "step": 567000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.00014100579183145583,
+      "loss": 1.6265,
+      "step": 567500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.0001408652133980619,
+      "loss": 1.6383,
+      "step": 568000
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.00014072463496466795,
+      "loss": 1.6294,
+      "step": 568500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.000140584056531274,
+      "loss": 1.6214,
+      "step": 569000
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.00014044347809788008,
+      "loss": 1.6521,
+      "step": 569500
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.00014030289966448614,
+      "loss": 1.6504,
+      "step": 570000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0001401623212310922,
+      "loss": 1.5651,
+      "step": 570500
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.00014002174279769823,
+      "loss": 1.5787,
+      "step": 571000
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.00013988116436430432,
+      "loss": 1.5698,
+      "step": 571500
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.00013974058593091039,
+      "loss": 1.5477,
+      "step": 572000
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.00013960000749751645,
+      "loss": 1.5727,
+      "step": 572500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.00013945942906412248,
+      "loss": 1.5762,
+      "step": 573000
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.00013931885063072854,
+      "loss": 1.5924,
+      "step": 573500
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.00013917827219733463,
+      "loss": 1.5913,
+      "step": 574000
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0001390376937639407,
+      "loss": 1.581,
+      "step": 574500
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.00013889711533054673,
+      "loss": 1.5518,
+      "step": 575000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0001387565368971528,
+      "loss": 1.5782,
+      "step": 575500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.00013861595846375888,
+      "loss": 1.5733,
+      "step": 576000
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.00013847538003036494,
+      "loss": 1.5869,
+      "step": 576500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.00013833480159697098,
+      "loss": 1.5828,
+      "step": 577000
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.00013819422316357704,
+      "loss": 1.5645,
+      "step": 577500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.00013805364473018313,
+      "loss": 1.5653,
+      "step": 578000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0001379130662967892,
+      "loss": 1.601,
+      "step": 578500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00013777248786339523,
+      "loss": 1.5955,
+      "step": 579000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0001376319094300013,
+      "loss": 1.5793,
+      "step": 579500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.00013749133099660735,
+      "loss": 1.5702,
+      "step": 580000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.00013735075256321344,
+      "loss": 1.5841,
+      "step": 580500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.00013721017412981947,
+      "loss": 1.5379,
+      "step": 581000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.00013706959569642554,
+      "loss": 1.5676,
+      "step": 581500
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0001369290172630316,
+      "loss": 1.5905,
+      "step": 582000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.00013678843882963769,
+      "loss": 1.6139,
+      "step": 582500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.00013664786039624372,
+      "loss": 1.608,
+      "step": 583000
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.00013650728196284978,
+      "loss": 1.5832,
+      "step": 583500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.00013636670352945584,
+      "loss": 1.6174,
+      "step": 584000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.00013622612509606193,
+      "loss": 1.5895,
+      "step": 584500
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.00013608554666266797,
+      "loss": 1.5861,
+      "step": 585000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.00013594496822927403,
+      "loss": 1.5796,
+      "step": 585500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0001358043897958801,
+      "loss": 1.5854,
+      "step": 586000
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.00013566381136248618,
+      "loss": 1.5782,
+      "step": 586500
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.00013552323292909222,
+      "loss": 1.5978,
+      "step": 587000
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.00013538265449569828,
+      "loss": 1.6099,
+      "step": 587500
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.00013524207606230434,
+      "loss": 1.6093,
+      "step": 588000
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.0001351014976289104,
+      "loss": 1.6228,
+      "step": 588500
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.0001349609191955165,
+      "loss": 1.5772,
+      "step": 589000
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.00013482034076212253,
+      "loss": 1.5884,
+      "step": 589500
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0001346797623287286,
+      "loss": 1.5894,
+      "step": 590000
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.00013453918389533465,
+      "loss": 1.5949,
+      "step": 590500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.00013439860546194074,
+      "loss": 1.5949,
+      "step": 591000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.00013425802702854677,
+      "loss": 1.6063,
+      "step": 591500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.00013411744859515284,
+      "loss": 1.5743,
+      "step": 592000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.0001339768701617589,
+      "loss": 1.586,
+      "step": 592500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.000133836291728365,
+      "loss": 1.5523,
+      "step": 593000
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.00013369571329497102,
+      "loss": 1.5871,
+      "step": 593500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.00013355513486157708,
+      "loss": 1.5904,
+      "step": 594000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.00013341455642818315,
+      "loss": 1.5936,
+      "step": 594500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.00013327397799478923,
+      "loss": 1.6254,
+      "step": 595000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.00013313339956139527,
+      "loss": 1.5805,
+      "step": 595500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.00013299282112800133,
+      "loss": 1.6077,
+      "step": 596000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0001328522426946074,
+      "loss": 1.5784,
+      "step": 596500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.00013271166426121345,
+      "loss": 1.585,
+      "step": 597000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.00013257108582781952,
+      "loss": 1.6174,
+      "step": 597500
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.00013243050739442558,
+      "loss": 1.5657,
+      "step": 598000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.00013228992896103164,
+      "loss": 1.6194,
+      "step": 598500
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.0001321493505276377,
+      "loss": 1.5821,
+      "step": 599000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.00013200877209424376,
+      "loss": 1.5954,
+      "step": 599500
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.00013186819366084983,
+      "loss": 1.6197,
+      "step": 600000
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.0001317276152274559,
+      "loss": 1.5936,
+      "step": 600500
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.00013158703679406195,
+      "loss": 1.6081,
+      "step": 601000
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.000131446458360668,
+      "loss": 1.6012,
+      "step": 601500
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.00013130587992727407,
+      "loss": 1.6169,
+      "step": 602000
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.00013116530149388014,
+      "loss": 1.6046,
+      "step": 602500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0001310247230604862,
+      "loss": 1.6007,
+      "step": 603000
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.00013088414462709226,
+      "loss": 1.5971,
+      "step": 603500
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00013074356619369832,
+      "loss": 1.5821,
+      "step": 604000
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.00013060298776030438,
+      "loss": 1.585,
+      "step": 604500
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.00013046240932691045,
+      "loss": 1.617,
+      "step": 605000
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0001303218308935165,
+      "loss": 1.6341,
+      "step": 605500
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.00013018125246012257,
+      "loss": 1.5642,
+      "step": 606000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.00013004067402672863,
+      "loss": 1.5107,
+      "step": 606500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0001299000955933347,
+      "loss": 1.5292,
+      "step": 607000
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.00012975951715994076,
+      "loss": 1.5316,
+      "step": 607500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.00012961893872654682,
+      "loss": 1.5349,
+      "step": 608000
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.00012947836029315288,
+      "loss": 1.5068,
+      "step": 608500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.00012933778185975894,
+      "loss": 1.5212,
+      "step": 609000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.000129197203426365,
+      "loss": 1.5319,
+      "step": 609500
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.00012905662499297107,
+      "loss": 1.5505,
+      "step": 610000
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00012891604655957713,
+      "loss": 1.4921,
+      "step": 610500
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0001287754681261832,
+      "loss": 1.5233,
+      "step": 611000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.00012863488969278925,
+      "loss": 1.5579,
+      "step": 611500
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0001284943112593953,
+      "loss": 1.5588,
+      "step": 612000
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00012835373282600137,
+      "loss": 1.5452,
+      "step": 612500
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.00012821315439260744,
+      "loss": 1.5415,
+      "step": 613000
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0001280725759592135,
+      "loss": 1.5208,
+      "step": 613500
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.00012793199752581956,
+      "loss": 1.5582,
+      "step": 614000
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.00012779141909242562,
+      "loss": 1.5683,
+      "step": 614500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.00012765084065903168,
+      "loss": 1.534,
+      "step": 615000
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.00012751026222563775,
+      "loss": 1.5108,
+      "step": 615500
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.0001273696837922438,
+      "loss": 1.5363,
+      "step": 616000
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.00012722910535884987,
+      "loss": 1.5752,
+      "step": 616500
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.00012708852692545593,
+      "loss": 1.5703,
+      "step": 617000
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.000126947948492062,
+      "loss": 1.5406,
+      "step": 617500
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00012680737005866806,
+      "loss": 1.5417,
+      "step": 618000
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.00012666679162527412,
+      "loss": 1.5452,
+      "step": 618500
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.00012652621319188018,
+      "loss": 1.5571,
+      "step": 619000
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.00012638563475848624,
+      "loss": 1.5484,
+      "step": 619500
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0001262450563250923,
+      "loss": 1.5979,
+      "step": 620000
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.00012610447789169837,
+      "loss": 1.5486,
+      "step": 620500
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.00012596389945830443,
+      "loss": 1.4974,
+      "step": 621000
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0001258233210249105,
+      "loss": 1.5237,
+      "step": 621500
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.00012568274259151655,
+      "loss": 1.5502,
+      "step": 622000
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.0001255421641581226,
+      "loss": 1.5373,
+      "step": 622500
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.00012540158572472868,
+      "loss": 1.5406,
+      "step": 623000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.00012526100729133474,
+      "loss": 1.5559,
+      "step": 623500
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.0001251204288579408,
+      "loss": 1.5506,
+      "step": 624000
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.00012497985042454686,
+      "loss": 1.5421,
+      "step": 624500
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.00012483927199115292,
+      "loss": 1.5631,
+      "step": 625000
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00012469869355775899,
+      "loss": 1.5374,
+      "step": 625500
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.00012455811512436505,
+      "loss": 1.5681,
+      "step": 626000
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.0001244175366909711,
+      "loss": 1.541,
+      "step": 626500
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.00012427695825757717,
+      "loss": 1.5325,
+      "step": 627000
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.00012413637982418323,
+      "loss": 1.5226,
+      "step": 627500
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0001239958013907893,
+      "loss": 1.5524,
+      "step": 628000
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.00012385522295739536,
+      "loss": 1.6071,
+      "step": 628500
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.00012371464452400142,
+      "loss": 1.5639,
+      "step": 629000
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.00012357406609060748,
+      "loss": 1.5428,
+      "step": 629500
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.00012343348765721354,
+      "loss": 1.5646,
+      "step": 630000
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0001232929092238196,
+      "loss": 1.5568,
+      "step": 630500
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00012315233079042567,
+      "loss": 1.5319,
+      "step": 631000
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.00012301175235703173,
+      "loss": 1.5503,
+      "step": 631500
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.0001228711739236378,
+      "loss": 1.5291,
+      "step": 632000
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.00012273059549024385,
+      "loss": 1.5911,
+      "step": 632500
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.00012259001705684991,
+      "loss": 1.5696,
+      "step": 633000
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.00012244943862345598,
+      "loss": 1.5498,
+      "step": 633500
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.00012230886019006204,
+      "loss": 1.5292,
+      "step": 634000
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0001221682817566681,
+      "loss": 1.5606,
+      "step": 634500
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.00012202770332327416,
+      "loss": 1.5626,
+      "step": 635000
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.00012188712488988022,
+      "loss": 1.5849,
+      "step": 635500
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.00012174654645648627,
+      "loss": 1.5534,
+      "step": 636000
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.00012160596802309233,
+      "loss": 1.5346,
+      "step": 636500
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00012146538958969841,
+      "loss": 1.5747,
+      "step": 637000
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.00012132481115630447,
+      "loss": 1.5572,
+      "step": 637500
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.00012118423272291052,
+      "loss": 1.5445,
+      "step": 638000
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.00012104365428951658,
+      "loss": 1.5555,
+      "step": 638500
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.00012090307585612264,
+      "loss": 1.5335,
+      "step": 639000
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00012076249742272872,
+      "loss": 1.5836,
+      "step": 639500
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.00012062191898933477,
+      "loss": 1.5538,
+      "step": 640000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.00012048134055594083,
+      "loss": 1.5627,
+      "step": 640500
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.00012034076212254689,
+      "loss": 1.5717,
+      "step": 641000
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.00012020018368915297,
+      "loss": 1.5152,
+      "step": 641500
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.00012005960525575902,
+      "loss": 1.4885,
+      "step": 642000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00011991902682236508,
+      "loss": 1.4731,
+      "step": 642500
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00011977844838897114,
+      "loss": 1.5073,
+      "step": 643000
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00011963786995557721,
+      "loss": 1.4917,
+      "step": 643500
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.00011949729152218326,
+      "loss": 1.4906,
+      "step": 644000
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.00011935671308878932,
+      "loss": 1.5095,
+      "step": 644500
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00011921613465539539,
+      "loss": 1.4915,
+      "step": 645000
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.00011907555622200146,
+      "loss": 1.5024,
+      "step": 645500
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00011893497778860751,
+      "loss": 1.5251,
+      "step": 646000
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.00011879439935521357,
+      "loss": 1.4836,
+      "step": 646500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.00011865382092181963,
+      "loss": 1.5216,
+      "step": 647000
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.0001185132424884257,
+      "loss": 1.4953,
+      "step": 647500
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.00011837266405503177,
+      "loss": 1.528,
+      "step": 648000
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.00011823208562163782,
+      "loss": 1.5212,
+      "step": 648500
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00011809150718824388,
+      "loss": 1.5065,
+      "step": 649000
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.00011795092875484994,
+      "loss": 1.5281,
+      "step": 649500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00011781035032145602,
+      "loss": 1.5104,
+      "step": 650000
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.00011766977188806207,
+      "loss": 1.4646,
+      "step": 650500
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00011752919345466813,
+      "loss": 1.4751,
+      "step": 651000
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.00011738861502127419,
+      "loss": 1.5132,
+      "step": 651500
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.00011724803658788027,
+      "loss": 1.4984,
+      "step": 652000
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00011710745815448632,
+      "loss": 1.4835,
+      "step": 652500
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.00011696687972109238,
+      "loss": 1.5207,
+      "step": 653000
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.00011682630128769844,
+      "loss": 1.4979,
+      "step": 653500
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.00011668572285430452,
+      "loss": 1.5274,
+      "step": 654000
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.00011654514442091056,
+      "loss": 1.5101,
+      "step": 654500
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.00011640456598751663,
+      "loss": 1.5121,
+      "step": 655000
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00011626398755412269,
+      "loss": 1.5177,
+      "step": 655500
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.00011612340912072875,
+      "loss": 1.4977,
+      "step": 656000
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00011598283068733481,
+      "loss": 1.5221,
+      "step": 656500
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.00011584225225394087,
+      "loss": 1.5066,
+      "step": 657000
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.00011570167382054694,
+      "loss": 1.4997,
+      "step": 657500
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.000115561095387153,
+      "loss": 1.5211,
+      "step": 658000
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.00011542051695375905,
+      "loss": 1.5144,
+      "step": 658500
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.00011527993852036512,
+      "loss": 1.4994,
+      "step": 659000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.00011513936008697118,
+      "loss": 1.5215,
+      "step": 659500
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.00011499878165357724,
+      "loss": 1.5035,
+      "step": 660000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00011485820322018329,
+      "loss": 1.5063,
+      "step": 660500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.00011471762478678937,
+      "loss": 1.4783,
+      "step": 661000
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.00011457704635339543,
+      "loss": 1.4942,
+      "step": 661500
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.00011443646792000149,
+      "loss": 1.528,
+      "step": 662000
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.00011429588948660754,
+      "loss": 1.5122,
+      "step": 662500
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.00011415531105321362,
+      "loss": 1.5361,
+      "step": 663000
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.00011401473261981968,
+      "loss": 1.5162,
+      "step": 663500
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.00011387415418642574,
+      "loss": 1.4806,
+      "step": 664000
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.00011373357575303179,
+      "loss": 1.496,
+      "step": 664500
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.00011359299731963786,
+      "loss": 1.5381,
+      "step": 665000
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.00011345241888624393,
+      "loss": 1.5123,
+      "step": 665500
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.00011331184045284999,
+      "loss": 1.5367,
+      "step": 666000
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.00011317126201945604,
+      "loss": 1.5061,
+      "step": 666500
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.0001130306835860621,
+      "loss": 1.4939,
+      "step": 667000
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.00011289010515266817,
+      "loss": 1.5094,
+      "step": 667500
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.00011274952671927424,
+      "loss": 1.5351,
+      "step": 668000
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.00011260894828588028,
+      "loss": 1.5246,
+      "step": 668500
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.00011246836985248635,
+      "loss": 1.5233,
+      "step": 669000
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.00011232779141909242,
+      "loss": 1.5041,
+      "step": 669500
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.00011218721298569848,
+      "loss": 1.5076,
+      "step": 670000
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.00011204663455230453,
+      "loss": 1.5281,
+      "step": 670500
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.0001119060561189106,
+      "loss": 1.5359,
+      "step": 671000
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.00011176547768551667,
+      "loss": 1.5241,
+      "step": 671500
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.00011162489925212273,
+      "loss": 1.5,
+      "step": 672000
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.00011148432081872878,
+      "loss": 1.4874,
+      "step": 672500
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.00011134374238533484,
+      "loss": 1.5024,
+      "step": 673000
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.00011120316395194092,
+      "loss": 1.5587,
+      "step": 673500
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00011106258551854698,
+      "loss": 1.5569,
+      "step": 674000
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.00011092200708515303,
+      "loss": 1.5419,
+      "step": 674500
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.00011078142865175909,
+      "loss": 1.4949,
+      "step": 675000
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00011064085021836515,
+      "loss": 1.5332,
+      "step": 675500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.00011050027178497123,
+      "loss": 1.5137,
+      "step": 676000
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.00011035969335157729,
+      "loss": 1.537,
+      "step": 676500
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.00011021911491818334,
+      "loss": 1.5142,
+      "step": 677000
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.0001100785364847894,
+      "loss": 1.4425,
+      "step": 677500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00010993795805139547,
+      "loss": 1.4679,
+      "step": 678000
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.00010979737961800154,
+      "loss": 1.441,
+      "step": 678500
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.00010965680118460758,
+      "loss": 1.4367,
+      "step": 679000
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.00010951622275121365,
+      "loss": 1.4158,
+      "step": 679500
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.00010937564431781972,
+      "loss": 1.467,
+      "step": 680000
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.00010923506588442578,
+      "loss": 1.4936,
+      "step": 680500
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.00010909448745103183,
+      "loss": 1.4517,
+      "step": 681000
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0001089539090176379,
+      "loss": 1.4532,
+      "step": 681500
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.00010881333058424396,
+      "loss": 1.4788,
+      "step": 682000
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.00010867275215085003,
+      "loss": 1.4477,
+      "step": 682500
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.00010853217371745608,
+      "loss": 1.4405,
+      "step": 683000
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.00010839159528406214,
+      "loss": 1.4977,
+      "step": 683500
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.0001082510168506682,
+      "loss": 1.4801,
+      "step": 684000
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00010811043841727428,
+      "loss": 1.4558,
+      "step": 684500
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00010796985998388033,
+      "loss": 1.4634,
+      "step": 685000
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.00010782928155048639,
+      "loss": 1.451,
+      "step": 685500
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.00010768870311709245,
+      "loss": 1.4647,
+      "step": 686000
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.00010754812468369853,
+      "loss": 1.458,
+      "step": 686500
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.00010740754625030458,
+      "loss": 1.4842,
+      "step": 687000
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 0.00010726696781691064,
+      "loss": 1.4794,
+      "step": 687500
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0001071263893835167,
+      "loss": 1.4627,
+      "step": 688000
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 0.00010698581095012277,
+      "loss": 1.4567,
+      "step": 688500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.00010684523251672882,
+      "loss": 1.4916,
+      "step": 689000
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.00010670465408333489,
+      "loss": 1.4865,
+      "step": 689500
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.00010656407564994095,
+      "loss": 1.4926,
+      "step": 690000
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 0.00010642349721654701,
+      "loss": 1.4923,
+      "step": 690500
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00010628291878315307,
+      "loss": 1.4813,
+      "step": 691000
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 0.00010614234034975913,
+      "loss": 1.4633,
+      "step": 691500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.0001060017619163652,
+      "loss": 1.4653,
+      "step": 692000
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00010586118348297126,
+      "loss": 1.4518,
+      "step": 692500
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.0001057206050495773,
+      "loss": 1.4856,
+      "step": 693000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.00010558002661618338,
+      "loss": 1.4646,
+      "step": 693500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.00010543944818278944,
+      "loss": 1.4714,
+      "step": 694000
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.0001052988697493955,
+      "loss": 1.4602,
+      "step": 694500
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 0.00010515829131600155,
+      "loss": 1.4555,
+      "step": 695000
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.00010501771288260763,
+      "loss": 1.4647,
+      "step": 695500
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.00010487713444921369,
+      "loss": 1.4686,
+      "step": 696000
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.00010473655601581975,
+      "loss": 1.4484,
+      "step": 696500
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 0.0001045959775824258,
+      "loss": 1.488,
+      "step": 697000
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.00010445539914903188,
+      "loss": 1.4916,
+      "step": 697500
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.00010431482071563794,
+      "loss": 1.4915,
+      "step": 698000
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.000104174242282244,
+      "loss": 1.4803,
+      "step": 698500
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.00010403366384885005,
+      "loss": 1.5062,
+      "step": 699000
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.00010389308541545612,
+      "loss": 1.4809,
+      "step": 699500
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.00010375250698206219,
+      "loss": 1.4766,
+      "step": 700000
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.00010361192854866825,
+      "loss": 1.4997,
+      "step": 700500
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.0001034713501152743,
+      "loss": 1.4345,
+      "step": 701000
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.00010333077168188036,
+      "loss": 1.4837,
+      "step": 701500
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00010319019324848643,
+      "loss": 1.5083,
+      "step": 702000
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.0001030496148150925,
+      "loss": 1.4603,
+      "step": 702500
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.00010290903638169854,
+      "loss": 1.52,
+      "step": 703000
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.0001027684579483046,
+      "loss": 1.4754,
+      "step": 703500
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.00010262787951491068,
+      "loss": 1.4833,
+      "step": 704000
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.00010248730108151674,
+      "loss": 1.484,
+      "step": 704500
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.0001023467226481228,
+      "loss": 1.4821,
+      "step": 705000
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.00010220614421472885,
+      "loss": 1.504,
+      "step": 705500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.00010206556578133493,
+      "loss": 1.4986,
+      "step": 706000
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.00010192498734794099,
+      "loss": 1.5064,
+      "step": 706500
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.00010178440891454705,
+      "loss": 1.4506,
+      "step": 707000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0001016438304811531,
+      "loss": 1.4618,
+      "step": 707500
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.00010150325204775918,
+      "loss": 1.4656,
+      "step": 708000
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00010136267361436524,
+      "loss": 1.477,
+      "step": 708500
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.0001012220951809713,
+      "loss": 1.4651,
+      "step": 709000
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00010108151674757735,
+      "loss": 1.484,
+      "step": 709500
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.00010094093831418341,
+      "loss": 1.4566,
+      "step": 710000
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.00010080035988078949,
+      "loss": 1.48,
+      "step": 710500
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.00010065978144739555,
+      "loss": 1.484,
+      "step": 711000
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.0001005192030140016,
+      "loss": 1.4938,
+      "step": 711500
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.00010037862458060766,
+      "loss": 1.4896,
+      "step": 712000
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.00010023804614721373,
+      "loss": 1.4827,
+      "step": 712500
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.0001000974677138198,
+      "loss": 1.4364,
+      "step": 713000
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 9.995688928042584e-05,
+      "loss": 1.4643,
+      "step": 713500
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 9.98163108470319e-05,
+      "loss": 1.4105,
+      "step": 714000
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.967573241363798e-05,
+      "loss": 1.4383,
+      "step": 714500
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.953515398024404e-05,
+      "loss": 1.4317,
+      "step": 715000
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.939457554685009e-05,
+      "loss": 1.4217,
+      "step": 715500
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.925399711345615e-05,
+      "loss": 1.4363,
+      "step": 716000
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.911341868006223e-05,
+      "loss": 1.4337,
+      "step": 716500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.897284024666829e-05,
+      "loss": 1.4153,
+      "step": 717000
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 9.883226181327434e-05,
+      "loss": 1.4394,
+      "step": 717500
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.86916833798804e-05,
+      "loss": 1.4216,
+      "step": 718000
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 9.855110494648646e-05,
+      "loss": 1.4515,
+      "step": 718500
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.841052651309254e-05,
+      "loss": 1.4232,
+      "step": 719000
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 9.826994807969859e-05,
+      "loss": 1.4117,
+      "step": 719500
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.812936964630465e-05,
+      "loss": 1.4224,
+      "step": 720000
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 9.798879121291071e-05,
+      "loss": 1.426,
+      "step": 720500
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.784821277951679e-05,
+      "loss": 1.4179,
+      "step": 721000
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 9.770763434612284e-05,
+      "loss": 1.4248,
+      "step": 721500
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.75670559127289e-05,
+      "loss": 1.4219,
+      "step": 722000
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 9.742647747933496e-05,
+      "loss": 1.4071,
+      "step": 722500
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.728589904594103e-05,
+      "loss": 1.4151,
+      "step": 723000
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.714532061254708e-05,
+      "loss": 1.4235,
+      "step": 723500
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.700474217915314e-05,
+      "loss": 1.4523,
+      "step": 724000
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.68641637457592e-05,
+      "loss": 1.4305,
+      "step": 724500
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.672358531236528e-05,
+      "loss": 1.4417,
+      "step": 725000
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.658300687897133e-05,
+      "loss": 1.4372,
+      "step": 725500
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 9.644242844557739e-05,
+      "loss": 1.4423,
+      "step": 726000
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.630185001218345e-05,
+      "loss": 1.428,
+      "step": 726500
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 9.616127157878952e-05,
+      "loss": 1.4481,
+      "step": 727000
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.602069314539558e-05,
+      "loss": 1.4533,
+      "step": 727500
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 9.588011471200164e-05,
+      "loss": 1.4555,
+      "step": 728000
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.57395362786077e-05,
+      "loss": 1.4279,
+      "step": 728500
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 9.559895784521376e-05,
+      "loss": 1.4715,
+      "step": 729000
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.545837941181981e-05,
+      "loss": 1.4165,
+      "step": 729500
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 9.531780097842589e-05,
+      "loss": 1.4176,
+      "step": 730000
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.517722254503195e-05,
+      "loss": 1.4379,
+      "step": 730500
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.503664411163801e-05,
+      "loss": 1.4531,
+      "step": 731000
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.489606567824406e-05,
+      "loss": 1.4247,
+      "step": 731500
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 9.475548724485014e-05,
+      "loss": 1.4398,
+      "step": 732000
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 9.46149088114562e-05,
+      "loss": 1.4422,
+      "step": 732500
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 9.447433037806226e-05,
+      "loss": 1.4273,
+      "step": 733000
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 9.433375194466831e-05,
+      "loss": 1.4482,
+      "step": 733500
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 9.419317351127438e-05,
+      "loss": 1.445,
+      "step": 734000
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 9.405259507788045e-05,
+      "loss": 1.4507,
+      "step": 734500
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 9.391201664448651e-05,
+      "loss": 1.4147,
+      "step": 735000
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 9.377143821109257e-05,
+      "loss": 1.4329,
+      "step": 735500
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 9.363085977769863e-05,
+      "loss": 1.4602,
+      "step": 736000
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 9.349028134430469e-05,
+      "loss": 1.4706,
+      "step": 736500
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 9.334970291091076e-05,
+      "loss": 1.4183,
+      "step": 737000
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 9.320912447751682e-05,
+      "loss": 1.4342,
+      "step": 737500
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 9.306854604412287e-05,
+      "loss": 1.4291,
+      "step": 738000
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 9.292796761072894e-05,
+      "loss": 1.4312,
+      "step": 738500
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 9.2787389177335e-05,
+      "loss": 1.4287,
+      "step": 739000
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 9.264681074394106e-05,
+      "loss": 1.4421,
+      "step": 739500
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 9.250623231054711e-05,
+      "loss": 1.4204,
+      "step": 740000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 9.236565387715319e-05,
+      "loss": 1.4279,
+      "step": 740500
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 9.222507544375925e-05,
+      "loss": 1.4129,
+      "step": 741000
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 9.208449701036531e-05,
+      "loss": 1.4563,
+      "step": 741500
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 9.194391857697136e-05,
+      "loss": 1.437,
+      "step": 742000
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 9.180334014357744e-05,
+      "loss": 1.4574,
+      "step": 742500
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 9.16627617101835e-05,
+      "loss": 1.4312,
+      "step": 743000
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 9.152218327678956e-05,
+      "loss": 1.4421,
+      "step": 743500
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 9.138160484339561e-05,
+      "loss": 1.4347,
+      "step": 744000
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 9.124102641000168e-05,
+      "loss": 1.4635,
+      "step": 744500
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 9.110044797660775e-05,
+      "loss": 1.4754,
+      "step": 745000
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 9.095986954321381e-05,
+      "loss": 1.4191,
+      "step": 745500
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 9.081929110981986e-05,
+      "loss": 1.4343,
+      "step": 746000
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 9.067871267642592e-05,
+      "loss": 1.4699,
+      "step": 746500
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 9.0538134243032e-05,
+      "loss": 1.4495,
+      "step": 747000
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 9.039755580963806e-05,
+      "loss": 1.4427,
+      "step": 747500
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 9.02569773762441e-05,
+      "loss": 1.4523,
+      "step": 748000
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 9.011639894285017e-05,
+      "loss": 1.4197,
+      "step": 748500
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.997582050945624e-05,
+      "loss": 1.4132,
+      "step": 749000
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.98352420760623e-05,
+      "loss": 1.3953,
+      "step": 749500
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 8.969466364266835e-05,
+      "loss": 1.4112,
+      "step": 750000
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 8.955408520927441e-05,
+      "loss": 1.3864,
+      "step": 750500
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 8.941350677588049e-05,
+      "loss": 1.3912,
+      "step": 751000
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 8.927292834248655e-05,
+      "loss": 1.3867,
+      "step": 751500
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 8.91323499090926e-05,
+      "loss": 1.4091,
+      "step": 752000
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 8.899177147569866e-05,
+      "loss": 1.4045,
+      "step": 752500
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 8.885119304230474e-05,
+      "loss": 1.4034,
+      "step": 753000
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 8.87106146089108e-05,
+      "loss": 1.3951,
+      "step": 753500
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 8.857003617551685e-05,
+      "loss": 1.397,
+      "step": 754000
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 8.842945774212291e-05,
+      "loss": 1.383,
+      "step": 754500
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 8.828887930872897e-05,
+      "loss": 1.4052,
+      "step": 755000
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 8.814830087533505e-05,
+      "loss": 1.405,
+      "step": 755500
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 8.80077224419411e-05,
+      "loss": 1.4064,
+      "step": 756000
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 8.786714400854716e-05,
+      "loss": 1.413,
+      "step": 756500
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 8.772656557515322e-05,
+      "loss": 1.3881,
+      "step": 757000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 8.75859871417593e-05,
+      "loss": 1.377,
+      "step": 757500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 8.744540870836534e-05,
+      "loss": 1.3939,
+      "step": 758000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 8.73048302749714e-05,
+      "loss": 1.4041,
+      "step": 758500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 8.716425184157747e-05,
+      "loss": 1.3923,
+      "step": 759000
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 8.702367340818354e-05,
+      "loss": 1.3887,
+      "step": 759500
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 8.688309497478959e-05,
+      "loss": 1.4045,
+      "step": 760000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 8.674251654139565e-05,
+      "loss": 1.3886,
+      "step": 760500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 8.660193810800171e-05,
+      "loss": 1.4225,
+      "step": 761000
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 8.646135967460779e-05,
+      "loss": 1.3489,
+      "step": 761500
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 8.632078124121384e-05,
+      "loss": 1.4059,
+      "step": 762000
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 8.61802028078199e-05,
+      "loss": 1.3909,
+      "step": 762500
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 8.603962437442596e-05,
+      "loss": 1.3911,
+      "step": 763000
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 8.589904594103202e-05,
+      "loss": 1.3751,
+      "step": 763500
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.57584675076381e-05,
+      "loss": 1.3928,
+      "step": 764000
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 8.561788907424415e-05,
+      "loss": 1.4066,
+      "step": 764500
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.547731064085021e-05,
+      "loss": 1.4075,
+      "step": 765000
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 8.533673220745627e-05,
+      "loss": 1.4234,
+      "step": 765500
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.519615377406235e-05,
+      "loss": 1.4039,
+      "step": 766000
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 8.50555753406684e-05,
+      "loss": 1.3861,
+      "step": 766500
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 8.491499690727446e-05,
+      "loss": 1.3816,
+      "step": 767000
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.477441847388052e-05,
+      "loss": 1.4111,
+      "step": 767500
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 8.46338400404866e-05,
+      "loss": 1.3977,
+      "step": 768000
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.449326160709264e-05,
+      "loss": 1.3999,
+      "step": 768500
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 8.43526831736987e-05,
+      "loss": 1.3902,
+      "step": 769000
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 8.421210474030477e-05,
+      "loss": 1.4278,
+      "step": 769500
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 8.407152630691084e-05,
+      "loss": 1.4353,
+      "step": 770000
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 8.393094787351689e-05,
+      "loss": 1.4072,
+      "step": 770500
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 8.379036944012295e-05,
+      "loss": 1.4325,
+      "step": 771000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 8.364979100672901e-05,
+      "loss": 1.3832,
+      "step": 771500
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 8.350921257333508e-05,
+      "loss": 1.3876,
+      "step": 772000
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.336863413994113e-05,
+      "loss": 1.3998,
+      "step": 772500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 8.32280557065472e-05,
+      "loss": 1.4081,
+      "step": 773000
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.308747727315326e-05,
+      "loss": 1.4361,
+      "step": 773500
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 8.294689883975932e-05,
+      "loss": 1.3762,
+      "step": 774000
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 8.280632040636537e-05,
+      "loss": 1.4322,
+      "step": 774500
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 8.266574197297145e-05,
+      "loss": 1.3878,
+      "step": 775000
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 8.252516353957751e-05,
+      "loss": 1.3681,
+      "step": 775500
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.238458510618357e-05,
+      "loss": 1.4005,
+      "step": 776000
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 8.224400667278962e-05,
+      "loss": 1.424,
+      "step": 776500
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 8.21034282393957e-05,
+      "loss": 1.4016,
+      "step": 777000
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 8.196284980600176e-05,
+      "loss": 1.3916,
+      "step": 777500
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 8.182227137260782e-05,
+      "loss": 1.449,
+      "step": 778000
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 8.168169293921387e-05,
+      "loss": 1.4048,
+      "step": 778500
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 8.154111450581994e-05,
+      "loss": 1.4209,
+      "step": 779000
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 8.1400536072426e-05,
+      "loss": 1.4021,
+      "step": 779500
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 8.125995763903207e-05,
+      "loss": 1.3813,
+      "step": 780000
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 8.111937920563812e-05,
+      "loss": 1.3991,
+      "step": 780500
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 8.097880077224418e-05,
+      "loss": 1.3895,
+      "step": 781000
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 8.083822233885025e-05,
+      "loss": 1.383,
+      "step": 781500
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.069764390545632e-05,
+      "loss": 1.3922,
+      "step": 782000
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 8.055706547206236e-05,
+      "loss": 1.3866,
+      "step": 782500
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 8.041648703866843e-05,
+      "loss": 1.404,
+      "step": 783000
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 8.02759086052745e-05,
+      "loss": 1.3895,
+      "step": 783500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 8.013533017188056e-05,
+      "loss": 1.4153,
+      "step": 784000
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 7.999475173848661e-05,
+      "loss": 1.339,
+      "step": 784500
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 7.985417330509267e-05,
+      "loss": 1.3512,
+      "step": 785000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 7.971359487169875e-05,
+      "loss": 1.3543,
+      "step": 785500
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 7.957301643830481e-05,
+      "loss": 1.3823,
+      "step": 786000
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 7.943243800491086e-05,
+      "loss": 1.3589,
+      "step": 786500
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 7.929185957151692e-05,
+      "loss": 1.3661,
+      "step": 787000
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 7.9151281138123e-05,
+      "loss": 1.3565,
+      "step": 787500
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 7.901070270472906e-05,
+      "loss": 1.3667,
+      "step": 788000
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 7.887012427133511e-05,
+      "loss": 1.3758,
+      "step": 788500
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 7.872954583794117e-05,
+      "loss": 1.3565,
+      "step": 789000
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 7.858896740454723e-05,
+      "loss": 1.366,
+      "step": 789500
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 7.84483889711533e-05,
+      "loss": 1.3763,
+      "step": 790000
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 7.830781053775935e-05,
+      "loss": 1.3381,
+      "step": 790500
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 7.816723210436542e-05,
+      "loss": 1.3546,
+      "step": 791000
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 7.802665367097148e-05,
+      "loss": 1.3326,
+      "step": 791500
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 7.788607523757755e-05,
+      "loss": 1.3528,
+      "step": 792000
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 7.77454968041836e-05,
+      "loss": 1.3642,
+      "step": 792500
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 7.760491837078966e-05,
+      "loss": 1.3849,
+      "step": 793000
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 7.746433993739573e-05,
+      "loss": 1.3709,
+      "step": 793500
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 7.73237615040018e-05,
+      "loss": 1.3701,
+      "step": 794000
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 7.718318307060786e-05,
+      "loss": 1.3805,
+      "step": 794500
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 7.704260463721391e-05,
+      "loss": 1.3518,
+      "step": 795000
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 7.690202620381997e-05,
+      "loss": 1.3907,
+      "step": 795500
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 7.676144777042605e-05,
+      "loss": 1.3648,
+      "step": 796000
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 7.662086933703211e-05,
+      "loss": 1.3384,
+      "step": 796500
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 7.648029090363816e-05,
+      "loss": 1.3522,
+      "step": 797000
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 7.633971247024422e-05,
+      "loss": 1.3454,
+      "step": 797500
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 7.619913403685028e-05,
+      "loss": 1.3731,
+      "step": 798000
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 7.605855560345636e-05,
+      "loss": 1.3833,
+      "step": 798500
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 7.591797717006241e-05,
+      "loss": 1.3655,
+      "step": 799000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 7.577739873666847e-05,
+      "loss": 1.3855,
+      "step": 799500
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 7.563682030327453e-05,
+      "loss": 1.3427,
+      "step": 800000
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 7.54962418698806e-05,
+      "loss": 1.3713,
+      "step": 800500
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 7.535566343648666e-05,
+      "loss": 1.3618,
+      "step": 801000
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 7.521508500309272e-05,
+      "loss": 1.3967,
+      "step": 801500
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 7.507450656969878e-05,
+      "loss": 1.3566,
+      "step": 802000
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 7.493392813630484e-05,
+      "loss": 1.3438,
+      "step": 802500
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 7.47933497029109e-05,
+      "loss": 1.37,
+      "step": 803000
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 7.465277126951696e-05,
+      "loss": 1.3862,
+      "step": 803500
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 7.451219283612303e-05,
+      "loss": 1.3876,
+      "step": 804000
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 7.437161440272909e-05,
+      "loss": 1.3405,
+      "step": 804500
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 7.423103596933515e-05,
+      "loss": 1.3513,
+      "step": 805000
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 7.409045753594121e-05,
+      "loss": 1.3928,
+      "step": 805500
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 7.394987910254727e-05,
+      "loss": 1.3646,
+      "step": 806000
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 7.380930066915334e-05,
+      "loss": 1.3238,
+      "step": 806500
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 7.36687222357594e-05,
+      "loss": 1.361,
+      "step": 807000
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 7.352814380236546e-05,
+      "loss": 1.3775,
+      "step": 807500
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 7.338756536897152e-05,
+      "loss": 1.3854,
+      "step": 808000
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 7.324698693557758e-05,
+      "loss": 1.367,
+      "step": 808500
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 7.310640850218365e-05,
+      "loss": 1.3572,
+      "step": 809000
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 7.296583006878971e-05,
+      "loss": 1.3752,
+      "step": 809500
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 7.282525163539577e-05,
+      "loss": 1.3821,
+      "step": 810000
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 7.268467320200183e-05,
+      "loss": 1.3616,
+      "step": 810500
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 7.25440947686079e-05,
+      "loss": 1.3904,
+      "step": 811000
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 7.240351633521396e-05,
+      "loss": 1.3583,
+      "step": 811500
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 7.226293790182002e-05,
+      "loss": 1.4081,
+      "step": 812000
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 7.212235946842608e-05,
+      "loss": 1.3632,
+      "step": 812500
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 7.198178103503214e-05,
+      "loss": 1.362,
+      "step": 813000
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 7.18412026016382e-05,
+      "loss": 1.3824,
+      "step": 813500
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 7.170062416824427e-05,
+      "loss": 1.3701,
+      "step": 814000
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 7.156004573485033e-05,
+      "loss": 1.3528,
+      "step": 814500
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 7.141946730145639e-05,
+      "loss": 1.3577,
+      "step": 815000
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 7.127888886806245e-05,
+      "loss": 1.3679,
+      "step": 815500
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 7.113831043466851e-05,
+      "loss": 1.3678,
+      "step": 816000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 7.099773200127456e-05,
+      "loss": 1.3471,
+      "step": 816500
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 7.085715356788064e-05,
+      "loss": 1.377,
+      "step": 817000
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 7.071657513448669e-05,
+      "loss": 1.36,
+      "step": 817500
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 7.057599670109276e-05,
+      "loss": 1.3942,
+      "step": 818000
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 7.043541826769881e-05,
+      "loss": 1.369,
+      "step": 818500
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 7.029483983430488e-05,
+      "loss": 1.367,
+      "step": 819000
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 7.015426140091093e-05,
+      "loss": 1.3907,
+      "step": 819500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 7.001368296751701e-05,
+      "loss": 1.3379,
+      "step": 820000
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 6.987310453412306e-05,
+      "loss": 1.335,
+      "step": 820500
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 6.973252610072913e-05,
+      "loss": 1.3061,
+      "step": 821000
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 6.959194766733518e-05,
+      "loss": 1.2759,
+      "step": 821500
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 6.945136923394126e-05,
+      "loss": 1.3039,
+      "step": 822000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 6.931079080054732e-05,
+      "loss": 1.329,
+      "step": 822500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 6.917021236715338e-05,
+      "loss": 1.3405,
+      "step": 823000
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 6.902963393375944e-05,
+      "loss": 1.3211,
+      "step": 823500
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 6.88890555003655e-05,
+      "loss": 1.3352,
+      "step": 824000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 6.874847706697157e-05,
+      "loss": 1.338,
+      "step": 824500
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 6.860789863357761e-05,
+      "loss": 1.3533,
+      "step": 825000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 6.846732020018369e-05,
+      "loss": 1.324,
+      "step": 825500
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 6.832674176678974e-05,
+      "loss": 1.3076,
+      "step": 826000
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 6.818616333339581e-05,
+      "loss": 1.3694,
+      "step": 826500
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 6.804558490000186e-05,
+      "loss": 1.3091,
+      "step": 827000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 6.790500646660794e-05,
+      "loss": 1.3354,
+      "step": 827500
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 6.776442803321399e-05,
+      "loss": 1.3473,
+      "step": 828000
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 6.762384959982006e-05,
+      "loss": 1.3365,
+      "step": 828500
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 6.748327116642611e-05,
+      "loss": 1.3606,
+      "step": 829000
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 6.734269273303219e-05,
+      "loss": 1.3394,
+      "step": 829500
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 6.720211429963823e-05,
+      "loss": 1.3267,
+      "step": 830000
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 6.706153586624431e-05,
+      "loss": 1.3294,
+      "step": 830500
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 6.692095743285036e-05,
+      "loss": 1.3472,
+      "step": 831000
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 6.678037899945643e-05,
+      "loss": 1.303,
+      "step": 831500
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 6.663980056606248e-05,
+      "loss": 1.3304,
+      "step": 832000
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 6.649922213266856e-05,
+      "loss": 1.3474,
+      "step": 832500
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 6.63586436992746e-05,
+      "loss": 1.3662,
+      "step": 833000
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 6.621806526588067e-05,
+      "loss": 1.3298,
+      "step": 833500
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 6.607748683248673e-05,
+      "loss": 1.3309,
+      "step": 834000
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 6.593690839909279e-05,
+      "loss": 1.3172,
+      "step": 834500
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 6.579632996569885e-05,
+      "loss": 1.3243,
+      "step": 835000
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 6.565575153230491e-05,
+      "loss": 1.3123,
+      "step": 835500
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 6.551517309891098e-05,
+      "loss": 1.3471,
+      "step": 836000
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 6.537459466551704e-05,
+      "loss": 1.3276,
+      "step": 836500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 6.52340162321231e-05,
+      "loss": 1.3227,
+      "step": 837000
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 6.509343779872916e-05,
+      "loss": 1.3373,
+      "step": 837500
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 6.495285936533522e-05,
+      "loss": 1.3258,
+      "step": 838000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 6.481228093194129e-05,
+      "loss": 1.3358,
+      "step": 838500
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 6.467170249854735e-05,
+      "loss": 1.3238,
+      "step": 839000
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 6.453112406515341e-05,
+      "loss": 1.3125,
+      "step": 839500
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 6.439054563175947e-05,
+      "loss": 1.3413,
+      "step": 840000
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 6.424996719836553e-05,
+      "loss": 1.331,
+      "step": 840500
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 6.41093887649716e-05,
+      "loss": 1.346,
+      "step": 841000
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 6.396881033157766e-05,
+      "loss": 1.3422,
+      "step": 841500
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 6.382823189818372e-05,
+      "loss": 1.3346,
+      "step": 842000
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 6.368765346478978e-05,
+      "loss": 1.3375,
+      "step": 842500
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 6.354707503139584e-05,
+      "loss": 1.334,
+      "step": 843000
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 6.34064965980019e-05,
+      "loss": 1.311,
+      "step": 843500
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 6.326591816460797e-05,
+      "loss": 1.3061,
+      "step": 844000
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 6.312533973121403e-05,
+      "loss": 1.3373,
+      "step": 844500
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 6.298476129782009e-05,
+      "loss": 1.3494,
+      "step": 845000
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 6.284418286442615e-05,
+      "loss": 1.326,
+      "step": 845500
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 6.270360443103222e-05,
+      "loss": 1.348,
+      "step": 846000
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 6.256302599763828e-05,
+      "loss": 1.3473,
+      "step": 846500
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 6.242244756424434e-05,
+      "loss": 1.3397,
+      "step": 847000
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 6.22818691308504e-05,
+      "loss": 1.3583,
+      "step": 847500
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 6.214129069745646e-05,
+      "loss": 1.3292,
+      "step": 848000
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 6.200071226406253e-05,
+      "loss": 1.3455,
+      "step": 848500
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 6.186013383066859e-05,
+      "loss": 1.3458,
+      "step": 849000
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 6.171955539727465e-05,
+      "loss": 1.3477,
+      "step": 849500
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 6.157897696388071e-05,
+      "loss": 1.3462,
+      "step": 850000
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 6.143839853048677e-05,
+      "loss": 1.328,
+      "step": 850500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 6.129782009709283e-05,
+      "loss": 1.3453,
+      "step": 851000
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 6.11572416636989e-05,
+      "loss": 1.3241,
+      "step": 851500
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 6.101666323030496e-05,
+      "loss": 1.3564,
+      "step": 852000
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 6.087608479691102e-05,
+      "loss": 1.3582,
+      "step": 852500
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 6.073550636351708e-05,
+      "loss": 1.3355,
+      "step": 853000
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 6.0594927930123144e-05,
+      "loss": 1.3438,
+      "step": 853500
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 6.0454349496729206e-05,
+      "loss": 1.3057,
+      "step": 854000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 6.031377106333526e-05,
+      "loss": 1.3214,
+      "step": 854500
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 6.017319262994133e-05,
+      "loss": 1.3221,
+      "step": 855000
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 6.0032614196547385e-05,
+      "loss": 1.2929,
+      "step": 855500
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 5.9892035763153454e-05,
+      "loss": 1.3036,
+      "step": 856000
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.975145732975951e-05,
+      "loss": 1.3002,
+      "step": 856500
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 5.961087889636558e-05,
+      "loss": 1.2967,
+      "step": 857000
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.947030046297163e-05,
+      "loss": 1.2968,
+      "step": 857500
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 5.93297220295777e-05,
+      "loss": 1.298,
+      "step": 858000
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.918914359618376e-05,
+      "loss": 1.2798,
+      "step": 858500
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 5.9048565162789826e-05,
+      "loss": 1.2808,
+      "step": 859000
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.890798672939588e-05,
+      "loss": 1.3283,
+      "step": 859500
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.876740829600195e-05,
+      "loss": 1.3048,
+      "step": 860000
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.8626829862608005e-05,
+      "loss": 1.2911,
+      "step": 860500
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.848625142921407e-05,
+      "loss": 1.2803,
+      "step": 861000
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.834567299582013e-05,
+      "loss": 1.3098,
+      "step": 861500
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.82050945624262e-05,
+      "loss": 1.3068,
+      "step": 862000
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 5.806451612903225e-05,
+      "loss": 1.2908,
+      "step": 862500
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.7923937695638314e-05,
+      "loss": 1.2579,
+      "step": 863000
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 5.7783359262244376e-05,
+      "loss": 1.3015,
+      "step": 863500
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.764278082885044e-05,
+      "loss": 1.2958,
+      "step": 864000
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 5.75022023954565e-05,
+      "loss": 1.3135,
+      "step": 864500
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.736162396206256e-05,
+      "loss": 1.2794,
+      "step": 865000
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 5.7221045528668624e-05,
+      "loss": 1.2913,
+      "step": 865500
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.7080467095274686e-05,
+      "loss": 1.2561,
+      "step": 866000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 5.693988866188075e-05,
+      "loss": 1.3083,
+      "step": 866500
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.679931022848681e-05,
+      "loss": 1.282,
+      "step": 867000
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.6658731795092865e-05,
+      "loss": 1.2826,
+      "step": 867500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.6518153361698934e-05,
+      "loss": 1.2866,
+      "step": 868000
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.637757492830499e-05,
+      "loss": 1.3105,
+      "step": 868500
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.623699649491106e-05,
+      "loss": 1.3,
+      "step": 869000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.609641806151711e-05,
+      "loss": 1.2891,
+      "step": 869500
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 5.595583962812318e-05,
+      "loss": 1.3011,
+      "step": 870000
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.5815261194729236e-05,
+      "loss": 1.3177,
+      "step": 870500
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 5.5674682761335305e-05,
+      "loss": 1.3146,
+      "step": 871000
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.553410432794136e-05,
+      "loss": 1.3113,
+      "step": 871500
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 5.539352589454743e-05,
+      "loss": 1.29,
+      "step": 872000
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.5252947461153484e-05,
+      "loss": 1.3293,
+      "step": 872500
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 5.511236902775955e-05,
+      "loss": 1.3065,
+      "step": 873000
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.497179059436561e-05,
+      "loss": 1.2612,
+      "step": 873500
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 5.483121216097168e-05,
+      "loss": 1.3006,
+      "step": 874000
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.469063372757773e-05,
+      "loss": 1.2845,
+      "step": 874500
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.45500552941838e-05,
+      "loss": 1.3217,
+      "step": 875000
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.4409476860789856e-05,
+      "loss": 1.3276,
+      "step": 875500
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.426889842739592e-05,
+      "loss": 1.2993,
+      "step": 876000
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.412831999400198e-05,
+      "loss": 1.2916,
+      "step": 876500
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.398774156060804e-05,
+      "loss": 1.3397,
+      "step": 877000
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.3847163127214103e-05,
+      "loss": 1.3067,
+      "step": 877500
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.3706584693820165e-05,
+      "loss": 1.3235,
+      "step": 878000
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 5.356600626042623e-05,
+      "loss": 1.2989,
+      "step": 878500
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.342542782703229e-05,
+      "loss": 1.2858,
+      "step": 879000
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 5.328484939363835e-05,
+      "loss": 1.3062,
+      "step": 879500
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.314427096024441e-05,
+      "loss": 1.2865,
+      "step": 880000
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.300369252685048e-05,
+      "loss": 1.2966,
+      "step": 880500
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.286311409345654e-05,
+      "loss": 1.2927,
+      "step": 881000
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.2722535660062606e-05,
+      "loss": 1.2981,
+      "step": 881500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.258195722666866e-05,
+      "loss": 1.3123,
+      "step": 882000
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 5.244137879327473e-05,
+      "loss": 1.3186,
+      "step": 882500
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.2300800359880785e-05,
+      "loss": 1.2962,
+      "step": 883000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.2160221926486853e-05,
+      "loss": 1.2899,
+      "step": 883500
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.201964349309291e-05,
+      "loss": 1.3057,
+      "step": 884000
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.187906505969897e-05,
+      "loss": 1.2949,
+      "step": 884500
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.173848662630503e-05,
+      "loss": 1.2874,
+      "step": 885000
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.1597908192911094e-05,
+      "loss": 1.3297,
+      "step": 885500
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.1457329759517156e-05,
+      "loss": 1.3324,
+      "step": 886000
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.131675132612322e-05,
+      "loss": 1.3237,
+      "step": 886500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.117617289272928e-05,
+      "loss": 1.3252,
+      "step": 887000
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.103559445933534e-05,
+      "loss": 1.2741,
+      "step": 887500
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.0895016025941404e-05,
+      "loss": 1.3208,
+      "step": 888000
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 5.0754437592547466e-05,
+      "loss": 1.2998,
+      "step": 888500
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.061385915915353e-05,
+      "loss": 1.2971,
+      "step": 889000
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 5.047328072575959e-05,
+      "loss": 1.3066,
+      "step": 889500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.0332702292365645e-05,
+      "loss": 1.3289,
+      "step": 890000
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 5.0192123858971714e-05,
+      "loss": 1.3317,
+      "step": 890500
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.005154542557777e-05,
+      "loss": 1.3158,
+      "step": 891000
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 4.991096699218384e-05,
+      "loss": 1.2461,
+      "step": 891500
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.977038855878989e-05,
+      "loss": 1.2825,
+      "step": 892000
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 4.962981012539596e-05,
+      "loss": 1.2938,
+      "step": 892500
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.9489231692002017e-05,
+      "loss": 1.2464,
+      "step": 893000
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.9348653258608085e-05,
+      "loss": 1.2535,
+      "step": 893500
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.920807482521414e-05,
+      "loss": 1.2678,
+      "step": 894000
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.906749639182021e-05,
+      "loss": 1.2809,
+      "step": 894500
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.8926917958426264e-05,
+      "loss": 1.2612,
+      "step": 895000
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.878633952503233e-05,
+      "loss": 1.2808,
+      "step": 895500
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 4.864576109163839e-05,
+      "loss": 1.2758,
+      "step": 896000
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.850518265824446e-05,
+      "loss": 1.267,
+      "step": 896500
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.836460422485051e-05,
+      "loss": 1.2714,
+      "step": 897000
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.822402579145658e-05,
+      "loss": 1.2657,
+      "step": 897500
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 4.8083447358062636e-05,
+      "loss": 1.2706,
+      "step": 898000
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.79428689246687e-05,
+      "loss": 1.2595,
+      "step": 898500
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 4.780229049127476e-05,
+      "loss": 1.2448,
+      "step": 899000
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.766171205788082e-05,
+      "loss": 1.2639,
+      "step": 899500
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 4.7521133624486884e-05,
+      "loss": 1.2664,
+      "step": 900000
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.7380555191092946e-05,
+      "loss": 1.2512,
+      "step": 900500
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.723997675769901e-05,
+      "loss": 1.2682,
+      "step": 901000
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.709939832430507e-05,
+      "loss": 1.2866,
+      "step": 901500
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.695881989091113e-05,
+      "loss": 1.2619,
+      "step": 902000
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.681824145751719e-05,
+      "loss": 1.2451,
+      "step": 902500
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.6677663024123255e-05,
+      "loss": 1.2759,
+      "step": 903000
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 4.653708459072932e-05,
+      "loss": 1.2691,
+      "step": 903500
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.639650615733537e-05,
+      "loss": 1.2742,
+      "step": 904000
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 4.625592772394144e-05,
+      "loss": 1.263,
+      "step": 904500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.6115349290547496e-05,
+      "loss": 1.2801,
+      "step": 905000
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 4.5974770857153565e-05,
+      "loss": 1.281,
+      "step": 905500
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.583419242375962e-05,
+      "loss": 1.2552,
+      "step": 906000
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 4.569361399036569e-05,
+      "loss": 1.2866,
+      "step": 906500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.5553035556971744e-05,
+      "loss": 1.2792,
+      "step": 907000
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 4.541245712357781e-05,
+      "loss": 1.2564,
+      "step": 907500
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.527187869018387e-05,
+      "loss": 1.2652,
+      "step": 908000
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.5131300256789936e-05,
+      "loss": 1.2889,
+      "step": 908500
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.499072182339599e-05,
+      "loss": 1.2696,
+      "step": 909000
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.485014339000206e-05,
+      "loss": 1.2568,
+      "step": 909500
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.470956495660812e-05,
+      "loss": 1.2891,
+      "step": 910000
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.4568986523214184e-05,
+      "loss": 1.2697,
+      "step": 910500
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.4428408089820246e-05,
+      "loss": 1.2713,
+      "step": 911000
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.428782965642631e-05,
+      "loss": 1.2878,
+      "step": 911500
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 4.414725122303237e-05,
+      "loss": 1.2697,
+      "step": 912000
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.4006672789638425e-05,
+      "loss": 1.2866,
+      "step": 912500
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 4.3866094356244494e-05,
+      "loss": 1.2519,
+      "step": 913000
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.372551592285055e-05,
+      "loss": 1.2736,
+      "step": 913500
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 4.358493748945662e-05,
+      "loss": 1.2847,
+      "step": 914000
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.344435905606267e-05,
+      "loss": 1.2802,
+      "step": 914500
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 4.330378062266874e-05,
+      "loss": 1.2639,
+      "step": 915000
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.31632021892748e-05,
+      "loss": 1.2613,
+      "step": 915500
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 4.3022623755880865e-05,
+      "loss": 1.2673,
+      "step": 916000
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.288204532248692e-05,
+      "loss": 1.2737,
+      "step": 916500
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.274146688909299e-05,
+      "loss": 1.2877,
+      "step": 917000
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.2600888455699044e-05,
+      "loss": 1.2975,
+      "step": 917500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.246031002230511e-05,
+      "loss": 1.2643,
+      "step": 918000
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.231973158891117e-05,
+      "loss": 1.3068,
+      "step": 918500
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.217915315551724e-05,
+      "loss": 1.2768,
+      "step": 919000
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 4.203857472212329e-05,
+      "loss": 1.2809,
+      "step": 919500
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.1897996288729354e-05,
+      "loss": 1.2895,
+      "step": 920000
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.1757417855335416e-05,
+      "loss": 1.289,
+      "step": 920500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.161683942194148e-05,
+      "loss": 1.2685,
+      "step": 921000
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 4.147626098854754e-05,
+      "loss": 1.2986,
+      "step": 921500
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.13356825551536e-05,
+      "loss": 1.2887,
+      "step": 922000
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 4.1195104121759664e-05,
+      "loss": 1.2693,
+      "step": 922500
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.1054525688365726e-05,
+      "loss": 1.2494,
+      "step": 923000
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.091394725497179e-05,
+      "loss": 1.2642,
+      "step": 923500
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.077336882157785e-05,
+      "loss": 1.2777,
+      "step": 924000
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.063279038818391e-05,
+      "loss": 1.2755,
+      "step": 924500
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.0492211954789973e-05,
+      "loss": 1.2565,
+      "step": 925000
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.035163352139603e-05,
+      "loss": 1.2715,
+      "step": 925500
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.02110550880021e-05,
+      "loss": 1.2794,
+      "step": 926000
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.007047665460815e-05,
+      "loss": 1.2859,
+      "step": 926500
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 3.992989822121422e-05,
+      "loss": 1.2242,
+      "step": 927000
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 3.9789319787820276e-05,
+      "loss": 1.244,
+      "step": 927500
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 3.9648741354426345e-05,
+      "loss": 1.2497,
+      "step": 928000
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 3.95081629210324e-05,
+      "loss": 1.2126,
+      "step": 928500
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 3.936758448763847e-05,
+      "loss": 1.2374,
+      "step": 929000
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 3.9227006054244524e-05,
+      "loss": 1.2321,
+      "step": 929500
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 3.908642762085059e-05,
+      "loss": 1.2191,
+      "step": 930000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3.894584918745665e-05,
+      "loss": 1.2377,
+      "step": 930500
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 3.8805270754062717e-05,
+      "loss": 1.2306,
+      "step": 931000
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 3.866469232066877e-05,
+      "loss": 1.2093,
+      "step": 931500
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 3.852411388727484e-05,
+      "loss": 1.2353,
+      "step": 932000
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 3.8383535453880896e-05,
+      "loss": 1.2532,
+      "step": 932500
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 3.8242957020486964e-05,
+      "loss": 1.2356,
+      "step": 933000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3.810237858709302e-05,
+      "loss": 1.2562,
+      "step": 933500
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 3.796180015369908e-05,
+      "loss": 1.2434,
+      "step": 934000
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 3.782122172030514e-05,
+      "loss": 1.2458,
+      "step": 934500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 3.7680643286911205e-05,
+      "loss": 1.2291,
+      "step": 935000
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 3.754006485351727e-05,
+      "loss": 1.2355,
+      "step": 935500
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 3.739948642012333e-05,
+      "loss": 1.2213,
+      "step": 936000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 3.725890798672939e-05,
+      "loss": 1.2332,
+      "step": 936500
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 3.711832955333545e-05,
+      "loss": 1.2243,
+      "step": 937000
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 3.6977751119941515e-05,
+      "loss": 1.2342,
+      "step": 937500
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 3.683717268654758e-05,
+      "loss": 1.276,
+      "step": 938000
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 3.669659425315364e-05,
+      "loss": 1.2475,
+      "step": 938500
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 3.65560158197597e-05,
+      "loss": 1.2522,
+      "step": 939000
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 3.641543738636576e-05,
+      "loss": 1.2453,
+      "step": 939500
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 3.6274858952971825e-05,
+      "loss": 1.2416,
+      "step": 940000
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3.6134280519577886e-05,
+      "loss": 1.2485,
+      "step": 940500
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 3.599370208618395e-05,
+      "loss": 1.2451,
+      "step": 941000
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 3.585312365279001e-05,
+      "loss": 1.2666,
+      "step": 941500
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 3.571254521939607e-05,
+      "loss": 1.2476,
+      "step": 942000
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 3.5571966786002134e-05,
+      "loss": 1.2161,
+      "step": 942500
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 3.5431388352608196e-05,
+      "loss": 1.2345,
+      "step": 943000
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3.529080991921426e-05,
+      "loss": 1.247,
+      "step": 943500
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 3.515023148582032e-05,
+      "loss": 1.2497,
+      "step": 944000
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 3.500965305242638e-05,
+      "loss": 1.2481,
+      "step": 944500
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 3.4869074619032444e-05,
+      "loss": 1.2761,
+      "step": 945000
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 3.4728496185638506e-05,
+      "loss": 1.2396,
+      "step": 945500
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 3.458791775224457e-05,
+      "loss": 1.2603,
+      "step": 946000
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 3.444733931885063e-05,
+      "loss": 1.2455,
+      "step": 946500
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 3.430676088545669e-05,
+      "loss": 1.2233,
+      "step": 947000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 3.416618245206275e-05,
+      "loss": 1.2498,
+      "step": 947500
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 3.402560401866881e-05,
+      "loss": 1.2853,
+      "step": 948000
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 3.388502558527487e-05,
+      "loss": 1.2309,
+      "step": 948500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 3.374444715188093e-05,
+      "loss": 1.2246,
+      "step": 949000
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 3.3603868718486994e-05,
+      "loss": 1.2179,
+      "step": 949500
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3.3463290285093056e-05,
+      "loss": 1.2463,
+      "step": 950000
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 3.332271185169912e-05,
+      "loss": 1.2507,
+      "step": 950500
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3.318213341830518e-05,
+      "loss": 1.242,
+      "step": 951000
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 3.304155498491124e-05,
+      "loss": 1.2434,
+      "step": 951500
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3.2900976551517304e-05,
+      "loss": 1.244,
+      "step": 952000
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 3.2760398118123366e-05,
+      "loss": 1.2368,
+      "step": 952500
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.261981968472943e-05,
+      "loss": 1.2472,
+      "step": 953000
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 3.247924125133549e-05,
+      "loss": 1.251,
+      "step": 953500
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3.233866281794156e-05,
+      "loss": 1.2534,
+      "step": 954000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3.219808438454762e-05,
+      "loss": 1.2408,
+      "step": 954500
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 3.205750595115368e-05,
+      "loss": 1.2493,
+      "step": 955000
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3.1916927517759744e-05,
+      "loss": 1.2163,
+      "step": 955500
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 3.17763490843658e-05,
+      "loss": 1.2495,
+      "step": 956000
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3.163577065097186e-05,
+      "loss": 1.2494,
+      "step": 956500
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 3.1495192217577923e-05,
+      "loss": 1.229,
+      "step": 957000
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3.1354613784183985e-05,
+      "loss": 1.257,
+      "step": 957500
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3.121403535079005e-05,
+      "loss": 1.2336,
+      "step": 958000
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3.107345691739611e-05,
+      "loss": 1.2409,
+      "step": 958500
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.093287848400217e-05,
+      "loss": 1.2139,
+      "step": 959000
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3.079230005060823e-05,
+      "loss": 1.2512,
+      "step": 959500
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3.0651721617214295e-05,
+      "loss": 1.2627,
+      "step": 960000
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 3.0511143183820357e-05,
+      "loss": 1.251,
+      "step": 960500
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3.037056475042642e-05,
+      "loss": 1.2024,
+      "step": 961000
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 3.022998631703248e-05,
+      "loss": 1.2204,
+      "step": 961500
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3.0089407883638543e-05,
+      "loss": 1.2593,
+      "step": 962000
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 2.9948829450244605e-05,
+      "loss": 1.2286,
+      "step": 962500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 2.9808251016850667e-05,
+      "loss": 1.2079,
+      "step": 963000
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 2.966767258345673e-05,
+      "loss": 1.2022,
+      "step": 963500
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 2.952709415006279e-05,
+      "loss": 1.1943,
+      "step": 964000
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 2.9386515716668852e-05,
+      "loss": 1.1952,
+      "step": 964500
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 2.9245937283274914e-05,
+      "loss": 1.2296,
+      "step": 965000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 2.9105358849880976e-05,
+      "loss": 1.2143,
+      "step": 965500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 2.8964780416487035e-05,
+      "loss": 1.2432,
+      "step": 966000
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 2.8824201983093097e-05,
+      "loss": 1.2106,
+      "step": 966500
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 2.868362354969916e-05,
+      "loss": 1.2221,
+      "step": 967000
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 2.854304511630522e-05,
+      "loss": 1.2327,
+      "step": 967500
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 2.8402466682911283e-05,
+      "loss": 1.2119,
+      "step": 968000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 2.8261888249517344e-05,
+      "loss": 1.2094,
+      "step": 968500
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 2.8121309816123406e-05,
+      "loss": 1.1851,
+      "step": 969000
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 2.798073138272947e-05,
+      "loss": 1.2177,
+      "step": 969500
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 2.784015294933553e-05,
+      "loss": 1.253,
+      "step": 970000
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 2.7699574515941592e-05,
+      "loss": 1.2167,
+      "step": 970500
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 2.7558996082547654e-05,
+      "loss": 1.2125,
+      "step": 971000
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 2.7418417649153716e-05,
+      "loss": 1.1955,
+      "step": 971500
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 2.7277839215759778e-05,
+      "loss": 1.1958,
+      "step": 972000
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 2.713726078236584e-05,
+      "loss": 1.229,
+      "step": 972500
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 2.69966823489719e-05,
+      "loss": 1.2004,
+      "step": 973000
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 2.685610391557796e-05,
+      "loss": 1.2221,
+      "step": 973500
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 2.6715525482184022e-05,
+      "loss": 1.2211,
+      "step": 974000
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 2.6574947048790084e-05,
+      "loss": 1.2279,
+      "step": 974500
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 2.6434368615396146e-05,
+      "loss": 1.229,
+      "step": 975000
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 2.6293790182002208e-05,
+      "loss": 1.2155,
+      "step": 975500
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 2.615321174860827e-05,
+      "loss": 1.2055,
+      "step": 976000
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 2.6012633315214332e-05,
+      "loss": 1.2284,
+      "step": 976500
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 2.5872054881820394e-05,
+      "loss": 1.2137,
+      "step": 977000
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 2.5731476448426456e-05,
+      "loss": 1.2012,
+      "step": 977500
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 2.5590898015032518e-05,
+      "loss": 1.2012,
+      "step": 978000
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 2.545031958163858e-05,
+      "loss": 1.2168,
+      "step": 978500
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 2.530974114824464e-05,
+      "loss": 1.2162,
+      "step": 979000
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 2.5169162714850704e-05,
+      "loss": 1.2203,
+      "step": 979500
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 2.5028584281456762e-05,
+      "loss": 1.1994,
+      "step": 980000
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 2.4888005848062824e-05,
+      "loss": 1.2206,
+      "step": 980500
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 2.4747427414668886e-05,
+      "loss": 1.2047,
+      "step": 981000
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 2.4606848981274948e-05,
+      "loss": 1.1918,
+      "step": 981500
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 2.446627054788101e-05,
+      "loss": 1.2074,
+      "step": 982000
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.4325692114487072e-05,
+      "loss": 1.2206,
+      "step": 982500
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 2.4185113681093137e-05,
+      "loss": 1.2193,
+      "step": 983000
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 2.40445352476992e-05,
+      "loss": 1.2269,
+      "step": 983500
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 2.390395681430526e-05,
+      "loss": 1.2168,
+      "step": 984000
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 2.3763378380911323e-05,
+      "loss": 1.2136,
+      "step": 984500
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 2.3622799947517385e-05,
+      "loss": 1.1938,
+      "step": 985000
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 2.3482221514123447e-05,
+      "loss": 1.2355,
+      "step": 985500
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 2.334164308072951e-05,
+      "loss": 1.2056,
+      "step": 986000
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 2.320106464733557e-05,
+      "loss": 1.2273,
+      "step": 986500
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 2.3060486213941633e-05,
+      "loss": 1.2103,
+      "step": 987000
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 2.2919907780547694e-05,
+      "loss": 1.1975,
+      "step": 987500
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 2.2779329347153753e-05,
+      "loss": 1.2034,
+      "step": 988000
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.2638750913759815e-05,
+      "loss": 1.2128,
+      "step": 988500
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 2.2498172480365877e-05,
+      "loss": 1.2214,
+      "step": 989000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 2.235759404697194e-05,
+      "loss": 1.2121,
+      "step": 989500
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 2.2217015613578e-05,
+      "loss": 1.2008,
+      "step": 990000
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 2.2076437180184063e-05,
+      "loss": 1.2235,
+      "step": 990500
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 2.1935858746790125e-05,
+      "loss": 1.224,
+      "step": 991000
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 2.1795280313396187e-05,
+      "loss": 1.2283,
+      "step": 991500
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 2.165470188000225e-05,
+      "loss": 1.1691,
+      "step": 992000
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 2.151412344660831e-05,
+      "loss": 1.2283,
+      "step": 992500
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 2.1373545013214372e-05,
+      "loss": 1.2298,
+      "step": 993000
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 2.1232966579820434e-05,
+      "loss": 1.2071,
+      "step": 993500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 2.1092388146426496e-05,
+      "loss": 1.201,
+      "step": 994000
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 2.0951809713032558e-05,
+      "loss": 1.2401,
+      "step": 994500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 2.0811231279638617e-05,
+      "loss": 1.2178,
+      "step": 995000
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 2.067065284624468e-05,
+      "loss": 1.2387,
+      "step": 995500
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 2.053007441285074e-05,
+      "loss": 1.2397,
+      "step": 996000
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 2.0389495979456802e-05,
+      "loss": 1.2406,
+      "step": 996500
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 2.0248917546062864e-05,
+      "loss": 1.2341,
+      "step": 997000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 2.0108339112668926e-05,
+      "loss": 1.2314,
+      "step": 997500
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 1.9967760679274988e-05,
+      "loss": 1.206,
+      "step": 998000
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 1.982718224588105e-05,
+      "loss": 1.1898,
+      "step": 998500
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 1.9686603812487112e-05,
+      "loss": 1.2027,
+      "step": 999000
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 1.9546025379093174e-05,
+      "loss": 1.2206,
+      "step": 999500
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 1.9405446945699236e-05,
+      "loss": 1.1764,
+      "step": 1000000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 1.9264868512305298e-05,
+      "loss": 1.1983,
+      "step": 1000500
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 1.912429007891136e-05,
+      "loss": 1.2146,
+      "step": 1001000
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 1.8983711645517422e-05,
+      "loss": 1.2142,
+      "step": 1001500
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 1.884313321212348e-05,
+      "loss": 1.212,
+      "step": 1002000
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 1.8702554778729542e-05,
+      "loss": 1.1655,
+      "step": 1002500
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 1.8561976345335604e-05,
+      "loss": 1.2233,
+      "step": 1003000
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 1.8421397911941666e-05,
+      "loss": 1.2042,
+      "step": 1003500
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 1.8280819478547728e-05,
+      "loss": 1.1557,
+      "step": 1004000
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 1.814024104515379e-05,
+      "loss": 1.2044,
+      "step": 1004500
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 1.7999662611759855e-05,
+      "loss": 1.2152,
+      "step": 1005000
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 1.7859084178365917e-05,
+      "loss": 1.1963,
+      "step": 1005500
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 1.7718505744971976e-05,
+      "loss": 1.1855,
+      "step": 1006000
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 1.7577927311578038e-05,
+      "loss": 1.2021,
+      "step": 1006500
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 1.74373488781841e-05,
+      "loss": 1.2276,
+      "step": 1007000
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 1.729677044479016e-05,
+      "loss": 1.2031,
+      "step": 1007500
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 1.7156192011396223e-05,
+      "loss": 1.197,
+      "step": 1008000
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 1.7015613578002285e-05,
+      "loss": 1.1679,
+      "step": 1008500
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 1.6875035144608347e-05,
+      "loss": 1.189,
+      "step": 1009000
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 1.673445671121441e-05,
+      "loss": 1.2135,
+      "step": 1009500
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 1.659387827782047e-05,
+      "loss": 1.1794,
+      "step": 1010000
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 1.6453299844426533e-05,
+      "loss": 1.1916,
+      "step": 1010500
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 1.6312721411032595e-05,
+      "loss": 1.1758,
+      "step": 1011000
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 1.6172142977638657e-05,
+      "loss": 1.1733,
+      "step": 1011500
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 1.603156454424472e-05,
+      "loss": 1.2269,
+      "step": 1012000
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 1.589098611085078e-05,
+      "loss": 1.2067,
+      "step": 1012500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 1.575040767745684e-05,
+      "loss": 1.1773,
+      "step": 1013000
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 1.56098292440629e-05,
+      "loss": 1.2163,
+      "step": 1013500
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 1.5469250810668963e-05,
+      "loss": 1.1957,
+      "step": 1014000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 1.5328672377275025e-05,
+      "loss": 1.2178,
+      "step": 1014500
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 1.5188093943881087e-05,
+      "loss": 1.1912,
+      "step": 1015000
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 1.5047515510487149e-05,
+      "loss": 1.1908,
+      "step": 1015500
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 1.4906937077093211e-05,
+      "loss": 1.2142,
+      "step": 1016000
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 1.4766358643699273e-05,
+      "loss": 1.1873,
+      "step": 1016500
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 1.4625780210305335e-05,
+      "loss": 1.171,
+      "step": 1017000
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 1.4485201776911397e-05,
+      "loss": 1.1901,
+      "step": 1017500
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 1.4344623343517457e-05,
+      "loss": 1.1877,
+      "step": 1018000
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 1.4204044910123519e-05,
+      "loss": 1.2195,
+      "step": 1018500
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 1.4063466476729581e-05,
+      "loss": 1.1885,
+      "step": 1019000
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 1.3922888043335643e-05,
+      "loss": 1.2032,
+      "step": 1019500
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 1.3782309609941706e-05,
+      "loss": 1.201,
+      "step": 1020000
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 1.3641731176547768e-05,
+      "loss": 1.1836,
+      "step": 1020500
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.350115274315383e-05,
+      "loss": 1.1866,
+      "step": 1021000
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 1.3360574309759892e-05,
+      "loss": 1.2105,
+      "step": 1021500
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 1.3219995876365954e-05,
+      "loss": 1.1799,
+      "step": 1022000
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 1.3079417442972014e-05,
+      "loss": 1.1936,
+      "step": 1022500
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 1.2938839009578076e-05,
+      "loss": 1.1856,
+      "step": 1023000
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 1.2798260576184138e-05,
+      "loss": 1.1831,
+      "step": 1023500
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.26576821427902e-05,
+      "loss": 1.1974,
+      "step": 1024000
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 1.2517103709396262e-05,
+      "loss": 1.2063,
+      "step": 1024500
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 1.2376525276002324e-05,
+      "loss": 1.1803,
+      "step": 1025000
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 1.2235946842608384e-05,
+      "loss": 1.2299,
+      "step": 1025500
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 1.2095368409214446e-05,
+      "loss": 1.185,
+      "step": 1026000
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.1954789975820508e-05,
+      "loss": 1.2083,
+      "step": 1026500
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 1.181421154242657e-05,
+      "loss": 1.1777,
+      "step": 1027000
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.1673633109032632e-05,
+      "loss": 1.2018,
+      "step": 1027500
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 1.1533054675638694e-05,
+      "loss": 1.1975,
+      "step": 1028000
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 1.1392476242244756e-05,
+      "loss": 1.1666,
+      "step": 1028500
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 1.1251897808850816e-05,
+      "loss": 1.1997,
+      "step": 1029000
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 1.1111319375456878e-05,
+      "loss": 1.2238,
+      "step": 1029500
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 1.097074094206294e-05,
+      "loss": 1.1982,
+      "step": 1030000
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 1.0830162508669002e-05,
+      "loss": 1.1779,
+      "step": 1030500
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 1.0689584075275064e-05,
+      "loss": 1.2023,
+      "step": 1031000
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 1.0549005641881126e-05,
+      "loss": 1.1825,
+      "step": 1031500
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 1.0408427208487188e-05,
+      "loss": 1.1994,
+      "step": 1032000
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 1.0267848775093248e-05,
+      "loss": 1.2162,
+      "step": 1032500
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 1.012727034169931e-05,
+      "loss": 1.2024,
+      "step": 1033000
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1069020,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 1.7996577873665385e+21,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}