diff --git "a/checkpoint-1069020/trainer_state.json" "b/checkpoint-1069020/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1069020/trainer_state.json"
@@ -0,0 +1,12849 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 29.99957905961919,
+  "eval_steps": 500,
+  "global_step": 1069020,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00015,
+      "loss": 33.6336,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0003,
+      "loss": 5.981,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002998595531918878,
+      "loss": 5.8563,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00029971910638377554,
+      "loss": 5.7975,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00029957865957566334,
+      "loss": 5.7589,
+      "step": 2500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00029943821276755115,
+      "loss": 5.7376,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00029929776595943896,
+      "loss": 5.7336,
+      "step": 3500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002991573191513267,
+      "loss": 5.6985,
+      "step": 4000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002990168723432145,
+      "loss": 5.6467,
+      "step": 4500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00029887642553510233,
+      "loss": 5.5431,
+      "step": 5000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002987359787269901,
+      "loss": 5.2658,
+      "step": 5500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002985955319188779,
+      "loss": 4.7901,
+      "step": 6000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002984550851107657,
+      "loss": 4.3833,
+      "step": 6500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002983146383026535,
+      "loss": 4.1403,
+      "step": 7000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00029817419149454126,
+      "loss": 3.8533,
+      "step": 7500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00029803374468642907,
+      "loss": 3.7172,
+      "step": 8000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0002978932978783169,
+      "loss": 3.6096,
+      "step": 8500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00029775285107020463,
+      "loss": 3.4788,
+      "step": 9000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00029761240426209244,
+      "loss": 3.3928,
+      "step": 9500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00029747195745398025,
+      "loss": 3.2847,
+      "step": 10000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00029733151064586806,
+      "loss": 3.2708,
+      "step": 10500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0002971910638377558,
+      "loss": 3.2045,
+      "step": 11000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002970506170296436,
+      "loss": 3.1776,
+      "step": 11500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002969101702215314,
+      "loss": 3.1367,
+      "step": 12000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002967697234134192,
+      "loss": 3.0834,
+      "step": 12500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000296629276605307,
+      "loss": 3.0597,
+      "step": 13000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0002964888297971948,
+      "loss": 3.047,
+      "step": 13500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0002963483829890826,
+      "loss": 3.0075,
+      "step": 14000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0002962079361809704,
+      "loss": 3.0121,
+      "step": 14500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00029606748937285816,
+      "loss": 2.9503,
+      "step": 15000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00029592704256474597,
+      "loss": 2.9327,
+      "step": 15500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0002957865957566337,
+      "loss": 2.9738,
+      "step": 16000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00029564614894852153,
+      "loss": 2.9376,
+      "step": 16500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00029550570214040934,
+      "loss": 2.8891,
+      "step": 17000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00029536525533229715,
+      "loss": 2.9003,
+      "step": 17500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0002952248085241849,
+      "loss": 2.8601,
+      "step": 18000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0002950843617160727,
+      "loss": 2.8786,
+      "step": 18500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0002949439149079605,
+      "loss": 2.8824,
+      "step": 19000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0002948034680998483,
+      "loss": 2.8358,
+      "step": 19500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0002946630212917361,
+      "loss": 2.8757,
+      "step": 20000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0002945225744836239,
+      "loss": 2.8083,
+      "step": 20500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00029438212767551164,
+      "loss": 2.7675,
+      "step": 21000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00029424168086739945,
+      "loss": 2.8155,
+      "step": 21500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00029410123405928726,
+      "loss": 2.7924,
+      "step": 22000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00029396078725117507,
+      "loss": 2.7935,
+      "step": 22500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002938203404430628,
+      "loss": 2.7332,
+      "step": 23000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00029367989363495063,
+      "loss": 2.7284,
+      "step": 23500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002935394468268384,
+      "loss": 2.757,
+      "step": 24000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002933990000187262,
+      "loss": 2.7514,
+      "step": 24500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000293258553210614,
+      "loss": 2.7135,
+      "step": 25000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0002931181064025018,
+      "loss": 2.7346,
+      "step": 25500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0002929776595943896,
+      "loss": 2.7471,
+      "step": 26000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00029283721278627737,
+      "loss": 2.7027,
+      "step": 26500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0002926967659781652,
+      "loss": 2.7163,
+      "step": 27000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00029255631917005293,
+      "loss": 2.7118,
+      "step": 27500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00029241587236194074,
+      "loss": 2.668,
+      "step": 28000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00029227542555382855,
+      "loss": 2.6531,
+      "step": 28500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00029213497874571636,
+      "loss": 2.6297,
+      "step": 29000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00029199453193760416,
+      "loss": 2.6589,
+      "step": 29500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0002918540851294919,
+      "loss": 2.6096,
+      "step": 30000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0002917136383213797,
+      "loss": 2.643,
+      "step": 30500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0002915731915132675,
+      "loss": 2.621,
+      "step": 31000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0002914327447051553,
+      "loss": 2.6194,
+      "step": 31500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0002912922978970431,
+      "loss": 2.649,
+      "step": 32000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0002911518510889309,
+      "loss": 2.6221,
+      "step": 32500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0002910114042808187,
+      "loss": 2.5794,
+      "step": 33000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00029087095747270646,
+      "loss": 2.6582,
+      "step": 33500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0002907305106645943,
+      "loss": 2.6264,
+      "step": 34000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000290590063856482,
+      "loss": 2.6119,
+      "step": 34500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00029044961704836983,
+      "loss": 2.5911,
+      "step": 35000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00029030917024025764,
+      "loss": 2.5688,
+      "step": 35500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00029016872343214545,
+      "loss": 2.5316,
+      "step": 36000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00029002827662403326,
+      "loss": 2.5186,
+      "step": 36500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.000289887829815921,
+      "loss": 2.4959,
+      "step": 37000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.0002897473830078088,
+      "loss": 2.5466,
+      "step": 37500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.0002896069361996966,
+      "loss": 2.5292,
+      "step": 38000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0002894664893915844,
+      "loss": 2.5283,
+      "step": 38500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0002893260425834722,
+      "loss": 2.5334,
+      "step": 39000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00028918559577536,
+      "loss": 2.5344,
+      "step": 39500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0002890451489672478,
+      "loss": 2.5227,
+      "step": 40000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00028890470215913556,
+      "loss": 2.5043,
+      "step": 40500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00028876425535102337,
+      "loss": 2.4737,
+      "step": 41000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0002886238085429112,
+      "loss": 2.5168,
+      "step": 41500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00028848336173479893,
+      "loss": 2.5557,
+      "step": 42000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00028834291492668674,
+      "loss": 2.4842,
+      "step": 42500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00028820246811857455,
+      "loss": 2.5212,
+      "step": 43000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00028806202131046235,
+      "loss": 2.5265,
+      "step": 43500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0002879215745023501,
+      "loss": 2.5013,
+      "step": 44000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0002877811276942379,
+      "loss": 2.5131,
+      "step": 44500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0002876406808861257,
+      "loss": 2.4899,
+      "step": 45000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0002875002340780135,
+      "loss": 2.5271,
+      "step": 45500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0002873597872699013,
+      "loss": 2.4909,
+      "step": 46000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.0002872193404617891,
+      "loss": 2.4901,
+      "step": 46500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0002870788936536769,
+      "loss": 2.4997,
+      "step": 47000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00028693844684556466,
+      "loss": 2.4705,
+      "step": 47500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00028679800003745246,
+      "loss": 2.453,
+      "step": 48000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00028665755322934027,
+      "loss": 2.4742,
+      "step": 48500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.000286517106421228,
+      "loss": 2.4825,
+      "step": 49000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00028637665961311583,
+      "loss": 2.4236,
+      "step": 49500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00028623621280500364,
+      "loss": 2.4295,
+      "step": 50000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00028609576599689145,
+      "loss": 2.4636,
+      "step": 50500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0002859553191887792,
+      "loss": 2.4772,
+      "step": 51000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000285814872380667,
+      "loss": 2.4243,
+      "step": 51500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0002856744255725548,
+      "loss": 2.4275,
+      "step": 52000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0002855339787644426,
+      "loss": 2.4304,
+      "step": 52500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0002853935319563304,
+      "loss": 2.4375,
+      "step": 53000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0002852530851482182,
+      "loss": 2.4304,
+      "step": 53500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00028511263834010594,
+      "loss": 2.4101,
+      "step": 54000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00028497219153199375,
+      "loss": 2.4256,
+      "step": 54500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00028483174472388156,
+      "loss": 2.4022,
+      "step": 55000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00028469129791576937,
+      "loss": 2.405,
+      "step": 55500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0002845508511076571,
+      "loss": 2.4053,
+      "step": 56000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00028441040429954493,
+      "loss": 2.4268,
+      "step": 56500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0002842699574914327,
+      "loss": 2.3901,
+      "step": 57000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0002841295106833205,
+      "loss": 2.4176,
+      "step": 57500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0002839890638752083,
+      "loss": 2.42,
+      "step": 58000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0002838486170670961,
+      "loss": 2.3913,
+      "step": 58500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0002837081702589839,
+      "loss": 2.382,
+      "step": 59000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00028356772345087167,
+      "loss": 2.3878,
+      "step": 59500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.0002834272766427595,
+      "loss": 2.4147,
+      "step": 60000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00028328682983464723,
+      "loss": 2.38,
+      "step": 60500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00028314638302653504,
+      "loss": 2.3794,
+      "step": 61000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00028300593621842285,
+      "loss": 2.3822,
+      "step": 61500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00028286548941031065,
+      "loss": 2.3941,
+      "step": 62000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00028272504260219846,
+      "loss": 2.3859,
+      "step": 62500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0002825845957940862,
+      "loss": 2.387,
+      "step": 63000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.000282444148985974,
+      "loss": 2.3683,
+      "step": 63500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0002823037021778618,
+      "loss": 2.3984,
+      "step": 64000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0002821632553697496,
+      "loss": 2.3278,
+      "step": 64500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.0002820228085616374,
+      "loss": 2.3648,
+      "step": 65000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0002818823617535252,
+      "loss": 2.3494,
+      "step": 65500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.000281741914945413,
+      "loss": 2.3724,
+      "step": 66000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00028160146813730076,
+      "loss": 2.3507,
+      "step": 66500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00028146102132918857,
+      "loss": 2.3449,
+      "step": 67000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0002813205745210763,
+      "loss": 2.371,
+      "step": 67500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00028118012771296413,
+      "loss": 2.3572,
+      "step": 68000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00028103968090485194,
+      "loss": 2.3779,
+      "step": 68500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00028089923409673975,
+      "loss": 2.3376,
+      "step": 69000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00028075878728862756,
+      "loss": 2.3384,
+      "step": 69500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0002806183404805153,
+      "loss": 2.3622,
+      "step": 70000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0002804778936724031,
+      "loss": 2.3338,
+      "step": 70500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0002803374468642909,
+      "loss": 2.3579,
+      "step": 71000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002801970000561787,
+      "loss": 2.3414,
+      "step": 71500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0002800565532480665,
+      "loss": 2.2625,
+      "step": 72000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0002799161064399543,
+      "loss": 2.2404,
+      "step": 72500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.0002797756596318421,
+      "loss": 2.2747,
+      "step": 73000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00027963521282372986,
+      "loss": 2.2791,
+      "step": 73500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00027949476601561767,
+      "loss": 2.3042,
+      "step": 74000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0002793543192075054,
+      "loss": 2.2652,
+      "step": 74500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00027921387239939323,
+      "loss": 2.236,
+      "step": 75000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00027907342559128104,
+      "loss": 2.2508,
+      "step": 75500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00027893297878316885,
+      "loss": 2.2145,
+      "step": 76000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00027879253197505665,
+      "loss": 2.2401,
+      "step": 76500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002786520851669444,
+      "loss": 2.2362,
+      "step": 77000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0002785116383588322,
+      "loss": 2.237,
+      "step": 77500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00027837119155071997,
+      "loss": 2.2818,
+      "step": 78000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0002782307447426078,
+      "loss": 2.2515,
+      "step": 78500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0002780902979344956,
+      "loss": 2.2559,
+      "step": 79000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002779498511263834,
+      "loss": 2.2788,
+      "step": 79500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0002778094043182712,
+      "loss": 2.2364,
+      "step": 80000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00027766895751015896,
+      "loss": 2.2632,
+      "step": 80500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00027752851070204676,
+      "loss": 2.2095,
+      "step": 81000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002773880638939345,
+      "loss": 2.256,
+      "step": 81500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.0002772476170858223,
+      "loss": 2.2417,
+      "step": 82000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00027710717027771013,
+      "loss": 2.2227,
+      "step": 82500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00027696672346959794,
+      "loss": 2.2372,
+      "step": 83000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00027682627666148575,
+      "loss": 2.2767,
+      "step": 83500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002766858298533735,
+      "loss": 2.2501,
+      "step": 84000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0002765453830452613,
+      "loss": 2.2325,
+      "step": 84500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00027640493623714906,
+      "loss": 2.2462,
+      "step": 85000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00027626448942903687,
+      "loss": 2.2773,
+      "step": 85500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.0002761240426209247,
+      "loss": 2.2363,
+      "step": 86000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.0002759835958128125,
+      "loss": 2.2692,
+      "step": 86500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002758431490047003,
+      "loss": 2.2882,
+      "step": 87000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00027570270219658805,
+      "loss": 2.2471,
+      "step": 87500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00027556225538847586,
+      "loss": 2.2483,
+      "step": 88000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002754218085803636,
+      "loss": 2.2677,
+      "step": 88500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002752813617722514,
+      "loss": 2.2497,
+      "step": 89000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00027514091496413923,
+      "loss": 2.2268,
+      "step": 89500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00027500046815602704,
+      "loss": 2.2396,
+      "step": 90000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002748600213479148,
+      "loss": 2.2197,
+      "step": 90500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.0002747195745398026,
+      "loss": 2.216,
+      "step": 91000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002745791277316904,
+      "loss": 2.2479,
+      "step": 91500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0002744386809235782,
+      "loss": 2.238,
+      "step": 92000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00027429823411546597,
+      "loss": 2.2426,
+      "step": 92500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002741577873073538,
+      "loss": 2.2297,
+      "step": 93000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00027401734049924153,
+      "loss": 2.183,
+      "step": 93500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00027387689369112934,
+      "loss": 2.2415,
+      "step": 94000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00027373644688301715,
+      "loss": 2.2219,
+      "step": 94500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00027359600007490495,
+      "loss": 2.2549,
+      "step": 95000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00027345555326679276,
+      "loss": 2.2516,
+      "step": 95500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0002733151064586805,
+      "loss": 2.2183,
+      "step": 96000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0002731746596505683,
+      "loss": 2.1943,
+      "step": 96500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002730342128424561,
+      "loss": 2.2435,
+      "step": 97000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0002728937660343439,
+      "loss": 2.2291,
+      "step": 97500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0002727533192262317,
+      "loss": 2.2371,
+      "step": 98000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0002726128724181195,
+      "loss": 2.235,
+      "step": 98500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002724724256100073,
+      "loss": 2.2207,
+      "step": 99000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00027233197880189506,
+      "loss": 2.26,
+      "step": 99500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00027219153199378287,
+      "loss": 2.2456,
+      "step": 100000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002720510851856706,
+      "loss": 2.251,
+      "step": 100500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00027191063837755843,
+      "loss": 2.2112,
+      "step": 101000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00027177019156944624,
+      "loss": 2.2283,
+      "step": 101500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00027162974476133405,
+      "loss": 2.2472,
+      "step": 102000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00027148929795322186,
+      "loss": 2.2258,
+      "step": 102500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0002713488511451096,
+      "loss": 2.225,
+      "step": 103000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002712084043369974,
+      "loss": 2.2183,
+      "step": 103500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.0002710679575288852,
+      "loss": 2.2316,
+      "step": 104000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.000270927510720773,
+      "loss": 2.2095,
+      "step": 104500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002707870639126608,
+      "loss": 2.2091,
+      "step": 105000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002706466171045486,
+      "loss": 2.2138,
+      "step": 105500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0002705061702964364,
+      "loss": 2.2082,
+      "step": 106000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00027036572348832416,
+      "loss": 2.2378,
+      "step": 106500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00027022527668021197,
+      "loss": 2.211,
+      "step": 107000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0002700848298720997,
+      "loss": 2.1808,
+      "step": 107500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00026994438306398753,
+      "loss": 2.1297,
+      "step": 108000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00026980393625587534,
+      "loss": 2.1099,
+      "step": 108500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00026966348944776314,
+      "loss": 2.1498,
+      "step": 109000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00026952304263965095,
+      "loss": 2.1412,
+      "step": 109500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0002693825958315387,
+      "loss": 2.1386,
+      "step": 110000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0002692421490234265,
+      "loss": 2.1541,
+      "step": 110500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00026910170221531427,
+      "loss": 2.1623,
+      "step": 111000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0002689612554072021,
+      "loss": 2.1315,
+      "step": 111500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0002688208085990899,
+      "loss": 2.1556,
+      "step": 112000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0002686803617909777,
+      "loss": 2.1764,
+      "step": 112500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0002685399149828655,
+      "loss": 2.1254,
+      "step": 113000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00026839946817475325,
+      "loss": 2.1201,
+      "step": 113500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00026825902136664106,
+      "loss": 2.1488,
+      "step": 114000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.0002681185745585288,
+      "loss": 2.1311,
+      "step": 114500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0002679781277504166,
+      "loss": 2.1196,
+      "step": 115000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00026783768094230443,
+      "loss": 2.1513,
+      "step": 115500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00026769723413419224,
+      "loss": 2.1543,
+      "step": 116000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00026755678732608005,
+      "loss": 2.1223,
+      "step": 116500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0002674163405179678,
+      "loss": 2.1536,
+      "step": 117000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.0002672758937098556,
+      "loss": 2.1219,
+      "step": 117500
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00026713544690174336,
+      "loss": 2.1435,
+      "step": 118000
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00026699500009363117,
+      "loss": 2.1299,
+      "step": 118500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.000266854553285519,
+      "loss": 2.1715,
+      "step": 119000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0002667141064774068,
+      "loss": 2.1372,
+      "step": 119500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0002665736596692946,
+      "loss": 2.1533,
+      "step": 120000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00026643321286118235,
+      "loss": 2.1325,
+      "step": 120500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00026629276605307016,
+      "loss": 2.1112,
+      "step": 121000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0002661523192449579,
+      "loss": 2.1307,
+      "step": 121500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0002660118724368457,
+      "loss": 2.1162,
+      "step": 122000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00026587142562873353,
+      "loss": 2.1148,
+      "step": 122500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00026573097882062134,
+      "loss": 2.1122,
+      "step": 123000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0002655905320125091,
+      "loss": 2.1272,
+      "step": 123500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0002654500852043969,
+      "loss": 2.1124,
+      "step": 124000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0002653096383962847,
+      "loss": 2.1317,
+      "step": 124500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00026516919158817246,
+      "loss": 2.0704,
+      "step": 125000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00026502874478006027,
+      "loss": 2.1439,
+      "step": 125500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.0002648882979719481,
+      "loss": 2.1099,
+      "step": 126000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00026474785116383583,
+      "loss": 2.1256,
+      "step": 126500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00026460740435572364,
+      "loss": 2.1312,
+      "step": 127000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00026446695754761145,
+      "loss": 2.1406,
+      "step": 127500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00026432651073949925,
+      "loss": 2.0957,
+      "step": 128000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.000264186063931387,
+      "loss": 2.1054,
+      "step": 128500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0002640456171232748,
+      "loss": 2.1584,
+      "step": 129000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0002639051703151626,
+      "loss": 2.119,
+      "step": 129500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0002637647235070504,
+      "loss": 2.1035,
+      "step": 130000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0002636242766989382,
+      "loss": 2.1215,
+      "step": 130500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.000263483829890826,
+      "loss": 2.1211,
+      "step": 131000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0002633433830827138,
+      "loss": 2.1282,
+      "step": 131500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00026320293627460155,
+      "loss": 2.1391,
+      "step": 132000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00026306248946648936,
+      "loss": 2.1249,
+      "step": 132500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00026292204265837717,
+      "loss": 2.1576,
+      "step": 133000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0002627815958502649,
+      "loss": 2.1529,
+      "step": 133500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00026264114904215273,
+      "loss": 2.119,
+      "step": 134000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00026250070223404054,
+      "loss": 2.1246,
+      "step": 134500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00026236025542592835,
+      "loss": 2.1074,
+      "step": 135000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0002622198086178161,
+      "loss": 2.1399,
+      "step": 135500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.0002620793618097039,
+      "loss": 2.1434,
+      "step": 136000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0002619389150015917,
+      "loss": 2.1012,
+      "step": 136500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00026179846819347947,
+      "loss": 2.103,
+      "step": 137000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0002616580213853673,
+      "loss": 2.1803,
+      "step": 137500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0002615175745772551,
+      "loss": 2.0987,
+      "step": 138000
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0002613771277691429,
+      "loss": 2.1171,
+      "step": 138500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00026123668096103065,
+      "loss": 2.1175,
+      "step": 139000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00026109623415291846,
+      "loss": 2.1562,
+      "step": 139500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00026095578734480627,
+      "loss": 2.1321,
+      "step": 140000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.000260815340536694,
+      "loss": 2.1195,
+      "step": 140500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00026067489372858183,
+      "loss": 2.1,
+      "step": 141000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00026053444692046964,
+      "loss": 2.1052,
+      "step": 141500
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00026039400011235744,
+      "loss": 2.0947,
+      "step": 142000
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00026025355330424525,
+      "loss": 2.1094,
+      "step": 142500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000260113106496133,
+      "loss": 2.0432,
+      "step": 143000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0002599726596880208,
+      "loss": 2.0555,
+      "step": 143500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00025983221287990857,
+      "loss": 2.007,
+      "step": 144000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0002596917660717964,
+      "loss": 2.0206,
+      "step": 144500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0002595513192636842,
+      "loss": 2.0597,
+      "step": 145000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000259410872455572,
+      "loss": 2.0644,
+      "step": 145500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0002592704256474598,
+      "loss": 2.0271,
+      "step": 146000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00025912997883934755,
+      "loss": 2.0555,
+      "step": 146500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00025898953203123536,
+      "loss": 2.0494,
+      "step": 147000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0002588490852231231,
+      "loss": 2.0369,
+      "step": 147500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0002587086384150109,
+      "loss": 2.0388,
+      "step": 148000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00025856819160689873,
+      "loss": 2.0225,
+      "step": 148500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00025842774479878654,
+      "loss": 2.0506,
+      "step": 149000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00025828729799067435,
+      "loss": 2.028,
+      "step": 149500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0002581468511825621,
+      "loss": 2.0405,
+      "step": 150000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0002580064043744499,
+      "loss": 2.0818,
+      "step": 150500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00025786595756633766,
+      "loss": 2.0557,
+      "step": 151000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00025772551075822547,
+      "loss": 2.0426,
+      "step": 151500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0002575850639501133,
+      "loss": 2.0414,
+      "step": 152000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0002574446171420011,
+      "loss": 2.0591,
+      "step": 152500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.0002573041703338889,
+      "loss": 2.0769,
+      "step": 153000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00025716372352577665,
+      "loss": 2.087,
+      "step": 153500
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00025702327671766446,
+      "loss": 2.0396,
+      "step": 154000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.0002568828299095522,
+      "loss": 2.0451,
+      "step": 154500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00025674238310144,
+      "loss": 2.0115,
+      "step": 155000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0002566019362933278,
+      "loss": 2.0424,
+      "step": 155500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00025646148948521563,
+      "loss": 2.0356,
+      "step": 156000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00025632104267710344,
+      "loss": 2.0349,
+      "step": 156500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0002561805958689912,
+      "loss": 2.0458,
+      "step": 157000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.000256040149060879,
+      "loss": 2.0379,
+      "step": 157500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00025589970225276676,
+      "loss": 2.0478,
+      "step": 158000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00025575925544465457,
+      "loss": 2.042,
+      "step": 158500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0002556188086365424,
+      "loss": 2.0609,
+      "step": 159000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0002554783618284302,
+      "loss": 2.0513,
+      "step": 159500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00025533791502031794,
+      "loss": 2.0489,
+      "step": 160000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00025519746821220574,
+      "loss": 2.049,
+      "step": 160500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00025505702140409355,
+      "loss": 2.0444,
+      "step": 161000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.0002549165745959813,
+      "loss": 2.0658,
+      "step": 161500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0002547761277878691,
+      "loss": 2.0258,
+      "step": 162000
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0002546356809797569,
+      "loss": 2.0551,
+      "step": 162500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.0002544952341716447,
+      "loss": 2.0555,
+      "step": 163000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0002543547873635325,
+      "loss": 2.0466,
+      "step": 163500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0002542143405554203,
+      "loss": 2.0336,
+      "step": 164000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0002540738937473081,
+      "loss": 2.0412,
+      "step": 164500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00025393344693919585,
+      "loss": 2.0094,
+      "step": 165000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00025379300013108366,
+      "loss": 2.0189,
+      "step": 165500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.0002536525533229714,
+      "loss": 2.0634,
+      "step": 166000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0002535121065148592,
+      "loss": 2.0543,
+      "step": 166500
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00025337165970674703,
+      "loss": 2.0689,
+      "step": 167000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00025323121289863484,
+      "loss": 2.0068,
+      "step": 167500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00025309076609052265,
+      "loss": 2.0627,
+      "step": 168000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0002529503192824104,
+      "loss": 2.0485,
+      "step": 168500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0002528098724742982,
+      "loss": 2.0589,
+      "step": 169000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000252669425666186,
+      "loss": 2.0407,
+      "step": 169500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00025252897885807377,
+      "loss": 2.0687,
+      "step": 170000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.0002523885320499616,
+      "loss": 2.0451,
+      "step": 170500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.0002522480852418494,
+      "loss": 2.0376,
+      "step": 171000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0002521076384337372,
+      "loss": 2.0302,
+      "step": 171500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00025196719162562495,
+      "loss": 2.0364,
+      "step": 172000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00025182674481751276,
+      "loss": 2.0172,
+      "step": 172500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00025168629800940057,
+      "loss": 2.0511,
+      "step": 173000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0002515458512012883,
+      "loss": 2.0463,
+      "step": 173500
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00025140540439317613,
+      "loss": 2.0661,
+      "step": 174000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00025126495758506394,
+      "loss": 2.0165,
+      "step": 174500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00025112451077695174,
+      "loss": 2.031,
+      "step": 175000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0002509840639688395,
+      "loss": 2.0353,
+      "step": 175500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0002508436171607273,
+      "loss": 2.0225,
+      "step": 176000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0002507031703526151,
+      "loss": 2.0854,
+      "step": 176500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00025056272354450287,
+      "loss": 2.0487,
+      "step": 177000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0002504222767363907,
+      "loss": 2.0354,
+      "step": 177500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0002502818299282785,
+      "loss": 2.04,
+      "step": 178000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0002501413831201663,
+      "loss": 2.0023,
+      "step": 178500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.00025000093631205404,
+      "loss": 1.9267,
+      "step": 179000
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00024986048950394185,
+      "loss": 1.9619,
+      "step": 179500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00024972004269582966,
+      "loss": 1.9616,
+      "step": 180000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0002495795958877174,
+      "loss": 1.9733,
+      "step": 180500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0002494391490796052,
+      "loss": 1.9445,
+      "step": 181000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00024929870227149303,
+      "loss": 1.9845,
+      "step": 181500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.00024915825546338084,
+      "loss": 1.9633,
+      "step": 182000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0002490178086552686,
+      "loss": 1.9904,
+      "step": 182500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0002488773618471564,
+      "loss": 1.9854,
+      "step": 183000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0002487369150390442,
+      "loss": 1.952,
+      "step": 183500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00024859646823093196,
+      "loss": 2.003,
+      "step": 184000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00024845602142281977,
+      "loss": 1.9374,
+      "step": 184500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0002483155746147076,
+      "loss": 1.9553,
+      "step": 185000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0002481751278065954,
+      "loss": 1.9394,
+      "step": 185500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00024803468099848314,
+      "loss": 1.9391,
+      "step": 186000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00024789423419037095,
+      "loss": 1.9604,
+      "step": 186500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.00024775378738225876,
+      "loss": 1.9474,
+      "step": 187000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0002476133405741465,
+      "loss": 1.9424,
+      "step": 187500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0002474728937660343,
+      "loss": 1.9742,
+      "step": 188000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0002473324469579221,
+      "loss": 1.9361,
+      "step": 188500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.00024719200014980993,
+      "loss": 1.9689,
+      "step": 189000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0002470515533416977,
+      "loss": 1.9674,
+      "step": 189500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0002469111065335855,
+      "loss": 1.9602,
+      "step": 190000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0002467706597254733,
+      "loss": 1.9455,
+      "step": 190500
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00024663021291736106,
+      "loss": 1.9798,
+      "step": 191000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.00024648976610924887,
+      "loss": 1.9809,
+      "step": 191500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0002463493193011367,
+      "loss": 1.9677,
+      "step": 192000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0002462088724930245,
+      "loss": 1.981,
+      "step": 192500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00024606842568491224,
+      "loss": 1.9866,
+      "step": 193000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.00024592797887680004,
+      "loss": 1.9806,
+      "step": 193500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00024578753206868785,
+      "loss": 1.9721,
+      "step": 194000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0002456470852605756,
+      "loss": 1.9762,
+      "step": 194500
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.0002455066384524634,
+      "loss": 2.0143,
+      "step": 195000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0002453661916443512,
+      "loss": 1.9659,
+      "step": 195500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.000245225744836239,
+      "loss": 1.9384,
+      "step": 196000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0002450852980281268,
+      "loss": 1.9937,
+      "step": 196500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0002449448512200146,
+      "loss": 1.9358,
+      "step": 197000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0002448044044119024,
+      "loss": 1.9805,
+      "step": 197500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.00024466395760379015,
+      "loss": 1.952,
+      "step": 198000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.00024452351079567796,
+      "loss": 2.03,
+      "step": 198500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0002443830639875657,
+      "loss": 2.0039,
+      "step": 199000
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0002442426171794535,
+      "loss": 1.9566,
+      "step": 199500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.00024410217037134136,
+      "loss": 1.9538,
+      "step": 200000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00024396172356322914,
+      "loss": 1.9754,
+      "step": 200500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00024382127675511695,
+      "loss": 1.977,
+      "step": 201000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0002436808299470047,
+      "loss": 1.9565,
+      "step": 201500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0002435403831388925,
+      "loss": 1.9479,
+      "step": 202000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0002433999363307803,
+      "loss": 2.0019,
+      "step": 202500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0002432594895226681,
+      "loss": 1.9587,
+      "step": 203000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0002431190427145559,
+      "loss": 1.9554,
+      "step": 203500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0002429785959064437,
+      "loss": 1.9694,
+      "step": 204000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00024283814909833147,
+      "loss": 1.9901,
+      "step": 204500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.00024269770229021925,
+      "loss": 1.9503,
+      "step": 205000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.00024255725548210706,
+      "loss": 1.9554,
+      "step": 205500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00024241680867399484,
+      "loss": 1.9525,
+      "step": 206000
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00024227636186588265,
+      "loss": 1.981,
+      "step": 206500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00024213591505777045,
+      "loss": 1.9451,
+      "step": 207000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0002419954682496582,
+      "loss": 1.9553,
+      "step": 207500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.00024185502144154602,
+      "loss": 1.9474,
+      "step": 208000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0002417145746334338,
+      "loss": 1.9642,
+      "step": 208500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0002415741278253216,
+      "loss": 1.9603,
+      "step": 209000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00024143368101720939,
+      "loss": 1.9093,
+      "step": 209500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0002412932342090972,
+      "loss": 1.9364,
+      "step": 210000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.000241152787400985,
+      "loss": 1.9891,
+      "step": 210500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00024101234059287276,
+      "loss": 1.9717,
+      "step": 211000
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00024087189378476056,
+      "loss": 1.9513,
+      "step": 211500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00024073144697664834,
+      "loss": 1.9571,
+      "step": 212000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00024059100016853615,
+      "loss": 1.9442,
+      "step": 212500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.00024045055336042393,
+      "loss": 1.9624,
+      "step": 213000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00024031010655231174,
+      "loss": 1.9352,
+      "step": 213500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00024016965974419955,
+      "loss": 1.9092,
+      "step": 214000
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0002400292129360873,
+      "loss": 1.8456,
+      "step": 214500
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0002398887661279751,
+      "loss": 1.8846,
+      "step": 215000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.0002397483193198629,
+      "loss": 1.8422,
+      "step": 215500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0002396078725117507,
+      "loss": 1.8811,
+      "step": 216000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00023946742570363848,
+      "loss": 1.8762,
+      "step": 216500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0002393269788955263,
+      "loss": 1.8813,
+      "step": 217000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0002391865320874141,
+      "loss": 1.8543,
+      "step": 217500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00023904608527930185,
+      "loss": 1.8821,
+      "step": 218000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00023890563847118966,
+      "loss": 1.8509,
+      "step": 218500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00023876519166307744,
+      "loss": 1.8799,
+      "step": 219000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.00023862474485496525,
+      "loss": 1.8657,
+      "step": 219500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00023848429804685306,
+      "loss": 1.8988,
+      "step": 220000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00023834385123874084,
+      "loss": 1.8873,
+      "step": 220500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.00023820340443062862,
+      "loss": 1.8665,
+      "step": 221000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0002380629576225164,
+      "loss": 1.8889,
+      "step": 221500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.0002379225108144042,
+      "loss": 1.8704,
+      "step": 222000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.000237782064006292,
+      "loss": 1.8989,
+      "step": 222500
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0002376416171981798,
+      "loss": 1.8646,
+      "step": 223000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0002375011703900676,
+      "loss": 1.8908,
+      "step": 223500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00023736072358195536,
+      "loss": 1.9154,
+      "step": 224000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.00023722027677384317,
+      "loss": 1.8759,
+      "step": 224500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00023707982996573095,
+      "loss": 1.8696,
+      "step": 225000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00023693938315761875,
+      "loss": 1.8735,
+      "step": 225500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.00023679893634950654,
+      "loss": 1.8781,
+      "step": 226000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00023665848954139434,
+      "loss": 1.8739,
+      "step": 226500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00023651804273328215,
+      "loss": 1.8942,
+      "step": 227000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0002363775959251699,
+      "loss": 1.8897,
+      "step": 227500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0002362371491170577,
+      "loss": 1.8956,
+      "step": 228000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0002360967023089455,
+      "loss": 1.8849,
+      "step": 228500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0002359562555008333,
+      "loss": 1.8823,
+      "step": 229000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00023581580869272108,
+      "loss": 1.8674,
+      "step": 229500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0002356753618846089,
+      "loss": 1.896,
+      "step": 230000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0002355349150764967,
+      "loss": 1.9278,
+      "step": 230500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.00023539446826838445,
+      "loss": 1.8707,
+      "step": 231000
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00023525402146027226,
+      "loss": 1.9087,
+      "step": 231500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00023511357465216004,
+      "loss": 1.8762,
+      "step": 232000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.00023497312784404785,
+      "loss": 1.9027,
+      "step": 232500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00023483268103593563,
+      "loss": 1.938,
+      "step": 233000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00023469223422782344,
+      "loss": 1.9021,
+      "step": 233500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.00023455178741971125,
+      "loss": 1.9112,
+      "step": 234000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.000234411340611599,
+      "loss": 1.904,
+      "step": 234500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0002342708938034868,
+      "loss": 1.9256,
+      "step": 235000
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0002341304469953746,
+      "loss": 1.9016,
+      "step": 235500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0002339900001872624,
+      "loss": 1.8928,
+      "step": 236000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00023384955337915018,
+      "loss": 1.8887,
+      "step": 236500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00023370910657103799,
+      "loss": 1.8664,
+      "step": 237000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.00023356865976292577,
+      "loss": 1.9367,
+      "step": 237500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00023342821295481355,
+      "loss": 1.9405,
+      "step": 238000
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00023328776614670136,
+      "loss": 1.8594,
+      "step": 238500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00023314731933858914,
+      "loss": 1.895,
+      "step": 239000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00023300687253047695,
+      "loss": 1.9265,
+      "step": 239500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.00023286642572236473,
+      "loss": 1.8762,
+      "step": 240000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0002327259789142525,
+      "loss": 1.9104,
+      "step": 240500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00023258553210614031,
+      "loss": 1.8946,
+      "step": 241000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0002324450852980281,
+      "loss": 1.9149,
+      "step": 241500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0002323046384899159,
+      "loss": 1.9136,
+      "step": 242000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00023216419168180368,
+      "loss": 1.88,
+      "step": 242500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0002320237448736915,
+      "loss": 1.8988,
+      "step": 243000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00023188329806557925,
+      "loss": 1.9184,
+      "step": 243500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00023174285125746705,
+      "loss": 1.8672,
+      "step": 244000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00023160240444935486,
+      "loss": 1.9284,
+      "step": 244500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.00023146195764124264,
+      "loss": 1.9108,
+      "step": 245000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.00023132151083313045,
+      "loss": 1.8983,
+      "step": 245500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00023118106402501823,
+      "loss": 1.8783,
+      "step": 246000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00023104061721690604,
+      "loss": 1.9412,
+      "step": 246500
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.00023090017040879385,
+      "loss": 1.9064,
+      "step": 247000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0002307597236006816,
+      "loss": 1.9071,
+      "step": 247500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0002306192767925694,
+      "loss": 1.9568,
+      "step": 248000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0002304788299844572,
+      "loss": 1.9261,
+      "step": 248500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.000230338383176345,
+      "loss": 1.8863,
+      "step": 249000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00023019793636823278,
+      "loss": 1.8946,
+      "step": 249500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0002300574895601206,
+      "loss": 1.7976,
+      "step": 250000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0002299170427520084,
+      "loss": 1.8215,
+      "step": 250500
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00022977659594389615,
+      "loss": 1.8041,
+      "step": 251000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00022963614913578396,
+      "loss": 1.8455,
+      "step": 251500
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.00022949570232767174,
+      "loss": 1.83,
+      "step": 252000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00022935525551955955,
+      "loss": 1.8258,
+      "step": 252500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00022921480871144733,
+      "loss": 1.7908,
+      "step": 253000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.00022907436190333514,
+      "loss": 1.8099,
+      "step": 253500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00022893391509522294,
+      "loss": 1.8051,
+      "step": 254000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0002287934682871107,
+      "loss": 1.8227,
+      "step": 254500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0002286530214789985,
+      "loss": 1.8217,
+      "step": 255000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0002285125746708863,
+      "loss": 1.818,
+      "step": 255500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0002283721278627741,
+      "loss": 1.8631,
+      "step": 256000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.00022823168105466188,
+      "loss": 1.8189,
+      "step": 256500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00022809123424654968,
+      "loss": 1.8172,
+      "step": 257000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.00022795078743843746,
+      "loss": 1.8375,
+      "step": 257500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00022781034063032525,
+      "loss": 1.841,
+      "step": 258000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.00022766989382221305,
+      "loss": 1.8825,
+      "step": 258500
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00022752944701410083,
+      "loss": 1.8211,
+      "step": 259000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.00022738900020598864,
+      "loss": 1.835,
+      "step": 259500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0002272485533978764,
+      "loss": 1.809,
+      "step": 260000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0002271081065897642,
+      "loss": 1.833,
+      "step": 260500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.000226967659781652,
+      "loss": 1.8221,
+      "step": 261000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0002268272129735398,
+      "loss": 1.842,
+      "step": 261500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0002266867661654276,
+      "loss": 1.8132,
+      "step": 262000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.00022654631935731538,
+      "loss": 1.8774,
+      "step": 262500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0002264058725492032,
+      "loss": 1.8435,
+      "step": 263000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.00022626542574109094,
+      "loss": 1.8275,
+      "step": 263500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.00022612497893297875,
+      "loss": 1.8354,
+      "step": 264000
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00022598453212486656,
+      "loss": 1.8332,
+      "step": 264500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00022584408531675434,
+      "loss": 1.8744,
+      "step": 265000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00022570363850864215,
+      "loss": 1.8174,
+      "step": 265500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00022556319170052993,
+      "loss": 1.819,
+      "step": 266000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.00022542274489241774,
+      "loss": 1.8235,
+      "step": 266500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0002252822980843055,
+      "loss": 1.8256,
+      "step": 267000
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0002251418512761933,
+      "loss": 1.8291,
+      "step": 267500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0002250014044680811,
+      "loss": 1.8595,
+      "step": 268000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0002248609576599689,
+      "loss": 1.839,
+      "step": 268500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0002247205108518567,
+      "loss": 1.8658,
+      "step": 269000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.00022458006404374448,
+      "loss": 1.8508,
+      "step": 269500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.00022443961723563229,
+      "loss": 1.8332,
+      "step": 270000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0002242991704275201,
+      "loss": 1.8551,
+      "step": 270500
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.00022415872361940785,
+      "loss": 1.8145,
+      "step": 271000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.00022401827681129566,
+      "loss": 1.8381,
+      "step": 271500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.00022387783000318344,
+      "loss": 1.8425,
+      "step": 272000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.00022373738319507124,
+      "loss": 1.8686,
+      "step": 272500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.00022359693638695903,
+      "loss": 1.8304,
+      "step": 273000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.00022345648957884683,
+      "loss": 1.8242,
+      "step": 273500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00022331604277073461,
+      "loss": 1.8065,
+      "step": 274000
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0002231755959626224,
+      "loss": 1.847,
+      "step": 274500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0002230351491545102,
+      "loss": 1.8359,
+      "step": 275000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.00022289470234639798,
+      "loss": 1.8148,
+      "step": 275500
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0002227542555382858,
+      "loss": 1.8743,
+      "step": 276000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.00022261380873017357,
+      "loss": 1.8471,
+      "step": 276500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.00022247336192206135,
+      "loss": 1.8661,
+      "step": 277000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.00022233291511394916,
+      "loss": 1.8514,
+      "step": 277500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.00022219246830583694,
+      "loss": 1.8565,
+      "step": 278000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.00022205202149772475,
+      "loss": 1.8542,
+      "step": 278500
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.00022191157468961253,
+      "loss": 1.8716,
+      "step": 279000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.00022177112788150034,
+      "loss": 1.8332,
+      "step": 279500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0002216306810733881,
+      "loss": 1.8361,
+      "step": 280000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0002214902342652759,
+      "loss": 1.8131,
+      "step": 280500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0002213497874571637,
+      "loss": 1.8443,
+      "step": 281000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0002212093406490515,
+      "loss": 1.8881,
+      "step": 281500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0002210688938409393,
+      "loss": 1.8587,
+      "step": 282000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.00022092844703282708,
+      "loss": 1.846,
+      "step": 282500
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0002207880002247149,
+      "loss": 1.8216,
+      "step": 283000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.00022064755341660264,
+      "loss": 1.8285,
+      "step": 283500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.00022050710660849045,
+      "loss": 1.8622,
+      "step": 284000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.00022036665980037826,
+      "loss": 1.8458,
+      "step": 284500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.00022022621299226604,
+      "loss": 1.8345,
+      "step": 285000
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.00022008576618415385,
+      "loss": 1.7663,
+      "step": 285500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.00021994531937604163,
+      "loss": 1.7715,
+      "step": 286000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.00021980487256792944,
+      "loss": 1.7414,
+      "step": 286500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.0002196644257598172,
+      "loss": 1.7456,
+      "step": 287000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.000219523978951705,
+      "loss": 1.7446,
+      "step": 287500
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0002193835321435928,
+      "loss": 1.76,
+      "step": 288000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.00021924308533548059,
+      "loss": 1.7493,
+      "step": 288500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0002191026385273684,
+      "loss": 1.761,
+      "step": 289000
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.00021896219171925617,
+      "loss": 1.7456,
+      "step": 289500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.00021882174491114398,
+      "loss": 1.7712,
+      "step": 290000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.00021868129810303174,
+      "loss": 1.7744,
+      "step": 290500
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.00021854085129491954,
+      "loss": 1.7679,
+      "step": 291000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.00021840040448680735,
+      "loss": 1.7527,
+      "step": 291500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.00021825995767869513,
+      "loss": 1.7428,
+      "step": 292000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.00021811951087058294,
+      "loss": 1.7666,
+      "step": 292500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.00021797906406247072,
+      "loss": 1.7579,
+      "step": 293000
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0002178386172543585,
+      "loss": 1.7789,
+      "step": 293500
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.00021769817044624628,
+      "loss": 1.7492,
+      "step": 294000
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0002175577236381341,
+      "loss": 1.7532,
+      "step": 294500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.0002174172768300219,
+      "loss": 1.7642,
+      "step": 295000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.00021727683002190968,
+      "loss": 1.7626,
+      "step": 295500
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0002171363832137975,
+      "loss": 1.7749,
+      "step": 296000
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.00021699593640568524,
+      "loss": 1.7803,
+      "step": 296500
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.00021685548959757305,
+      "loss": 1.7891,
+      "step": 297000
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.00021671504278946086,
+      "loss": 1.7569,
+      "step": 297500
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.00021657459598134864,
+      "loss": 1.7607,
+      "step": 298000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.00021643414917323645,
+      "loss": 1.7525,
+      "step": 298500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.00021629370236512423,
+      "loss": 1.8031,
+      "step": 299000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.00021615325555701204,
+      "loss": 1.768,
+      "step": 299500
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0002160128087488998,
+      "loss": 1.7674,
+      "step": 300000
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0002158723619407876,
+      "loss": 1.7714,
+      "step": 300500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0002157319151326754,
+      "loss": 1.7616,
+      "step": 301000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.0002155914683245632,
+      "loss": 1.8042,
+      "step": 301500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.000215451021516451,
+      "loss": 1.7405,
+      "step": 302000
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.00021531057470833878,
+      "loss": 1.793,
+      "step": 302500
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.00021517012790022658,
+      "loss": 1.7702,
+      "step": 303000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.00021502968109211434,
+      "loss": 1.7444,
+      "step": 303500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.00021488923428400215,
+      "loss": 1.7789,
+      "step": 304000
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.00021474878747588995,
+      "loss": 1.775,
+      "step": 304500
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.00021460834066777774,
+      "loss": 1.7749,
+      "step": 305000
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.00021446789385966554,
+      "loss": 1.7767,
+      "step": 305500
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.00021432744705155332,
+      "loss": 1.7808,
+      "step": 306000
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.00021418700024344113,
+      "loss": 1.7688,
+      "step": 306500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.00021404655343532889,
+      "loss": 1.7917,
+      "step": 307000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0002139061066272167,
+      "loss": 1.7924,
+      "step": 307500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0002137656598191045,
+      "loss": 1.7847,
+      "step": 308000
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.00021362521301099228,
+      "loss": 1.7892,
+      "step": 308500
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0002134847662028801,
+      "loss": 1.7915,
+      "step": 309000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.00021334431939476787,
+      "loss": 1.7895,
+      "step": 309500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.00021320387258665565,
+      "loss": 1.7896,
+      "step": 310000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.00021306342577854343,
+      "loss": 1.785,
+      "step": 310500
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.00021292297897043124,
+      "loss": 1.8127,
+      "step": 311000
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.00021278253216231905,
+      "loss": 1.7889,
+      "step": 311500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.00021264208535420683,
+      "loss": 1.7924,
+      "step": 312000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.00021250163854609464,
+      "loss": 1.8108,
+      "step": 312500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0002123611917379824,
+      "loss": 1.8106,
+      "step": 313000
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0002122207449298702,
+      "loss": 1.7852,
+      "step": 313500
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.00021208029812175798,
+      "loss": 1.7817,
+      "step": 314000
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.0002119398513136458,
+      "loss": 1.7985,
+      "step": 314500
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.0002117994045055336,
+      "loss": 1.7904,
+      "step": 315000
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.00021165895769742138,
+      "loss": 1.8134,
+      "step": 315500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.0002115185108893092,
+      "loss": 1.812,
+      "step": 316000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.00021137806408119694,
+      "loss": 1.7939,
+      "step": 316500
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.00021123761727308475,
+      "loss": 1.7814,
+      "step": 317000
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.00021109717046497253,
+      "loss": 1.7592,
+      "step": 317500
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.00021095672365686034,
+      "loss": 1.7919,
+      "step": 318000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.00021081627684874815,
+      "loss": 1.7763,
+      "step": 318500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.00021067583004063593,
+      "loss": 1.8105,
+      "step": 319000
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.00021053538323252373,
+      "loss": 1.7793,
+      "step": 319500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.0002103949364244115,
+      "loss": 1.8149,
+      "step": 320000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0002102544896162993,
+      "loss": 1.7625,
+      "step": 320500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.0002101140428081871,
+      "loss": 1.6988,
+      "step": 321000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.00020997359600007489,
+      "loss": 1.6651,
+      "step": 321500
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0002098331491919627,
+      "loss": 1.6747,
+      "step": 322000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.00020969270238385047,
+      "loss": 1.7048,
+      "step": 322500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.00020955225557573828,
+      "loss": 1.6829,
+      "step": 323000
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.00020941180876762604,
+      "loss": 1.7218,
+      "step": 323500
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.00020927136195951384,
+      "loss": 1.7071,
+      "step": 324000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.00020913091515140165,
+      "loss": 1.7048,
+      "step": 324500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.00020899046834328943,
+      "loss": 1.7058,
+      "step": 325000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.00020885002153517724,
+      "loss": 1.7024,
+      "step": 325500
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.00020870957472706502,
+      "loss": 1.6729,
+      "step": 326000
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.00020856912791895283,
+      "loss": 1.7195,
+      "step": 326500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.00020842868111084058,
+      "loss": 1.723,
+      "step": 327000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0002082882343027284,
+      "loss": 1.7097,
+      "step": 327500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.0002081477874946162,
+      "loss": 1.7152,
+      "step": 328000
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.00020800734068650398,
+      "loss": 1.7013,
+      "step": 328500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0002078668938783918,
+      "loss": 1.6876,
+      "step": 329000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.00020772644707027954,
+      "loss": 1.6929,
+      "step": 329500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.00020758600026216735,
+      "loss": 1.7123,
+      "step": 330000
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.00020744555345405513,
+      "loss": 1.6763,
+      "step": 330500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.00020730510664594294,
+      "loss": 1.7054,
+      "step": 331000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.00020716465983783075,
+      "loss": 1.7192,
+      "step": 331500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.00020702421302971853,
+      "loss": 1.7188,
+      "step": 332000
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.00020688376622160634,
+      "loss": 1.7263,
+      "step": 332500
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0002067433194134941,
+      "loss": 1.7191,
+      "step": 333000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0002066028726053819,
+      "loss": 1.7121,
+      "step": 333500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.00020646242579726968,
+      "loss": 1.7201,
+      "step": 334000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0002063219789891575,
+      "loss": 1.7105,
+      "step": 334500
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.0002061815321810453,
+      "loss": 1.6985,
+      "step": 335000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.00020604108537293308,
+      "loss": 1.7075,
+      "step": 335500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.00020590063856482088,
+      "loss": 1.6683,
+      "step": 336000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.00020576019175670864,
+      "loss": 1.7711,
+      "step": 336500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.00020561974494859645,
+      "loss": 1.7415,
+      "step": 337000
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.00020547929814048423,
+      "loss": 1.7425,
+      "step": 337500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.00020533885133237203,
+      "loss": 1.7103,
+      "step": 338000
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.00020519840452425984,
+      "loss": 1.7212,
+      "step": 338500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.00020505795771614762,
+      "loss": 1.7515,
+      "step": 339000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.00020491751090803543,
+      "loss": 1.7124,
+      "step": 339500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.00020477706409992319,
+      "loss": 1.7298,
+      "step": 340000
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.000204636617291811,
+      "loss": 1.6947,
+      "step": 340500
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.00020449617048369877,
+      "loss": 1.7412,
+      "step": 341000
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.00020435572367558658,
+      "loss": 1.7487,
+      "step": 341500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.0002042152768674744,
+      "loss": 1.7391,
+      "step": 342000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.00020407483005936217,
+      "loss": 1.7039,
+      "step": 342500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.00020393438325124998,
+      "loss": 1.7487,
+      "step": 343000
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.00020379393644313773,
+      "loss": 1.7047,
+      "step": 343500
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.00020365348963502554,
+      "loss": 1.7137,
+      "step": 344000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.00020351304282691332,
+      "loss": 1.7259,
+      "step": 344500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.00020337259601880113,
+      "loss": 1.7112,
+      "step": 345000
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.00020323214921068894,
+      "loss": 1.7499,
+      "step": 345500
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.00020309170240257672,
+      "loss": 1.741,
+      "step": 346000
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0002029512555944645,
+      "loss": 1.6921,
+      "step": 346500
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.00020281080878635228,
+      "loss": 1.7582,
+      "step": 347000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.0002026703619782401,
+      "loss": 1.7167,
+      "step": 347500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.0002025299151701279,
+      "loss": 1.7155,
+      "step": 348000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.00020238946836201568,
+      "loss": 1.7296,
+      "step": 348500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.00020224902155390349,
+      "loss": 1.7333,
+      "step": 349000
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.00020210857474579124,
+      "loss": 1.6987,
+      "step": 349500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.00020196812793767905,
+      "loss": 1.7412,
+      "step": 350000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.00020182768112956683,
+      "loss": 1.7516,
+      "step": 350500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.00020168723432145464,
+      "loss": 1.7316,
+      "step": 351000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.00020154678751334244,
+      "loss": 1.7409,
+      "step": 351500
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.00020140634070523023,
+      "loss": 1.7477,
+      "step": 352000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.00020126589389711803,
+      "loss": 1.7584,
+      "step": 352500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.0002011254470890058,
+      "loss": 1.7328,
+      "step": 353000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0002009850002808936,
+      "loss": 1.7421,
+      "step": 353500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.00020084455347278138,
+      "loss": 1.719,
+      "step": 354000
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.00020070410666466918,
+      "loss": 1.7224,
+      "step": 354500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.000200563659856557,
+      "loss": 1.7464,
+      "step": 355000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.00020042321304844477,
+      "loss": 1.73,
+      "step": 355500
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.00020028276624033258,
+      "loss": 1.7005,
+      "step": 356000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.00020014231943222034,
+      "loss": 1.7133,
+      "step": 356500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.00020000187262410814,
+      "loss": 1.6669,
+      "step": 357000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.00019986142581599592,
+      "loss": 1.6335,
+      "step": 357500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.00019972097900788373,
+      "loss": 1.641,
+      "step": 358000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.00019958053219977154,
+      "loss": 1.6476,
+      "step": 358500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.00019944008539165932,
+      "loss": 1.6382,
+      "step": 359000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.00019929963858354713,
+      "loss": 1.6289,
+      "step": 359500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.00019915919177543488,
+      "loss": 1.6553,
+      "step": 360000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0001990187449673227,
+      "loss": 1.6524,
+      "step": 360500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.00019887829815921047,
+      "loss": 1.6656,
+      "step": 361000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.00019873785135109828,
+      "loss": 1.6646,
+      "step": 361500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0001985974045429861,
+      "loss": 1.6444,
+      "step": 362000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.00019845695773487387,
+      "loss": 1.6786,
+      "step": 362500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.00019831651092676165,
+      "loss": 1.6439,
+      "step": 363000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.00019817606411864943,
+      "loss": 1.6175,
+      "step": 363500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.00019803561731053724,
+      "loss": 1.6708,
+      "step": 364000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.00019789517050242502,
+      "loss": 1.6821,
+      "step": 364500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.00019775472369431283,
+      "loss": 1.6408,
+      "step": 365000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.00019761427688620064,
+      "loss": 1.6807,
+      "step": 365500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0001974738300780884,
+      "loss": 1.6527,
+      "step": 366000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0001973333832699762,
+      "loss": 1.6521,
+      "step": 366500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.00019719293646186398,
+      "loss": 1.6517,
+      "step": 367000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.00019705248965375179,
+      "loss": 1.6752,
+      "step": 367500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.00019691204284563957,
+      "loss": 1.6419,
+      "step": 368000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.00019677159603752738,
+      "loss": 1.6271,
+      "step": 368500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.00019663114922941518,
+      "loss": 1.6934,
+      "step": 369000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.00019649070242130294,
+      "loss": 1.6474,
+      "step": 369500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.00019635025561319075,
+      "loss": 1.6374,
+      "step": 370000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.00019620980880507853,
+      "loss": 1.6623,
+      "step": 370500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.00019606936199696633,
+      "loss": 1.6459,
+      "step": 371000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.00019592891518885412,
+      "loss": 1.6437,
+      "step": 371500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.00019578846838074192,
+      "loss": 1.6553,
+      "step": 372000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.00019564802157262973,
+      "loss": 1.655,
+      "step": 372500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.00019550757476451748,
+      "loss": 1.6601,
+      "step": 373000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.0001953671279564053,
+      "loss": 1.6885,
+      "step": 373500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.00019522668114829307,
+      "loss": 1.6711,
+      "step": 374000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.00019508623434018088,
+      "loss": 1.6831,
+      "step": 374500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0001949457875320687,
+      "loss": 1.6198,
+      "step": 375000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.00019480534072395647,
+      "loss": 1.6693,
+      "step": 375500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.00019466489391584428,
+      "loss": 1.651,
+      "step": 376000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.00019452444710773203,
+      "loss": 1.6937,
+      "step": 376500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.00019438400029961984,
+      "loss": 1.6631,
+      "step": 377000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.00019424355349150762,
+      "loss": 1.6963,
+      "step": 377500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.00019410310668339543,
+      "loss": 1.6647,
+      "step": 378000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.00019396265987528324,
+      "loss": 1.6662,
+      "step": 378500
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.00019382221306717102,
+      "loss": 1.68,
+      "step": 379000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0001936817662590588,
+      "loss": 1.6921,
+      "step": 379500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.00019354131945094658,
+      "loss": 1.6737,
+      "step": 380000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.0001934008726428344,
+      "loss": 1.6782,
+      "step": 380500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.00019326042583472217,
+      "loss": 1.667,
+      "step": 381000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.00019311997902660998,
+      "loss": 1.6799,
+      "step": 381500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.00019297953221849779,
+      "loss": 1.6438,
+      "step": 382000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.00019283908541038554,
+      "loss": 1.6626,
+      "step": 382500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.00019269863860227335,
+      "loss": 1.6698,
+      "step": 383000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.00019255819179416113,
+      "loss": 1.6847,
+      "step": 383500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.00019241774498604894,
+      "loss": 1.725,
+      "step": 384000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.00019227729817793672,
+      "loss": 1.6462,
+      "step": 384500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.00019213685136982452,
+      "loss": 1.6951,
+      "step": 385000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.00019199640456171233,
+      "loss": 1.6711,
+      "step": 385500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0001918559577536001,
+      "loss": 1.6783,
+      "step": 386000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0001917155109454879,
+      "loss": 1.6566,
+      "step": 386500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.00019157506413737568,
+      "loss": 1.6606,
+      "step": 387000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.00019143461732926348,
+      "loss": 1.6585,
+      "step": 387500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.00019129417052115126,
+      "loss": 1.7094,
+      "step": 388000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.00019115372371303907,
+      "loss": 1.638,
+      "step": 388500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.00019101327690492688,
+      "loss": 1.6686,
+      "step": 389000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.00019087283009681463,
+      "loss": 1.6696,
+      "step": 389500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.00019073238328870244,
+      "loss": 1.6516,
+      "step": 390000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.00019059193648059022,
+      "loss": 1.6625,
+      "step": 390500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.00019045148967247803,
+      "loss": 1.6865,
+      "step": 391000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0001903110428643658,
+      "loss": 1.6903,
+      "step": 391500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.00019017059605625362,
+      "loss": 1.6686,
+      "step": 392000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.00019003014924814143,
+      "loss": 1.5473,
+      "step": 392500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.00018988970244002918,
+      "loss": 1.5807,
+      "step": 393000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.000189749255631917,
+      "loss": 1.5696,
+      "step": 393500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.00018960880882380477,
+      "loss": 1.5851,
+      "step": 394000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.00018946836201569258,
+      "loss": 1.5706,
+      "step": 394500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.00018932791520758036,
+      "loss": 1.5829,
+      "step": 395000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.00018918746839946817,
+      "loss": 1.6146,
+      "step": 395500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.00018904702159135598,
+      "loss": 1.6072,
+      "step": 396000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.00018890657478324373,
+      "loss": 1.587,
+      "step": 396500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.00018876612797513154,
+      "loss": 1.6045,
+      "step": 397000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.00018862568116701932,
+      "loss": 1.5912,
+      "step": 397500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.00018848523435890713,
+      "loss": 1.6025,
+      "step": 398000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.00018834478755079493,
+      "loss": 1.568,
+      "step": 398500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.0001882043407426827,
+      "loss": 1.5908,
+      "step": 399000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0001880638939345705,
+      "loss": 1.5785,
+      "step": 399500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.00018792344712645828,
+      "loss": 1.5831,
+      "step": 400000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.00018778300031834609,
+      "loss": 1.6117,
+      "step": 400500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.00018764255351023387,
+      "loss": 1.5902,
+      "step": 401000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.00018750210670212167,
+      "loss": 1.6121,
+      "step": 401500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.00018736165989400948,
+      "loss": 1.619,
+      "step": 402000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.00018722121308589724,
+      "loss": 1.5974,
+      "step": 402500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.00018708076627778504,
+      "loss": 1.6051,
+      "step": 403000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.00018694031946967283,
+      "loss": 1.6172,
+      "step": 403500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.00018679987266156063,
+      "loss": 1.603,
+      "step": 404000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.00018665942585344841,
+      "loss": 1.6111,
+      "step": 404500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.00018651897904533622,
+      "loss": 1.5969,
+      "step": 405000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.00018637853223722403,
+      "loss": 1.6005,
+      "step": 405500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.00018623808542911178,
+      "loss": 1.6083,
+      "step": 406000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.0001860976386209996,
+      "loss": 1.6204,
+      "step": 406500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.00018595719181288737,
+      "loss": 1.6001,
+      "step": 407000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.00018581674500477518,
+      "loss": 1.5974,
+      "step": 407500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.00018567629819666296,
+      "loss": 1.6192,
+      "step": 408000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.00018553585138855077,
+      "loss": 1.6349,
+      "step": 408500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.00018539540458043858,
+      "loss": 1.6468,
+      "step": 409000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.00018525495777232633,
+      "loss": 1.5827,
+      "step": 409500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.00018511451096421414,
+      "loss": 1.5988,
+      "step": 410000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.00018497406415610192,
+      "loss": 1.6065,
+      "step": 410500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.00018483361734798973,
+      "loss": 1.628,
+      "step": 411000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.0001846931705398775,
+      "loss": 1.6265,
+      "step": 411500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.00018455272373176532,
+      "loss": 1.5977,
+      "step": 412000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.00018441227692365313,
+      "loss": 1.5836,
+      "step": 412500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.00018427183011554088,
+      "loss": 1.5906,
+      "step": 413000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.0001841313833074287,
+      "loss": 1.6158,
+      "step": 413500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.00018399093649931647,
+      "loss": 1.6166,
+      "step": 414000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.00018385048969120428,
+      "loss": 1.6381,
+      "step": 414500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.00018371004288309206,
+      "loss": 1.646,
+      "step": 415000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.00018356959607497987,
+      "loss": 1.6012,
+      "step": 415500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.00018342914926686765,
+      "loss": 1.6277,
+      "step": 416000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.00018328870245875543,
+      "loss": 1.6731,
+      "step": 416500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.00018314825565064324,
+      "loss": 1.6306,
+      "step": 417000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.00018300780884253102,
+      "loss": 1.6024,
+      "step": 417500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.00018286736203441882,
+      "loss": 1.6096,
+      "step": 418000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0001827269152263066,
+      "loss": 1.6349,
+      "step": 418500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.00018258646841819439,
+      "loss": 1.611,
+      "step": 419000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.0001824460216100822,
+      "loss": 1.6175,
+      "step": 419500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.00018230557480196997,
+      "loss": 1.6383,
+      "step": 420000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.00018216512799385778,
+      "loss": 1.6082,
+      "step": 420500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.00018202468118574556,
+      "loss": 1.6501,
+      "step": 421000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.00018188423437763337,
+      "loss": 1.6195,
+      "step": 421500
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.00018174378756952113,
+      "loss": 1.6106,
+      "step": 422000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.00018160334076140893,
+      "loss": 1.5999,
+      "step": 422500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.00018146289395329674,
+      "loss": 1.6126,
+      "step": 423000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.00018132244714518452,
+      "loss": 1.6388,
+      "step": 423500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.00018118200033707233,
+      "loss": 1.586,
+      "step": 424000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0001810415535289601,
+      "loss": 1.6462,
+      "step": 424500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.00018090110672084792,
+      "loss": 1.6112,
+      "step": 425000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.00018076065991273573,
+      "loss": 1.5967,
+      "step": 425500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.00018062021310462348,
+      "loss": 1.6225,
+      "step": 426000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0001804797662965113,
+      "loss": 1.6086,
+      "step": 426500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.00018033931948839907,
+      "loss": 1.6326,
+      "step": 427000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.00018019887268028688,
+      "loss": 1.6163,
+      "step": 427500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.00018005842587217466,
+      "loss": 1.5382,
+      "step": 428000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.00017991797906406247,
+      "loss": 1.5531,
+      "step": 428500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.00017977753225595028,
+      "loss": 1.5373,
+      "step": 429000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.00017963708544783803,
+      "loss": 1.5424,
+      "step": 429500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.00017949663863972584,
+      "loss": 1.5394,
+      "step": 430000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.00017935619183161362,
+      "loss": 1.5331,
+      "step": 430500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.00017921574502350143,
+      "loss": 1.5342,
+      "step": 431000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.0001790752982153892,
+      "loss": 1.524,
+      "step": 431500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.00017893485140727702,
+      "loss": 1.5285,
+      "step": 432000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.0001787944045991648,
+      "loss": 1.5422,
+      "step": 432500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.00017865395779105258,
+      "loss": 1.5073,
+      "step": 433000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.00017851351098294038,
+      "loss": 1.5426,
+      "step": 433500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.00017837306417482817,
+      "loss": 1.5532,
+      "step": 434000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.00017823261736671597,
+      "loss": 1.5142,
+      "step": 434500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.00017809217055860375,
+      "loss": 1.5332,
+      "step": 435000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.00017795172375049154,
+      "loss": 1.5548,
+      "step": 435500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.00017781127694237934,
+      "loss": 1.5717,
+      "step": 436000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.00017767083013426712,
+      "loss": 1.5633,
+      "step": 436500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.00017753038332615493,
+      "loss": 1.5532,
+      "step": 437000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.00017738993651804271,
+      "loss": 1.562,
+      "step": 437500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.00017724948970993052,
+      "loss": 1.5514,
+      "step": 438000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.00017710904290181828,
+      "loss": 1.5566,
+      "step": 438500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.00017696859609370608,
+      "loss": 1.5624,
+      "step": 439000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0001768281492855939,
+      "loss": 1.51,
+      "step": 439500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.00017668770247748167,
+      "loss": 1.5532,
+      "step": 440000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.00017654725566936948,
+      "loss": 1.5621,
+      "step": 440500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.00017640680886125726,
+      "loss": 1.5415,
+      "step": 441000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.00017626636205314507,
+      "loss": 1.5547,
+      "step": 441500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.00017612591524503282,
+      "loss": 1.5488,
+      "step": 442000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.00017598546843692063,
+      "loss": 1.5488,
+      "step": 442500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.00017584502162880844,
+      "loss": 1.5959,
+      "step": 443000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.00017570457482069622,
+      "loss": 1.5253,
+      "step": 443500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.00017556412801258403,
+      "loss": 1.5767,
+      "step": 444000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.0001754236812044718,
+      "loss": 1.557,
+      "step": 444500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.00017528323439635962,
+      "loss": 1.5658,
+      "step": 445000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.00017514278758824737,
+      "loss": 1.5549,
+      "step": 445500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.00017500234078013518,
+      "loss": 1.555,
+      "step": 446000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.000174861893972023,
+      "loss": 1.5539,
+      "step": 446500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.00017472144716391077,
+      "loss": 1.5777,
+      "step": 447000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.00017458100035579858,
+      "loss": 1.543,
+      "step": 447500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.00017444055354768636,
+      "loss": 1.5948,
+      "step": 448000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.00017430010673957416,
+      "loss": 1.5819,
+      "step": 448500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.00017415965993146195,
+      "loss": 1.5539,
+      "step": 449000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.00017401921312334973,
+      "loss": 1.566,
+      "step": 449500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.00017387876631523753,
+      "loss": 1.5644,
+      "step": 450000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.00017373831950712532,
+      "loss": 1.556,
+      "step": 450500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.00017359787269901312,
+      "loss": 1.5502,
+      "step": 451000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0001734574258909009,
+      "loss": 1.581,
+      "step": 451500
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.00017331697908278869,
+      "loss": 1.5689,
+      "step": 452000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0001731765322746765,
+      "loss": 1.594,
+      "step": 452500
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.00017303608546656427,
+      "loss": 1.5912,
+      "step": 453000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.00017289563865845208,
+      "loss": 1.5534,
+      "step": 453500
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.00017275519185033986,
+      "loss": 1.5707,
+      "step": 454000
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.00017261474504222767,
+      "loss": 1.5652,
+      "step": 454500
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.00017247429823411543,
+      "loss": 1.5836,
+      "step": 455000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.00017233385142600323,
+      "loss": 1.5521,
+      "step": 455500
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.00017219340461789104,
+      "loss": 1.5621,
+      "step": 456000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.00017205295780977882,
+      "loss": 1.5712,
+      "step": 456500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.00017191251100166663,
+      "loss": 1.5442,
+      "step": 457000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.0001717720641935544,
+      "loss": 1.5823,
+      "step": 457500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.00017163161738544222,
+      "loss": 1.5797,
+      "step": 458000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.00017149117057732997,
+      "loss": 1.5101,
+      "step": 458500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.00017135072376921778,
+      "loss": 1.5733,
+      "step": 459000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0001712102769611056,
+      "loss": 1.5874,
+      "step": 459500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.00017106983015299337,
+      "loss": 1.5539,
+      "step": 460000
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.00017092938334488118,
+      "loss": 1.5474,
+      "step": 460500
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.00017078893653676896,
+      "loss": 1.5657,
+      "step": 461000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.00017064848972865677,
+      "loss": 1.5742,
+      "step": 461500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.00017050804292054452,
+      "loss": 1.5759,
+      "step": 462000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.00017036759611243233,
+      "loss": 1.5867,
+      "step": 462500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.00017022714930432014,
+      "loss": 1.5741,
+      "step": 463000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.00017008670249620792,
+      "loss": 1.5301,
+      "step": 463500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.00016994625568809573,
+      "loss": 1.4812,
+      "step": 464000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0001698058088799835,
+      "loss": 1.4959,
+      "step": 464500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.00016966536207187131,
+      "loss": 1.4608,
+      "step": 465000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.00016952491526375907,
+      "loss": 1.5079,
+      "step": 465500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.00016938446845564688,
+      "loss": 1.4575,
+      "step": 466000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.00016924402164753468,
+      "loss": 1.4994,
+      "step": 466500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.00016910357483942247,
+      "loss": 1.5024,
+      "step": 467000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.00016896312803131027,
+      "loss": 1.4771,
+      "step": 467500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.00016882268122319805,
+      "loss": 1.4988,
+      "step": 468000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.00016868223441508583,
+      "loss": 1.495,
+      "step": 468500
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.00016854178760697362,
+      "loss": 1.4725,
+      "step": 469000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.00016840134079886142,
+      "loss": 1.4715,
+      "step": 469500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.00016826089399074923,
+      "loss": 1.5111,
+      "step": 470000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.000168120447182637,
+      "loss": 1.5205,
+      "step": 470500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.00016798000037452482,
+      "loss": 1.5102,
+      "step": 471000
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.00016783955356641257,
+      "loss": 1.4871,
+      "step": 471500
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.00016769910675830038,
+      "loss": 1.4828,
+      "step": 472000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.00016755865995018816,
+      "loss": 1.4935,
+      "step": 472500
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.00016741821314207597,
+      "loss": 1.4811,
+      "step": 473000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.00016727776633396378,
+      "loss": 1.4903,
+      "step": 473500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.00016713731952585156,
+      "loss": 1.5206,
+      "step": 474000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.00016699687271773937,
+      "loss": 1.4896,
+      "step": 474500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.00016685642590962712,
+      "loss": 1.5221,
+      "step": 475000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.00016671597910151493,
+      "loss": 1.5004,
+      "step": 475500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.00016657553229340274,
+      "loss": 1.5036,
+      "step": 476000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.00016643508548529052,
+      "loss": 1.5516,
+      "step": 476500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.00016629463867717833,
+      "loss": 1.5031,
+      "step": 477000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0001661541918690661,
+      "loss": 1.516,
+      "step": 477500
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.00016601374506095392,
+      "loss": 1.5192,
+      "step": 478000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.00016587329825284167,
+      "loss": 1.5089,
+      "step": 478500
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.00016573285144472948,
+      "loss": 1.5197,
+      "step": 479000
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.00016559240463661729,
+      "loss": 1.5161,
+      "step": 479500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.00016545195782850507,
+      "loss": 1.5045,
+      "step": 480000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.00016531151102039288,
+      "loss": 1.4987,
+      "step": 480500
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.00016517106421228066,
+      "loss": 1.5233,
+      "step": 481000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.00016503061740416846,
+      "loss": 1.5311,
+      "step": 481500
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.00016489017059605622,
+      "loss": 1.507,
+      "step": 482000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.00016474972378794403,
+      "loss": 1.5194,
+      "step": 482500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.00016460927697983183,
+      "loss": 1.5057,
+      "step": 483000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.00016446883017171961,
+      "loss": 1.5187,
+      "step": 483500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.00016432838336360742,
+      "loss": 1.4982,
+      "step": 484000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0001641879365554952,
+      "loss": 1.5115,
+      "step": 484500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.000164047489747383,
+      "loss": 1.4948,
+      "step": 485000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.00016390704293927077,
+      "loss": 1.5035,
+      "step": 485500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.00016376659613115857,
+      "loss": 1.4884,
+      "step": 486000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.00016362614932304638,
+      "loss": 1.518,
+      "step": 486500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.00016348570251493416,
+      "loss": 1.494,
+      "step": 487000
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.00016334525570682197,
+      "loss": 1.4975,
+      "step": 487500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.00016320480889870975,
+      "loss": 1.5047,
+      "step": 488000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.00016306436209059753,
+      "loss": 1.5242,
+      "step": 488500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0001629239152824853,
+      "loss": 1.5266,
+      "step": 489000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.00016278346847437312,
+      "loss": 1.5154,
+      "step": 489500
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.00016264302166626093,
+      "loss": 1.5013,
+      "step": 490000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0001625025748581487,
+      "loss": 1.5039,
+      "step": 490500
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.00016236212805003652,
+      "loss": 1.4944,
+      "step": 491000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.00016222168124192427,
+      "loss": 1.5096,
+      "step": 491500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.00016208123443381208,
+      "loss": 1.5127,
+      "step": 492000
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.00016194078762569986,
+      "loss": 1.4777,
+      "step": 492500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.00016180034081758767,
+      "loss": 1.5085,
+      "step": 493000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.00016165989400947548,
+      "loss": 1.5391,
+      "step": 493500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.00016151944720136326,
+      "loss": 1.5089,
+      "step": 494000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.00016137900039325107,
+      "loss": 1.5116,
+      "step": 494500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.00016123855358513882,
+      "loss": 1.5482,
+      "step": 495000
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.00016109810677702663,
+      "loss": 1.5358,
+      "step": 495500
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.0001609576599689144,
+      "loss": 1.507,
+      "step": 496000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.00016081721316080222,
+      "loss": 1.5343,
+      "step": 496500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.00016067676635269002,
+      "loss": 1.5384,
+      "step": 497000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.0001605363195445778,
+      "loss": 1.4906,
+      "step": 497500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.00016039587273646561,
+      "loss": 1.5225,
+      "step": 498000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.00016025542592835337,
+      "loss": 1.5258,
+      "step": 498500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.00016011497912024118,
+      "loss": 1.5214,
+      "step": 499000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.00015997453231212896,
+      "loss": 1.44,
+      "step": 499500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.00015983408550401676,
+      "loss": 1.4353,
+      "step": 500000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.00015969363869590457,
+      "loss": 1.4431,
+      "step": 500500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.00015955319188779235,
+      "loss": 1.4284,
+      "step": 501000
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.00015941274507968016,
+      "loss": 1.4001,
+      "step": 501500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.00015927229827156792,
+      "loss": 1.453,
+      "step": 502000
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.00015913185146345572,
+      "loss": 1.4384,
+      "step": 502500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.00015899140465534353,
+      "loss": 1.4264,
+      "step": 503000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.0001588509578472313,
+      "loss": 1.4246,
+      "step": 503500
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.00015871051103911912,
+      "loss": 1.4032,
+      "step": 504000
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0001585700642310069,
+      "loss": 1.4715,
+      "step": 504500
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.00015842961742289468,
+      "loss": 1.4753,
+      "step": 505000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.00015828917061478246,
+      "loss": 1.4413,
+      "step": 505500
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.00015814872380667027,
+      "loss": 1.4542,
+      "step": 506000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.00015800827699855808,
+      "loss": 1.4645,
+      "step": 506500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.00015786783019044586,
+      "loss": 1.4672,
+      "step": 507000
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.00015772738338233367,
+      "loss": 1.4443,
+      "step": 507500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.00015758693657422142,
+      "loss": 1.4625,
+      "step": 508000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.00015744648976610923,
+      "loss": 1.4585,
+      "step": 508500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.000157306042957997,
+      "loss": 1.4591,
+      "step": 509000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.00015716559614988482,
+      "loss": 1.4824,
+      "step": 509500
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.00015702514934177263,
+      "loss": 1.4764,
+      "step": 510000
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0001568847025336604,
+      "loss": 1.476,
+      "step": 510500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.00015674425572554822,
+      "loss": 1.4494,
+      "step": 511000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.00015660380891743597,
+      "loss": 1.4923,
+      "step": 511500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.00015646336210932378,
+      "loss": 1.4689,
+      "step": 512000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.00015632291530121156,
+      "loss": 1.4711,
+      "step": 512500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.00015618246849309937,
+      "loss": 1.4899,
+      "step": 513000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.00015604202168498717,
+      "loss": 1.4408,
+      "step": 513500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.00015590157487687496,
+      "loss": 1.4241,
+      "step": 514000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.00015576112806876276,
+      "loss": 1.4586,
+      "step": 514500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.00015562068126065052,
+      "loss": 1.4789,
+      "step": 515000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.00015548023445253833,
+      "loss": 1.4828,
+      "step": 515500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0001553397876444261,
+      "loss": 1.4457,
+      "step": 516000
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.00015519934083631391,
+      "loss": 1.4695,
+      "step": 516500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.00015505889402820172,
+      "loss": 1.4421,
+      "step": 517000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0001549184472200895,
+      "loss": 1.4464,
+      "step": 517500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.0001547780004119773,
+      "loss": 1.439,
+      "step": 518000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.00015463755360386506,
+      "loss": 1.4784,
+      "step": 518500
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.00015449710679575287,
+      "loss": 1.423,
+      "step": 519000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.00015435665998764065,
+      "loss": 1.4751,
+      "step": 519500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.00015421621317952846,
+      "loss": 1.4603,
+      "step": 520000
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.00015407576637141627,
+      "loss": 1.4713,
+      "step": 520500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00015393531956330405,
+      "loss": 1.4577,
+      "step": 521000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.00015379487275519183,
+      "loss": 1.4449,
+      "step": 521500
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0001536544259470796,
+      "loss": 1.4836,
+      "step": 522000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.00015351397913896742,
+      "loss": 1.4734,
+      "step": 522500
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0001533735323308552,
+      "loss": 1.4549,
+      "step": 523000
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.000153233085522743,
+      "loss": 1.4572,
+      "step": 523500
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.00015309263871463082,
+      "loss": 1.4503,
+      "step": 524000
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.00015295219190651857,
+      "loss": 1.4661,
+      "step": 524500
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.00015281174509840638,
+      "loss": 1.4359,
+      "step": 525000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.00015267129829029416,
+      "loss": 1.4545,
+      "step": 525500
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.00015253085148218197,
+      "loss": 1.4696,
+      "step": 526000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.00015239040467406978,
+      "loss": 1.4677,
+      "step": 526500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.00015224995786595756,
+      "loss": 1.4685,
+      "step": 527000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.00015210951105784537,
+      "loss": 1.4631,
+      "step": 527500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.00015196906424973312,
+      "loss": 1.4639,
+      "step": 528000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.00015182861744162093,
+      "loss": 1.4791,
+      "step": 528500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.0001516881706335087,
+      "loss": 1.4634,
+      "step": 529000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.00015154772382539652,
+      "loss": 1.4521,
+      "step": 529500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.00015140727701728432,
+      "loss": 1.4717,
+      "step": 530000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0001512668302091721,
+      "loss": 1.4843,
+      "step": 530500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0001511263834010599,
+      "loss": 1.4988,
+      "step": 531000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.00015098593659294767,
+      "loss": 1.4741,
+      "step": 531500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.00015084548978483547,
+      "loss": 1.471,
+      "step": 532000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.00015070504297672326,
+      "loss": 1.4618,
+      "step": 532500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.00015056459616861106,
+      "loss": 1.4727,
+      "step": 533000
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.00015042414936049887,
+      "loss": 1.441,
+      "step": 533500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.00015028370255238665,
+      "loss": 1.513,
+      "step": 534000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.00015014325574427446,
+      "loss": 1.4718,
+      "step": 534500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.00015000280893616221,
+      "loss": 1.4022,
+      "step": 535000
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.00014986236212805002,
+      "loss": 1.3613,
+      "step": 535500
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.00014972191531993783,
+      "loss": 1.3951,
+      "step": 536000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0001495814685118256,
+      "loss": 1.4099,
+      "step": 536500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0001494410217037134,
+      "loss": 1.4049,
+      "step": 537000
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.0001493005748956012,
+      "loss": 1.3861,
+      "step": 537500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.00014916012808748898,
+      "loss": 1.4068,
+      "step": 538000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0001490196812793768,
+      "loss": 1.3857,
+      "step": 538500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.00014887923447126457,
+      "loss": 1.3978,
+      "step": 539000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.00014873878766315235,
+      "loss": 1.4054,
+      "step": 539500
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.00014859834085504016,
+      "loss": 1.4441,
+      "step": 540000
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.00014845789404692794,
+      "loss": 1.4118,
+      "step": 540500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.00014831744723881572,
+      "loss": 1.395,
+      "step": 541000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.00014817700043070353,
+      "loss": 1.3928,
+      "step": 541500
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.00014803655362259134,
+      "loss": 1.4088,
+      "step": 542000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.00014789610681447912,
+      "loss": 1.4245,
+      "step": 542500
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.0001477556600063669,
+      "loss": 1.3877,
+      "step": 543000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0001476152131982547,
+      "loss": 1.4311,
+      "step": 543500
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.0001474747663901425,
+      "loss": 1.4027,
+      "step": 544000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.00014733431958203027,
+      "loss": 1.4169,
+      "step": 544500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.00014719387277391808,
+      "loss": 1.3918,
+      "step": 545000
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.00014705342596580588,
+      "loss": 1.4059,
+      "step": 545500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.00014691297915769367,
+      "loss": 1.4136,
+      "step": 546000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.00014677253234958145,
+      "loss": 1.3865,
+      "step": 546500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.00014663208554146925,
+      "loss": 1.3966,
+      "step": 547000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.00014649163873335704,
+      "loss": 1.4183,
+      "step": 547500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.00014635119192524482,
+      "loss": 1.4159,
+      "step": 548000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.00014621074511713262,
+      "loss": 1.4426,
+      "step": 548500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.00014607029830902043,
+      "loss": 1.4249,
+      "step": 549000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0001459298515009082,
+      "loss": 1.4265,
+      "step": 549500
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.000145789404692796,
+      "loss": 1.4052,
+      "step": 550000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.0001456489578846838,
+      "loss": 1.3974,
+      "step": 550500
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.00014550851107657158,
+      "loss": 1.4252,
+      "step": 551000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.00014536806426845936,
+      "loss": 1.4379,
+      "step": 551500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.00014522761746034717,
+      "loss": 1.448,
+      "step": 552000
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.00014508717065223498,
+      "loss": 1.3834,
+      "step": 552500
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.00014494672384412276,
+      "loss": 1.44,
+      "step": 553000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.00014480627703601054,
+      "loss": 1.4171,
+      "step": 553500
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.00014466583022789835,
+      "loss": 1.3753,
+      "step": 554000
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.00014452538341978613,
+      "loss": 1.4184,
+      "step": 554500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0001443849366116739,
+      "loss": 1.4294,
+      "step": 555000
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.00014424448980356172,
+      "loss": 1.4299,
+      "step": 555500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.00014410404299544953,
+      "loss": 1.4341,
+      "step": 556000
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0001439635961873373,
+      "loss": 1.4158,
+      "step": 556500
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.0001438231493792251,
+      "loss": 1.3985,
+      "step": 557000
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0001436827025711129,
+      "loss": 1.4194,
+      "step": 557500
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.00014354225576300068,
+      "loss": 1.4301,
+      "step": 558000
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.0001434018089548885,
+      "loss": 1.4088,
+      "step": 558500
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.00014326136214677627,
+      "loss": 1.4206,
+      "step": 559000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.00014312091533866405,
+      "loss": 1.4305,
+      "step": 559500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.00014298046853055186,
+      "loss": 1.4144,
+      "step": 560000
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.00014284002172243964,
+      "loss": 1.3909,
+      "step": 560500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.00014269957491432742,
+      "loss": 1.3907,
+      "step": 561000
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.00014255912810621523,
+      "loss": 1.42,
+      "step": 561500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.00014241868129810303,
+      "loss": 1.4351,
+      "step": 562000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.00014227823448999082,
+      "loss": 1.4395,
+      "step": 562500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.0001421377876818786,
+      "loss": 1.4346,
+      "step": 563000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.0001419973408737664,
+      "loss": 1.4151,
+      "step": 563500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.00014185689406565419,
+      "loss": 1.411,
+      "step": 564000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.00014171644725754197,
+      "loss": 1.4364,
+      "step": 564500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.00014157600044942977,
+      "loss": 1.4227,
+      "step": 565000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.00014143555364131758,
+      "loss": 1.4068,
+      "step": 565500
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.00014129510683320536,
+      "loss": 1.4267,
+      "step": 566000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.00014115466002509314,
+      "loss": 1.4291,
+      "step": 566500
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.00014101421321698095,
+      "loss": 1.4357,
+      "step": 567000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.00014087376640886873,
+      "loss": 1.3937,
+      "step": 567500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.00014073331960075651,
+      "loss": 1.4021,
+      "step": 568000
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.00014059287279264432,
+      "loss": 1.4269,
+      "step": 568500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.00014045242598453213,
+      "loss": 1.4335,
+      "step": 569000
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.0001403119791764199,
+      "loss": 1.4595,
+      "step": 569500
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.0001401715323683077,
+      "loss": 1.4421,
+      "step": 570000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0001400310855601955,
+      "loss": 1.3778,
+      "step": 570500
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.00013989063875208328,
+      "loss": 1.3223,
+      "step": 571000
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.00013975019194397106,
+      "loss": 1.3708,
+      "step": 571500
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.00013960974513585887,
+      "loss": 1.3409,
+      "step": 572000
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.00013946929832774668,
+      "loss": 1.3737,
+      "step": 572500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.00013932885151963446,
+      "loss": 1.3662,
+      "step": 573000
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.00013918840471152224,
+      "loss": 1.3599,
+      "step": 573500
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.00013904795790341005,
+      "loss": 1.3449,
+      "step": 574000
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.00013890751109529783,
+      "loss": 1.3371,
+      "step": 574500
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.0001387670642871856,
+      "loss": 1.3623,
+      "step": 575000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.00013862661747907342,
+      "loss": 1.3336,
+      "step": 575500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.0001384861706709612,
+      "loss": 1.3426,
+      "step": 576000
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.000138345723862849,
+      "loss": 1.389,
+      "step": 576500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0001382052770547368,
+      "loss": 1.3399,
+      "step": 577000
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.00013806483024662457,
+      "loss": 1.3722,
+      "step": 577500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.00013792438343851238,
+      "loss": 1.3475,
+      "step": 578000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.00013778393663040016,
+      "loss": 1.3582,
+      "step": 578500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.00013764348982228796,
+      "loss": 1.3363,
+      "step": 579000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.00013750304301417575,
+      "loss": 1.3506,
+      "step": 579500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.00013736259620606355,
+      "loss": 1.3908,
+      "step": 580000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.00013722214939795133,
+      "loss": 1.3727,
+      "step": 580500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.00013708170258983912,
+      "loss": 1.3497,
+      "step": 581000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.00013694125578172692,
+      "loss": 1.357,
+      "step": 581500
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0001368008089736147,
+      "loss": 1.3516,
+      "step": 582000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.0001366603621655025,
+      "loss": 1.3448,
+      "step": 582500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.0001365199153573903,
+      "loss": 1.3756,
+      "step": 583000
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0001363794685492781,
+      "loss": 1.3671,
+      "step": 583500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.00013623902174116588,
+      "loss": 1.3616,
+      "step": 584000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.00013609857493305366,
+      "loss": 1.3418,
+      "step": 584500
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.00013595812812494147,
+      "loss": 1.3471,
+      "step": 585000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.00013581768131682928,
+      "loss": 1.358,
+      "step": 585500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.00013567723450871706,
+      "loss": 1.357,
+      "step": 586000
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.00013553678770060484,
+      "loss": 1.3636,
+      "step": 586500
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.00013539634089249265,
+      "loss": 1.3472,
+      "step": 587000
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.00013525589408438043,
+      "loss": 1.3594,
+      "step": 587500
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0001351154472762682,
+      "loss": 1.3551,
+      "step": 588000
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.00013497500046815602,
+      "loss": 1.3657,
+      "step": 588500
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.00013483455366004383,
+      "loss": 1.3729,
+      "step": 589000
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.0001346941068519316,
+      "loss": 1.3573,
+      "step": 589500
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0001345536600438194,
+      "loss": 1.336,
+      "step": 590000
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0001344132132357072,
+      "loss": 1.3719,
+      "step": 590500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.00013427276642759498,
+      "loss": 1.3597,
+      "step": 591000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.00013413231961948276,
+      "loss": 1.3644,
+      "step": 591500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.00013399187281137057,
+      "loss": 1.3769,
+      "step": 592000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.00013385142600325835,
+      "loss": 1.372,
+      "step": 592500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.00013371097919514616,
+      "loss": 1.3404,
+      "step": 593000
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.00013357053238703394,
+      "loss": 1.3573,
+      "step": 593500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.00013343008557892172,
+      "loss": 1.3646,
+      "step": 594000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.00013328963877080953,
+      "loss": 1.3726,
+      "step": 594500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0001331491919626973,
+      "loss": 1.347,
+      "step": 595000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.0001330087451545851,
+      "loss": 1.3938,
+      "step": 595500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0001328682983464729,
+      "loss": 1.3703,
+      "step": 596000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0001327278515383607,
+      "loss": 1.3553,
+      "step": 596500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.00013258740473024848,
+      "loss": 1.3985,
+      "step": 597000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.00013244695792213627,
+      "loss": 1.3638,
+      "step": 597500
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.00013230651111402407,
+      "loss": 1.3921,
+      "step": 598000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.00013216606430591185,
+      "loss": 1.3586,
+      "step": 598500
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.00013202561749779966,
+      "loss": 1.3723,
+      "step": 599000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.00013188517068968744,
+      "loss": 1.3877,
+      "step": 599500
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.00013174472388157525,
+      "loss": 1.396,
+      "step": 600000
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.00013160427707346303,
+      "loss": 1.3886,
+      "step": 600500
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.0001314638302653508,
+      "loss": 1.3702,
+      "step": 601000
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.00013132338345723862,
+      "loss": 1.3806,
+      "step": 601500
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.0001311829366491264,
+      "loss": 1.3646,
+      "step": 602000
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0001310424898410142,
+      "loss": 1.3682,
+      "step": 602500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.000130902043032902,
+      "loss": 1.3821,
+      "step": 603000
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.0001307615962247898,
+      "loss": 1.3607,
+      "step": 603500
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.00013062114941667758,
+      "loss": 1.3596,
+      "step": 604000
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.00013048070260856536,
+      "loss": 1.3993,
+      "step": 604500
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.00013034025580045317,
+      "loss": 1.3803,
+      "step": 605000
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.00013019980899234095,
+      "loss": 1.3975,
+      "step": 605500
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.00013005936218422876,
+      "loss": 1.3645,
+      "step": 606000
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.00012991891537611654,
+      "loss": 1.2937,
+      "step": 606500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.00012977846856800435,
+      "loss": 1.3025,
+      "step": 607000
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.00012963802175989213,
+      "loss": 1.2941,
+      "step": 607500
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.0001294975749517799,
+      "loss": 1.3042,
+      "step": 608000
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.00012935712814366772,
+      "loss": 1.3002,
+      "step": 608500
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.0001292166813355555,
+      "loss": 1.3202,
+      "step": 609000
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0001290762345274433,
+      "loss": 1.2977,
+      "step": 609500
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.00012893578771933109,
+      "loss": 1.2959,
+      "step": 610000
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.00012879534091121887,
+      "loss": 1.3038,
+      "step": 610500
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.00012865489410310668,
+      "loss": 1.3174,
+      "step": 611000
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.00012851444729499446,
+      "loss": 1.2808,
+      "step": 611500
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.00012837400048688224,
+      "loss": 1.3028,
+      "step": 612000
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.00012823355367877005,
+      "loss": 1.3325,
+      "step": 612500
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.00012809310687065785,
+      "loss": 1.3061,
+      "step": 613000
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.00012795266006254563,
+      "loss": 1.3116,
+      "step": 613500
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.00012781221325443341,
+      "loss": 1.3119,
+      "step": 614000
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.00012767176644632122,
+      "loss": 1.303,
+      "step": 614500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.000127531319638209,
+      "loss": 1.276,
+      "step": 615000
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.00012739087283009678,
+      "loss": 1.3066,
+      "step": 615500
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.0001272504260219846,
+      "loss": 1.3132,
+      "step": 616000
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.0001271099792138724,
+      "loss": 1.2656,
+      "step": 616500
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.00012696953240576018,
+      "loss": 1.3112,
+      "step": 617000
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.00012682908559764796,
+      "loss": 1.337,
+      "step": 617500
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.00012668863878953577,
+      "loss": 1.3286,
+      "step": 618000
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.00012654819198142355,
+      "loss": 1.2837,
+      "step": 618500
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.00012640774517331133,
+      "loss": 1.3011,
+      "step": 619000
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.00012626729836519914,
+      "loss": 1.3076,
+      "step": 619500
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.00012612685155708695,
+      "loss": 1.3343,
+      "step": 620000
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.00012598640474897473,
+      "loss": 1.2962,
+      "step": 620500
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.0001258459579408625,
+      "loss": 1.3236,
+      "step": 621000
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.00012570551113275032,
+      "loss": 1.3082,
+      "step": 621500
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.0001255650643246381,
+      "loss": 1.3086,
+      "step": 622000
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.0001254246175165259,
+      "loss": 1.3151,
+      "step": 622500
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.0001252841707084137,
+      "loss": 1.3091,
+      "step": 623000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0001251437239003015,
+      "loss": 1.3105,
+      "step": 623500
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.00012500327709218928,
+      "loss": 1.3166,
+      "step": 624000
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.00012486283028407706,
+      "loss": 1.3014,
+      "step": 624500
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.00012472238347596487,
+      "loss": 1.3021,
+      "step": 625000
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.00012458193666785265,
+      "loss": 1.3102,
+      "step": 625500
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.00012444148985974046,
+      "loss": 1.3234,
+      "step": 626000
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.00012430104305162824,
+      "loss": 1.336,
+      "step": 626500
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.00012416059624351604,
+      "loss": 1.315,
+      "step": 627000
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.00012402014943540382,
+      "loss": 1.3038,
+      "step": 627500
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0001238797026272916,
+      "loss": 1.3132,
+      "step": 628000
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.00012373925581917941,
+      "loss": 1.3037,
+      "step": 628500
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.0001235988090110672,
+      "loss": 1.3109,
+      "step": 629000
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.000123458362202955,
+      "loss": 1.3006,
+      "step": 629500
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.00012331791539484278,
+      "loss": 1.3587,
+      "step": 630000
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.00012317746858673056,
+      "loss": 1.3146,
+      "step": 630500
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.00012303702177861837,
+      "loss": 1.3138,
+      "step": 631000
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.00012289657497050615,
+      "loss": 1.2895,
+      "step": 631500
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.00012275612816239393,
+      "loss": 1.3425,
+      "step": 632000
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.00012261568135428174,
+      "loss": 1.3281,
+      "step": 632500
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.00012247523454616955,
+      "loss": 1.326,
+      "step": 633000
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.00012233478773805733,
+      "loss": 1.3228,
+      "step": 633500
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0001221943409299451,
+      "loss": 1.3185,
+      "step": 634000
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0001220538941218329,
+      "loss": 1.349,
+      "step": 634500
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0001219134473137207,
+      "loss": 1.3335,
+      "step": 635000
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.0001217730005056085,
+      "loss": 1.3331,
+      "step": 635500
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0001216325536974963,
+      "loss": 1.3346,
+      "step": 636000
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.00012149210688938408,
+      "loss": 1.3286,
+      "step": 636500
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.00012135166008127188,
+      "loss": 1.3264,
+      "step": 637000
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.00012121121327315967,
+      "loss": 1.3262,
+      "step": 637500
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.00012107076646504745,
+      "loss": 1.333,
+      "step": 638000
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.00012093031965693525,
+      "loss": 1.3542,
+      "step": 638500
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.00012078987284882304,
+      "loss": 1.3194,
+      "step": 639000
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.00012064942604071085,
+      "loss": 1.3137,
+      "step": 639500
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.00012050897923259863,
+      "loss": 1.3317,
+      "step": 640000
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.00012036853242448643,
+      "loss": 1.3195,
+      "step": 640500
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.00012022808561637422,
+      "loss": 1.3313,
+      "step": 641000
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.000120087638808262,
+      "loss": 1.3185,
+      "step": 641500
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.0001199471920001498,
+      "loss": 1.2436,
+      "step": 642000
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.00011980674519203758,
+      "loss": 1.2658,
+      "step": 642500
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.00011966629838392539,
+      "loss": 1.2675,
+      "step": 643000
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.00011952585157581318,
+      "loss": 1.2203,
+      "step": 643500
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.00011938540476770097,
+      "loss": 1.2593,
+      "step": 644000
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.00011924495795958876,
+      "loss": 1.2784,
+      "step": 644500
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.00011910451115147655,
+      "loss": 1.2633,
+      "step": 645000
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.00011896406434336434,
+      "loss": 1.2395,
+      "step": 645500
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.00011882361753525213,
+      "loss": 1.2544,
+      "step": 646000
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.00011868317072713993,
+      "loss": 1.2599,
+      "step": 646500
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.00011854272391902773,
+      "loss": 1.2445,
+      "step": 647000
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.00011840227711091552,
+      "loss": 1.2267,
+      "step": 647500
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.0001182618303028033,
+      "loss": 1.2634,
+      "step": 648000
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.0001181213834946911,
+      "loss": 1.2239,
+      "step": 648500
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.00011798093668657889,
+      "loss": 1.2746,
+      "step": 649000
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.00011784048987846669,
+      "loss": 1.2392,
+      "step": 649500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.00011770004307035448,
+      "loss": 1.2407,
+      "step": 650000
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.00011755959626224228,
+      "loss": 1.2554,
+      "step": 650500
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.00011741914945413006,
+      "loss": 1.2624,
+      "step": 651000
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.00011727870264601785,
+      "loss": 1.2498,
+      "step": 651500
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.00011713825583790565,
+      "loss": 1.2597,
+      "step": 652000
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.00011699780902979343,
+      "loss": 1.278,
+      "step": 652500
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.00011685736222168123,
+      "loss": 1.2565,
+      "step": 653000
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.00011671691541356903,
+      "loss": 1.2496,
+      "step": 653500
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.00011657646860545682,
+      "loss": 1.2582,
+      "step": 654000
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0001164360217973446,
+      "loss": 1.2733,
+      "step": 654500
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0001162955749892324,
+      "loss": 1.2515,
+      "step": 655000
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.00011615512818112019,
+      "loss": 1.2596,
+      "step": 655500
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.00011601468137300797,
+      "loss": 1.2828,
+      "step": 656000
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.00011587423456489578,
+      "loss": 1.2634,
+      "step": 656500
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.00011573378775678358,
+      "loss": 1.2501,
+      "step": 657000
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.00011559334094867137,
+      "loss": 1.2474,
+      "step": 657500
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.00011545289414055915,
+      "loss": 1.2746,
+      "step": 658000
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.00011531244733244695,
+      "loss": 1.2862,
+      "step": 658500
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.00011517200052433474,
+      "loss": 1.2696,
+      "step": 659000
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.00011503155371622252,
+      "loss": 1.2729,
+      "step": 659500
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.00011489110690811033,
+      "loss": 1.255,
+      "step": 660000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.00011475066009999812,
+      "loss": 1.2736,
+      "step": 660500
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0001146102132918859,
+      "loss": 1.2411,
+      "step": 661000
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.0001144697664837737,
+      "loss": 1.2661,
+      "step": 661500
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.0001143293196756615,
+      "loss": 1.2637,
+      "step": 662000
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.00011418887286754927,
+      "loss": 1.2766,
+      "step": 662500
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.00011404842605943708,
+      "loss": 1.2677,
+      "step": 663000
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.00011390797925132488,
+      "loss": 1.2798,
+      "step": 663500
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.00011376753244321267,
+      "loss": 1.2858,
+      "step": 664000
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.00011362708563510045,
+      "loss": 1.2779,
+      "step": 664500
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.00011348663882698825,
+      "loss": 1.2734,
+      "step": 665000
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.00011334619201887604,
+      "loss": 1.2503,
+      "step": 665500
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.00011320574521076382,
+      "loss": 1.2903,
+      "step": 666000
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.00011306529840265163,
+      "loss": 1.2798,
+      "step": 666500
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.00011292485159453943,
+      "loss": 1.2993,
+      "step": 667000
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0001127844047864272,
+      "loss": 1.259,
+      "step": 667500
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.000112643957978315,
+      "loss": 1.2531,
+      "step": 668000
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.0001125035111702028,
+      "loss": 1.2713,
+      "step": 668500
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.00011236306436209058,
+      "loss": 1.2919,
+      "step": 669000
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.00011222261755397837,
+      "loss": 1.2812,
+      "step": 669500
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.00011208217074586618,
+      "loss": 1.2909,
+      "step": 670000
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.00011194172393775397,
+      "loss": 1.2807,
+      "step": 670500
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.00011180127712964175,
+      "loss": 1.2726,
+      "step": 671000
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.00011166083032152955,
+      "loss": 1.2782,
+      "step": 671500
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.00011152038351341734,
+      "loss": 1.2554,
+      "step": 672000
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.00011137993670530512,
+      "loss": 1.2748,
+      "step": 672500
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.00011123948989719293,
+      "loss": 1.2875,
+      "step": 673000
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.00011109904308908073,
+      "loss": 1.2593,
+      "step": 673500
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.00011095859628096852,
+      "loss": 1.3059,
+      "step": 674000
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0001108181494728563,
+      "loss": 1.2666,
+      "step": 674500
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.0001106777026647441,
+      "loss": 1.2965,
+      "step": 675000
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.00011053725585663189,
+      "loss": 1.2656,
+      "step": 675500
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.00011039680904851967,
+      "loss": 1.2737,
+      "step": 676000
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.00011025636224040748,
+      "loss": 1.3124,
+      "step": 676500
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.00011011591543229527,
+      "loss": 1.2881,
+      "step": 677000
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.00010997546862418305,
+      "loss": 1.2242,
+      "step": 677500
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.00010983502181607085,
+      "loss": 1.2085,
+      "step": 678000
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.00010969457500795864,
+      "loss": 1.2096,
+      "step": 678500
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.00010955412819984642,
+      "loss": 1.175,
+      "step": 679000
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.00010941368139173422,
+      "loss": 1.2158,
+      "step": 679500
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.00010927323458362203,
+      "loss": 1.192,
+      "step": 680000
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.00010913278777550982,
+      "loss": 1.211,
+      "step": 680500
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.0001089923409673976,
+      "loss": 1.1971,
+      "step": 681000
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0001088518941592854,
+      "loss": 1.2256,
+      "step": 681500
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.00010871144735117319,
+      "loss": 1.2046,
+      "step": 682000
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.00010857100054306097,
+      "loss": 1.1963,
+      "step": 682500
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.00010843055373494877,
+      "loss": 1.1934,
+      "step": 683000
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 0.00010829010692683657,
+      "loss": 1.2209,
+      "step": 683500
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 0.00010814966011872437,
+      "loss": 1.1987,
+      "step": 684000
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 0.00010800921331061215,
+      "loss": 1.1989,
+      "step": 684500
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 0.00010786876650249994,
+      "loss": 1.2023,
+      "step": 685000
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 0.00010772831969438774,
+      "loss": 1.2116,
+      "step": 685500
+    },
+    {
+      "epoch": 19.25,
+      "learning_rate": 0.00010758787288627552,
+      "loss": 1.2047,
+      "step": 686000
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 0.00010744742607816333,
+      "loss": 1.2377,
+      "step": 686500
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 0.00010730697927005112,
+      "loss": 1.189,
+      "step": 687000
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 0.0001071665324619389,
+      "loss": 1.2157,
+      "step": 687500
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 0.0001070260856538267,
+      "loss": 1.2216,
+      "step": 688000
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 0.00010688563884571449,
+      "loss": 1.177,
+      "step": 688500
+    },
+    {
+      "epoch": 19.34,
+      "learning_rate": 0.00010674519203760227,
+      "loss": 1.2182,
+      "step": 689000
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 0.00010660474522949007,
+      "loss": 1.1988,
+      "step": 689500
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 0.00010646429842137788,
+      "loss": 1.2144,
+      "step": 690000
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 0.00010632385161326567,
+      "loss": 1.2286,
+      "step": 690500
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 0.00010618340480515345,
+      "loss": 1.2298,
+      "step": 691000
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 0.00010604295799704125,
+      "loss": 1.1789,
+      "step": 691500
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 0.00010590251118892904,
+      "loss": 1.2358,
+      "step": 692000
+    },
+    {
+      "epoch": 19.43,
+      "learning_rate": 0.00010576206438081682,
+      "loss": 1.2112,
+      "step": 692500
+    },
+    {
+      "epoch": 19.45,
+      "learning_rate": 0.00010562161757270462,
+      "loss": 1.2054,
+      "step": 693000
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 0.00010548117076459242,
+      "loss": 1.2202,
+      "step": 693500
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 0.0001053407239564802,
+      "loss": 1.2374,
+      "step": 694000
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 0.000105200277148368,
+      "loss": 1.2075,
+      "step": 694500
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 0.0001050598303402558,
+      "loss": 1.2234,
+      "step": 695000
+    },
+    {
+      "epoch": 19.52,
+      "learning_rate": 0.00010491938353214357,
+      "loss": 1.2231,
+      "step": 695500
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 0.00010477893672403137,
+      "loss": 1.2446,
+      "step": 696000
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 0.00010463848991591916,
+      "loss": 1.2135,
+      "step": 696500
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 0.00010449804310780697,
+      "loss": 1.2248,
+      "step": 697000
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 0.00010435759629969475,
+      "loss": 1.2104,
+      "step": 697500
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 0.00010421714949158255,
+      "loss": 1.2013,
+      "step": 698000
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 0.00010407670268347034,
+      "loss": 1.2264,
+      "step": 698500
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 0.00010393625587535812,
+      "loss": 1.2195,
+      "step": 699000
+    },
+    {
+      "epoch": 19.63,
+      "learning_rate": 0.00010379580906724592,
+      "loss": 1.2195,
+      "step": 699500
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 0.00010365536225913372,
+      "loss": 1.2198,
+      "step": 700000
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 0.00010351491545102152,
+      "loss": 1.1913,
+      "step": 700500
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 0.0001033744686429093,
+      "loss": 1.2231,
+      "step": 701000
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 0.0001032340218347971,
+      "loss": 1.2456,
+      "step": 701500
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 0.00010309357502668489,
+      "loss": 1.2426,
+      "step": 702000
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 0.00010295312821857267,
+      "loss": 1.2245,
+      "step": 702500
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 0.00010281268141046046,
+      "loss": 1.28,
+      "step": 703000
+    },
+    {
+      "epoch": 19.74,
+      "learning_rate": 0.00010267223460234827,
+      "loss": 1.253,
+      "step": 703500
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 0.00010253178779423605,
+      "loss": 1.2213,
+      "step": 704000
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 0.00010239134098612385,
+      "loss": 1.2145,
+      "step": 704500
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 0.00010225089417801164,
+      "loss": 1.2306,
+      "step": 705000
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 0.00010211044736989942,
+      "loss": 1.241,
+      "step": 705500
+    },
+    {
+      "epoch": 19.81,
+      "learning_rate": 0.00010197000056178722,
+      "loss": 1.2447,
+      "step": 706000
+    },
+    {
+      "epoch": 19.83,
+      "learning_rate": 0.00010182955375367501,
+      "loss": 1.2606,
+      "step": 706500
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 0.00010168910694556282,
+      "loss": 1.2593,
+      "step": 707000
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 0.0001015486601374506,
+      "loss": 1.2424,
+      "step": 707500
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 0.0001014082133293384,
+      "loss": 1.2421,
+      "step": 708000
+    },
+    {
+      "epoch": 19.88,
+      "learning_rate": 0.00010126776652122619,
+      "loss": 1.2328,
+      "step": 708500
+    },
+    {
+      "epoch": 19.9,
+      "learning_rate": 0.00010112731971311397,
+      "loss": 1.2323,
+      "step": 709000
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 0.00010098687290500177,
+      "loss": 1.2401,
+      "step": 709500
+    },
+    {
+      "epoch": 19.92,
+      "learning_rate": 0.00010084642609688956,
+      "loss": 1.2343,
+      "step": 710000
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 0.00010070597928877737,
+      "loss": 1.1958,
+      "step": 710500
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 0.00010056553248066515,
+      "loss": 1.2349,
+      "step": 711000
+    },
+    {
+      "epoch": 19.97,
+      "learning_rate": 0.00010042508567255294,
+      "loss": 1.2477,
+      "step": 711500
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 0.00010028463886444074,
+      "loss": 1.2248,
+      "step": 712000
+    },
+    {
+      "epoch": 19.99,
+      "learning_rate": 0.00010014419205632852,
+      "loss": 1.2459,
+      "step": 712500
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 0.00010000374524821631,
+      "loss": 1.1723,
+      "step": 713000
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 9.986329844010412e-05,
+      "loss": 1.1634,
+      "step": 713500
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 9.97228516319919e-05,
+      "loss": 1.1365,
+      "step": 714000
+    },
+    {
+      "epoch": 20.05,
+      "learning_rate": 9.95824048238797e-05,
+      "loss": 1.1556,
+      "step": 714500
+    },
+    {
+      "epoch": 20.06,
+      "learning_rate": 9.944195801576749e-05,
+      "loss": 1.179,
+      "step": 715000
+    },
+    {
+      "epoch": 20.08,
+      "learning_rate": 9.930151120765527e-05,
+      "loss": 1.1533,
+      "step": 715500
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 9.916106439954307e-05,
+      "loss": 1.1557,
+      "step": 716000
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 9.902061759143086e-05,
+      "loss": 1.1516,
+      "step": 716500
+    },
+    {
+      "epoch": 20.12,
+      "learning_rate": 9.888017078331867e-05,
+      "loss": 1.1769,
+      "step": 717000
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 9.873972397520645e-05,
+      "loss": 1.1663,
+      "step": 717500
+    },
+    {
+      "epoch": 20.15,
+      "learning_rate": 9.859927716709424e-05,
+      "loss": 1.1744,
+      "step": 718000
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 9.845883035898204e-05,
+      "loss": 1.1447,
+      "step": 718500
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 9.831838355086982e-05,
+      "loss": 1.1439,
+      "step": 719000
+    },
+    {
+      "epoch": 20.19,
+      "learning_rate": 9.817793674275761e-05,
+      "loss": 1.1868,
+      "step": 719500
+    },
+    {
+      "epoch": 20.21,
+      "learning_rate": 9.803748993464541e-05,
+      "loss": 1.181,
+      "step": 720000
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 9.78970431265332e-05,
+      "loss": 1.1764,
+      "step": 720500
+    },
+    {
+      "epoch": 20.23,
+      "learning_rate": 9.7756596318421e-05,
+      "loss": 1.1794,
+      "step": 721000
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 9.761614951030879e-05,
+      "loss": 1.1583,
+      "step": 721500
+    },
+    {
+      "epoch": 20.26,
+      "learning_rate": 9.747570270219657e-05,
+      "loss": 1.174,
+      "step": 722000
+    },
+    {
+      "epoch": 20.28,
+      "learning_rate": 9.733525589408437e-05,
+      "loss": 1.1712,
+      "step": 722500
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 9.719480908597216e-05,
+      "loss": 1.1656,
+      "step": 723000
+    },
+    {
+      "epoch": 20.3,
+      "learning_rate": 9.705436227785997e-05,
+      "loss": 1.1593,
+      "step": 723500
+    },
+    {
+      "epoch": 20.32,
+      "learning_rate": 9.691391546974775e-05,
+      "loss": 1.1696,
+      "step": 724000
+    },
+    {
+      "epoch": 20.33,
+      "learning_rate": 9.677346866163554e-05,
+      "loss": 1.1739,
+      "step": 724500
+    },
+    {
+      "epoch": 20.35,
+      "learning_rate": 9.663302185352334e-05,
+      "loss": 1.1932,
+      "step": 725000
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 9.649257504541112e-05,
+      "loss": 1.1506,
+      "step": 725500
+    },
+    {
+      "epoch": 20.37,
+      "learning_rate": 9.635212823729891e-05,
+      "loss": 1.1521,
+      "step": 726000
+    },
+    {
+      "epoch": 20.39,
+      "learning_rate": 9.621168142918671e-05,
+      "loss": 1.177,
+      "step": 726500
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 9.607123462107452e-05,
+      "loss": 1.1874,
+      "step": 727000
+    },
+    {
+      "epoch": 20.42,
+      "learning_rate": 9.59307878129623e-05,
+      "loss": 1.1543,
+      "step": 727500
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 9.579034100485009e-05,
+      "loss": 1.1798,
+      "step": 728000
+    },
+    {
+      "epoch": 20.44,
+      "learning_rate": 9.564989419673789e-05,
+      "loss": 1.1665,
+      "step": 728500
+    },
+    {
+      "epoch": 20.46,
+      "learning_rate": 9.550944738862567e-05,
+      "loss": 1.1723,
+      "step": 729000
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 9.536900058051346e-05,
+      "loss": 1.1711,
+      "step": 729500
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 9.522855377240126e-05,
+      "loss": 1.1574,
+      "step": 730000
+    },
+    {
+      "epoch": 20.5,
+      "learning_rate": 9.508810696428905e-05,
+      "loss": 1.2109,
+      "step": 730500
+    },
+    {
+      "epoch": 20.51,
+      "learning_rate": 9.494766015617685e-05,
+      "loss": 1.1871,
+      "step": 731000
+    },
+    {
+      "epoch": 20.53,
+      "learning_rate": 9.480721334806464e-05,
+      "loss": 1.1752,
+      "step": 731500
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 9.466676653995242e-05,
+      "loss": 1.1446,
+      "step": 732000
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 9.452631973184022e-05,
+      "loss": 1.2237,
+      "step": 732500
+    },
+    {
+      "epoch": 20.57,
+      "learning_rate": 9.438587292372801e-05,
+      "loss": 1.1726,
+      "step": 733000
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 9.424542611561579e-05,
+      "loss": 1.1836,
+      "step": 733500
+    },
+    {
+      "epoch": 20.6,
+      "learning_rate": 9.41049793075036e-05,
+      "loss": 1.1865,
+      "step": 734000
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 9.39645324993914e-05,
+      "loss": 1.1597,
+      "step": 734500
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 9.382408569127919e-05,
+      "loss": 1.1935,
+      "step": 735000
+    },
+    {
+      "epoch": 20.64,
+      "learning_rate": 9.368363888316697e-05,
+      "loss": 1.1654,
+      "step": 735500
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 9.354319207505476e-05,
+      "loss": 1.1634,
+      "step": 736000
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 9.340274526694256e-05,
+      "loss": 1.1966,
+      "step": 736500
+    },
+    {
+      "epoch": 20.68,
+      "learning_rate": 9.326229845883035e-05,
+      "loss": 1.1762,
+      "step": 737000
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 9.312185165071815e-05,
+      "loss": 1.1922,
+      "step": 737500
+    },
+    {
+      "epoch": 20.71,
+      "learning_rate": 9.298140484260594e-05,
+      "loss": 1.183,
+      "step": 738000
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 9.284095803449372e-05,
+      "loss": 1.171,
+      "step": 738500
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 9.270051122638152e-05,
+      "loss": 1.1857,
+      "step": 739000
+    },
+    {
+      "epoch": 20.75,
+      "learning_rate": 9.256006441826931e-05,
+      "loss": 1.1882,
+      "step": 739500
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 9.241961761015709e-05,
+      "loss": 1.1782,
+      "step": 740000
+    },
+    {
+      "epoch": 20.78,
+      "learning_rate": 9.22791708020449e-05,
+      "loss": 1.1772,
+      "step": 740500
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 9.21387239939327e-05,
+      "loss": 1.1686,
+      "step": 741000
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 9.199827718582049e-05,
+      "loss": 1.1965,
+      "step": 741500
+    },
+    {
+      "epoch": 20.82,
+      "learning_rate": 9.185783037770827e-05,
+      "loss": 1.1715,
+      "step": 742000
+    },
+    {
+      "epoch": 20.84,
+      "learning_rate": 9.171738356959606e-05,
+      "loss": 1.1701,
+      "step": 742500
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 9.157693676148386e-05,
+      "loss": 1.1616,
+      "step": 743000
+    },
+    {
+      "epoch": 20.86,
+      "learning_rate": 9.143648995337164e-05,
+      "loss": 1.2022,
+      "step": 743500
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 9.129604314525945e-05,
+      "loss": 1.1882,
+      "step": 744000
+    },
+    {
+      "epoch": 20.89,
+      "learning_rate": 9.115559633714724e-05,
+      "loss": 1.1873,
+      "step": 744500
+    },
+    {
+      "epoch": 20.91,
+      "learning_rate": 9.101514952903504e-05,
+      "loss": 1.1965,
+      "step": 745000
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 9.087470272092282e-05,
+      "loss": 1.2093,
+      "step": 745500
+    },
+    {
+      "epoch": 20.93,
+      "learning_rate": 9.073425591281061e-05,
+      "loss": 1.1754,
+      "step": 746000
+    },
+    {
+      "epoch": 20.95,
+      "learning_rate": 9.05938091046984e-05,
+      "loss": 1.1968,
+      "step": 746500
+    },
+    {
+      "epoch": 20.96,
+      "learning_rate": 9.045336229658619e-05,
+      "loss": 1.1894,
+      "step": 747000
+    },
+    {
+      "epoch": 20.98,
+      "learning_rate": 9.0312915488474e-05,
+      "loss": 1.1938,
+      "step": 747500
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 9.017246868036179e-05,
+      "loss": 1.1966,
+      "step": 748000
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 9.003202187224957e-05,
+      "loss": 1.175,
+      "step": 748500
+    },
+    {
+      "epoch": 21.02,
+      "learning_rate": 8.989157506413737e-05,
+      "loss": 1.1007,
+      "step": 749000
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 8.975112825602516e-05,
+      "loss": 1.1129,
+      "step": 749500
+    },
+    {
+      "epoch": 21.05,
+      "learning_rate": 8.961068144791294e-05,
+      "loss": 1.1079,
+      "step": 750000
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 8.947023463980075e-05,
+      "loss": 1.1114,
+      "step": 750500
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 8.932978783168854e-05,
+      "loss": 1.1245,
+      "step": 751000
+    },
+    {
+      "epoch": 21.09,
+      "learning_rate": 8.918934102357634e-05,
+      "loss": 1.1261,
+      "step": 751500
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 8.904889421546412e-05,
+      "loss": 1.1244,
+      "step": 752000
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 8.890844740735191e-05,
+      "loss": 1.1137,
+      "step": 752500
+    },
+    {
+      "epoch": 21.13,
+      "learning_rate": 8.876800059923971e-05,
+      "loss": 1.1279,
+      "step": 753000
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 8.862755379112749e-05,
+      "loss": 1.1199,
+      "step": 753500
+    },
+    {
+      "epoch": 21.16,
+      "learning_rate": 8.84871069830153e-05,
+      "loss": 1.124,
+      "step": 754000
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 8.834666017490309e-05,
+      "loss": 1.0937,
+      "step": 754500
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 8.820621336679089e-05,
+      "loss": 1.134,
+      "step": 755000
+    },
+    {
+      "epoch": 21.2,
+      "learning_rate": 8.806576655867867e-05,
+      "loss": 1.1194,
+      "step": 755500
+    },
+    {
+      "epoch": 21.22,
+      "learning_rate": 8.792531975056646e-05,
+      "loss": 1.1378,
+      "step": 756000
+    },
+    {
+      "epoch": 21.23,
+      "learning_rate": 8.778487294245426e-05,
+      "loss": 1.1242,
+      "step": 756500
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 8.764442613434204e-05,
+      "loss": 1.1631,
+      "step": 757000
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 8.750397932622984e-05,
+      "loss": 1.103,
+      "step": 757500
+    },
+    {
+      "epoch": 21.27,
+      "learning_rate": 8.736353251811764e-05,
+      "loss": 1.1169,
+      "step": 758000
+    },
+    {
+      "epoch": 21.29,
+      "learning_rate": 8.722308571000542e-05,
+      "loss": 1.1532,
+      "step": 758500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 8.708263890189321e-05,
+      "loss": 1.1296,
+      "step": 759000
+    },
+    {
+      "epoch": 21.31,
+      "learning_rate": 8.694219209378101e-05,
+      "loss": 1.1365,
+      "step": 759500
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 8.680174528566879e-05,
+      "loss": 1.1171,
+      "step": 760000
+    },
+    {
+      "epoch": 21.34,
+      "learning_rate": 8.666129847755658e-05,
+      "loss": 1.1402,
+      "step": 760500
+    },
+    {
+      "epoch": 21.36,
+      "learning_rate": 8.652085166944439e-05,
+      "loss": 1.1393,
+      "step": 761000
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 8.638040486133219e-05,
+      "loss": 1.1442,
+      "step": 761500
+    },
+    {
+      "epoch": 21.38,
+      "learning_rate": 8.623995805321997e-05,
+      "loss": 1.143,
+      "step": 762000
+    },
+    {
+      "epoch": 21.4,
+      "learning_rate": 8.609951124510776e-05,
+      "loss": 1.1264,
+      "step": 762500
+    },
+    {
+      "epoch": 21.41,
+      "learning_rate": 8.595906443699556e-05,
+      "loss": 1.1292,
+      "step": 763000
+    },
+    {
+      "epoch": 21.43,
+      "learning_rate": 8.581861762888334e-05,
+      "loss": 1.1087,
+      "step": 763500
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 8.567817082077115e-05,
+      "loss": 1.1085,
+      "step": 764000
+    },
+    {
+      "epoch": 21.45,
+      "learning_rate": 8.553772401265894e-05,
+      "loss": 1.1242,
+      "step": 764500
+    },
+    {
+      "epoch": 21.47,
+      "learning_rate": 8.539727720454672e-05,
+      "loss": 1.1303,
+      "step": 765000
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 8.525683039643451e-05,
+      "loss": 1.099,
+      "step": 765500
+    },
+    {
+      "epoch": 21.5,
+      "learning_rate": 8.511638358832231e-05,
+      "loss": 1.1538,
+      "step": 766000
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 8.497593678021009e-05,
+      "loss": 1.1517,
+      "step": 766500
+    },
+    {
+      "epoch": 21.52,
+      "learning_rate": 8.483548997209788e-05,
+      "loss": 1.124,
+      "step": 767000
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 8.469504316398569e-05,
+      "loss": 1.1305,
+      "step": 767500
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 8.455459635587349e-05,
+      "loss": 1.1466,
+      "step": 768000
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 8.441414954776127e-05,
+      "loss": 1.1146,
+      "step": 768500
+    },
+    {
+      "epoch": 21.58,
+      "learning_rate": 8.427370273964906e-05,
+      "loss": 1.122,
+      "step": 769000
+    },
+    {
+      "epoch": 21.59,
+      "learning_rate": 8.413325593153686e-05,
+      "loss": 1.1454,
+      "step": 769500
+    },
+    {
+      "epoch": 21.61,
+      "learning_rate": 8.399280912342464e-05,
+      "loss": 1.16,
+      "step": 770000
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 8.385236231531243e-05,
+      "loss": 1.1593,
+      "step": 770500
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 8.371191550720024e-05,
+      "loss": 1.1463,
+      "step": 771000
+    },
+    {
+      "epoch": 21.65,
+      "learning_rate": 8.357146869908803e-05,
+      "loss": 1.1309,
+      "step": 771500
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 8.343102189097582e-05,
+      "loss": 1.1496,
+      "step": 772000
+    },
+    {
+      "epoch": 21.68,
+      "learning_rate": 8.329057508286361e-05,
+      "loss": 1.1591,
+      "step": 772500
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 8.31501282747514e-05,
+      "loss": 1.105,
+      "step": 773000
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 8.300968146663919e-05,
+      "loss": 1.1542,
+      "step": 773500
+    },
+    {
+      "epoch": 21.72,
+      "learning_rate": 8.286923465852698e-05,
+      "loss": 1.17,
+      "step": 774000
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 8.272878785041479e-05,
+      "loss": 1.1318,
+      "step": 774500
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 8.258834104230257e-05,
+      "loss": 1.1279,
+      "step": 775000
+    },
+    {
+      "epoch": 21.76,
+      "learning_rate": 8.244789423419036e-05,
+      "loss": 1.136,
+      "step": 775500
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 8.230744742607816e-05,
+      "loss": 1.1214,
+      "step": 776000
+    },
+    {
+      "epoch": 21.79,
+      "learning_rate": 8.216700061796594e-05,
+      "loss": 1.1429,
+      "step": 776500
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 8.202655380985373e-05,
+      "loss": 1.1407,
+      "step": 777000
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 8.188610700174154e-05,
+      "loss": 1.1617,
+      "step": 777500
+    },
+    {
+      "epoch": 21.83,
+      "learning_rate": 8.174566019362934e-05,
+      "loss": 1.1193,
+      "step": 778000
+    },
+    {
+      "epoch": 21.85,
+      "learning_rate": 8.160521338551712e-05,
+      "loss": 1.1362,
+      "step": 778500
+    },
+    {
+      "epoch": 21.86,
+      "learning_rate": 8.146476657740491e-05,
+      "loss": 1.1325,
+      "step": 779000
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 8.13243197692927e-05,
+      "loss": 1.1055,
+      "step": 779500
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 8.118387296118049e-05,
+      "loss": 1.1304,
+      "step": 780000
+    },
+    {
+      "epoch": 21.9,
+      "learning_rate": 8.104342615306828e-05,
+      "loss": 1.1545,
+      "step": 780500
+    },
+    {
+      "epoch": 21.92,
+      "learning_rate": 8.090297934495609e-05,
+      "loss": 1.1518,
+      "step": 781000
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 8.076253253684388e-05,
+      "loss": 1.1682,
+      "step": 781500
+    },
+    {
+      "epoch": 21.95,
+      "learning_rate": 8.062208572873166e-05,
+      "loss": 1.1433,
+      "step": 782000
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 8.048163892061946e-05,
+      "loss": 1.1372,
+      "step": 782500
+    },
+    {
+      "epoch": 21.97,
+      "learning_rate": 8.034119211250724e-05,
+      "loss": 1.1321,
+      "step": 783000
+    },
+    {
+      "epoch": 21.99,
+      "learning_rate": 8.020074530439503e-05,
+      "loss": 1.1505,
+      "step": 783500
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 8.006029849628283e-05,
+      "loss": 1.1292,
+      "step": 784000
+    },
+    {
+      "epoch": 22.02,
+      "learning_rate": 7.991985168817064e-05,
+      "loss": 1.0754,
+      "step": 784500
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 7.977940488005842e-05,
+      "loss": 1.0455,
+      "step": 785000
+    },
+    {
+      "epoch": 22.04,
+      "learning_rate": 7.963895807194621e-05,
+      "loss": 1.0852,
+      "step": 785500
+    },
+    {
+      "epoch": 22.06,
+      "learning_rate": 7.9498511263834e-05,
+      "loss": 1.0497,
+      "step": 786000
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 7.935806445572179e-05,
+      "loss": 1.1157,
+      "step": 786500
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 7.921761764760958e-05,
+      "loss": 1.085,
+      "step": 787000
+    },
+    {
+      "epoch": 22.1,
+      "learning_rate": 7.907717083949739e-05,
+      "loss": 1.0713,
+      "step": 787500
+    },
+    {
+      "epoch": 22.11,
+      "learning_rate": 7.893672403138518e-05,
+      "loss": 1.1042,
+      "step": 788000
+    },
+    {
+      "epoch": 22.13,
+      "learning_rate": 7.879627722327297e-05,
+      "loss": 1.1009,
+      "step": 788500
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 7.865583041516076e-05,
+      "loss": 1.0994,
+      "step": 789000
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 7.851538360704855e-05,
+      "loss": 1.081,
+      "step": 789500
+    },
+    {
+      "epoch": 22.17,
+      "learning_rate": 7.837493679893634e-05,
+      "loss": 1.0888,
+      "step": 790000
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 7.823448999082413e-05,
+      "loss": 1.1001,
+      "step": 790500
+    },
+    {
+      "epoch": 22.2,
+      "learning_rate": 7.809404318271194e-05,
+      "loss": 1.0861,
+      "step": 791000
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 7.795359637459972e-05,
+      "loss": 1.0728,
+      "step": 791500
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 7.781314956648751e-05,
+      "loss": 1.0877,
+      "step": 792000
+    },
+    {
+      "epoch": 22.24,
+      "learning_rate": 7.767270275837531e-05,
+      "loss": 1.1109,
+      "step": 792500
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 7.753225595026309e-05,
+      "loss": 1.0825,
+      "step": 793000
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 7.739180914215088e-05,
+      "loss": 1.0772,
+      "step": 793500
+    },
+    {
+      "epoch": 22.28,
+      "learning_rate": 7.725136233403868e-05,
+      "loss": 1.114,
+      "step": 794000
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 7.711091552592649e-05,
+      "loss": 1.0956,
+      "step": 794500
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 7.697046871781427e-05,
+      "loss": 1.0918,
+      "step": 795000
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 7.683002190970206e-05,
+      "loss": 1.099,
+      "step": 795500
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 7.668957510158986e-05,
+      "loss": 1.0841,
+      "step": 796000
+    },
+    {
+      "epoch": 22.35,
+      "learning_rate": 7.654912829347764e-05,
+      "loss": 1.0937,
+      "step": 796500
+    },
+    {
+      "epoch": 22.37,
+      "learning_rate": 7.640868148536543e-05,
+      "loss": 1.0962,
+      "step": 797000
+    },
+    {
+      "epoch": 22.38,
+      "learning_rate": 7.626823467725323e-05,
+      "loss": 1.1158,
+      "step": 797500
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 7.612778786914103e-05,
+      "loss": 1.1044,
+      "step": 798000
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 7.598734106102881e-05,
+      "loss": 1.0872,
+      "step": 798500
+    },
+    {
+      "epoch": 22.42,
+      "learning_rate": 7.584689425291661e-05,
+      "loss": 1.1139,
+      "step": 799000
+    },
+    {
+      "epoch": 22.44,
+      "learning_rate": 7.57064474448044e-05,
+      "loss": 1.0982,
+      "step": 799500
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 7.556600063669218e-05,
+      "loss": 1.0748,
+      "step": 800000
+    },
+    {
+      "epoch": 22.46,
+      "learning_rate": 7.542555382857998e-05,
+      "loss": 1.0869,
+      "step": 800500
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 7.528510702046779e-05,
+      "loss": 1.1141,
+      "step": 801000
+    },
+    {
+      "epoch": 22.49,
+      "learning_rate": 7.514466021235557e-05,
+      "loss": 1.0838,
+      "step": 801500
+    },
+    {
+      "epoch": 22.51,
+      "learning_rate": 7.500421340424336e-05,
+      "loss": 1.0729,
+      "step": 802000
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 7.486376659613116e-05,
+      "loss": 1.0518,
+      "step": 802500
+    },
+    {
+      "epoch": 22.53,
+      "learning_rate": 7.472331978801894e-05,
+      "loss": 1.0645,
+      "step": 803000
+    },
+    {
+      "epoch": 22.55,
+      "learning_rate": 7.458287297990675e-05,
+      "loss": 1.0993,
+      "step": 803500
+    },
+    {
+      "epoch": 22.56,
+      "learning_rate": 7.444242617179453e-05,
+      "loss": 1.0817,
+      "step": 804000
+    },
+    {
+      "epoch": 22.58,
+      "learning_rate": 7.430197936368232e-05,
+      "loss": 1.1001,
+      "step": 804500
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 7.416153255557012e-05,
+      "loss": 1.0691,
+      "step": 805000
+    },
+    {
+      "epoch": 22.6,
+      "learning_rate": 7.402108574745791e-05,
+      "loss": 1.0951,
+      "step": 805500
+    },
+    {
+      "epoch": 22.62,
+      "learning_rate": 7.38806389393457e-05,
+      "loss": 1.1136,
+      "step": 806000
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 7.374019213123348e-05,
+      "loss": 1.0895,
+      "step": 806500
+    },
+    {
+      "epoch": 22.65,
+      "learning_rate": 7.359974532312129e-05,
+      "loss": 1.0736,
+      "step": 807000
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 7.345929851500907e-05,
+      "loss": 1.1013,
+      "step": 807500
+    },
+    {
+      "epoch": 22.67,
+      "learning_rate": 7.331885170689687e-05,
+      "loss": 1.07,
+      "step": 808000
+    },
+    {
+      "epoch": 22.69,
+      "learning_rate": 7.317840489878466e-05,
+      "loss": 1.0795,
+      "step": 808500
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 7.303795809067246e-05,
+      "loss": 1.0871,
+      "step": 809000
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 7.289751128256024e-05,
+      "loss": 1.074,
+      "step": 809500
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 7.275706447444803e-05,
+      "loss": 1.0948,
+      "step": 810000
+    },
+    {
+      "epoch": 22.74,
+      "learning_rate": 7.261661766633583e-05,
+      "loss": 1.1008,
+      "step": 810500
+    },
+    {
+      "epoch": 22.76,
+      "learning_rate": 7.247617085822362e-05,
+      "loss": 1.1233,
+      "step": 811000
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 7.233572405011142e-05,
+      "loss": 1.0846,
+      "step": 811500
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 7.219527724199921e-05,
+      "loss": 1.1133,
+      "step": 812000
+    },
+    {
+      "epoch": 22.8,
+      "learning_rate": 7.2054830433887e-05,
+      "loss": 1.1052,
+      "step": 812500
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 7.191438362577479e-05,
+      "loss": 1.0933,
+      "step": 813000
+    },
+    {
+      "epoch": 22.83,
+      "learning_rate": 7.177393681766258e-05,
+      "loss": 1.0879,
+      "step": 813500
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 7.163349000955037e-05,
+      "loss": 1.0886,
+      "step": 814000
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 7.149304320143817e-05,
+      "loss": 1.0757,
+      "step": 814500
+    },
+    {
+      "epoch": 22.87,
+      "learning_rate": 7.135259639332596e-05,
+      "loss": 1.0979,
+      "step": 815000
+    },
+    {
+      "epoch": 22.89,
+      "learning_rate": 7.121214958521376e-05,
+      "loss": 1.1325,
+      "step": 815500
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 7.107170277710155e-05,
+      "loss": 1.1131,
+      "step": 816000
+    },
+    {
+      "epoch": 22.91,
+      "learning_rate": 7.093125596898933e-05,
+      "loss": 1.107,
+      "step": 816500
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 7.079080916087714e-05,
+      "loss": 1.0754,
+      "step": 817000
+    },
+    {
+      "epoch": 22.94,
+      "learning_rate": 7.065036235276492e-05,
+      "loss": 1.1159,
+      "step": 817500
+    },
+    {
+      "epoch": 22.96,
+      "learning_rate": 7.050991554465272e-05,
+      "loss": 1.1028,
+      "step": 818000
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 7.03694687365405e-05,
+      "loss": 1.1204,
+      "step": 818500
+    },
+    {
+      "epoch": 22.98,
+      "learning_rate": 7.02290219284283e-05,
+      "loss": 1.0885,
+      "step": 819000
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 7.008857512031609e-05,
+      "loss": 1.0953,
+      "step": 819500
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 6.994812831220388e-05,
+      "loss": 1.056,
+      "step": 820000
+    },
+    {
+      "epoch": 23.03,
+      "learning_rate": 6.980768150409168e-05,
+      "loss": 1.031,
+      "step": 820500
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 6.966723469597947e-05,
+      "loss": 1.0594,
+      "step": 821000
+    },
+    {
+      "epoch": 23.05,
+      "learning_rate": 6.952678788786726e-05,
+      "loss": 0.9976,
+      "step": 821500
+    },
+    {
+      "epoch": 23.07,
+      "learning_rate": 6.938634107975505e-05,
+      "loss": 1.0471,
+      "step": 822000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 6.924589427164285e-05,
+      "loss": 1.0296,
+      "step": 822500
+    },
+    {
+      "epoch": 23.1,
+      "learning_rate": 6.910544746353063e-05,
+      "loss": 1.0404,
+      "step": 823000
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 6.896500065541843e-05,
+      "loss": 1.0662,
+      "step": 823500
+    },
+    {
+      "epoch": 23.12,
+      "learning_rate": 6.882455384730622e-05,
+      "loss": 1.0507,
+      "step": 824000
+    },
+    {
+      "epoch": 23.14,
+      "learning_rate": 6.868410703919402e-05,
+      "loss": 1.0617,
+      "step": 824500
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 6.854366023108181e-05,
+      "loss": 1.0324,
+      "step": 825000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 6.840321342296961e-05,
+      "loss": 1.0515,
+      "step": 825500
+    },
+    {
+      "epoch": 23.18,
+      "learning_rate": 6.82627666148574e-05,
+      "loss": 1.051,
+      "step": 826000
+    },
+    {
+      "epoch": 23.19,
+      "learning_rate": 6.812231980674518e-05,
+      "loss": 1.0416,
+      "step": 826500
+    },
+    {
+      "epoch": 23.21,
+      "learning_rate": 6.798187299863298e-05,
+      "loss": 1.0278,
+      "step": 827000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 6.784142619052077e-05,
+      "loss": 1.0625,
+      "step": 827500
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 6.770097938240857e-05,
+      "loss": 1.0301,
+      "step": 828000
+    },
+    {
+      "epoch": 23.25,
+      "learning_rate": 6.756053257429635e-05,
+      "loss": 1.0468,
+      "step": 828500
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 6.742008576618415e-05,
+      "loss": 1.0368,
+      "step": 829000
+    },
+    {
+      "epoch": 23.28,
+      "learning_rate": 6.727963895807194e-05,
+      "loss": 1.0442,
+      "step": 829500
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 6.713919214995973e-05,
+      "loss": 1.039,
+      "step": 830000
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 6.699874534184752e-05,
+      "loss": 1.0589,
+      "step": 830500
+    },
+    {
+      "epoch": 23.32,
+      "learning_rate": 6.685829853373532e-05,
+      "loss": 1.0398,
+      "step": 831000
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 6.671785172562311e-05,
+      "loss": 1.0442,
+      "step": 831500
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 6.65774049175109e-05,
+      "loss": 1.0707,
+      "step": 832000
+    },
+    {
+      "epoch": 23.36,
+      "learning_rate": 6.64369581093987e-05,
+      "loss": 1.0625,
+      "step": 832500
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 6.629651130128648e-05,
+      "loss": 1.0501,
+      "step": 833000
+    },
+    {
+      "epoch": 23.39,
+      "learning_rate": 6.615606449317428e-05,
+      "loss": 1.084,
+      "step": 833500
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 6.601561768506207e-05,
+      "loss": 1.0412,
+      "step": 834000
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 6.587517087694987e-05,
+      "loss": 1.0433,
+      "step": 834500
+    },
+    {
+      "epoch": 23.43,
+      "learning_rate": 6.573472406883766e-05,
+      "loss": 1.0554,
+      "step": 835000
+    },
+    {
+      "epoch": 23.45,
+      "learning_rate": 6.559427726072546e-05,
+      "loss": 1.0543,
+      "step": 835500
+    },
+    {
+      "epoch": 23.46,
+      "learning_rate": 6.545383045261324e-05,
+      "loss": 1.0653,
+      "step": 836000
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 6.531338364450103e-05,
+      "loss": 1.0521,
+      "step": 836500
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 6.517293683638883e-05,
+      "loss": 1.0521,
+      "step": 837000
+    },
+    {
+      "epoch": 23.5,
+      "learning_rate": 6.503249002827662e-05,
+      "loss": 1.0654,
+      "step": 837500
+    },
+    {
+      "epoch": 23.52,
+      "learning_rate": 6.489204322016441e-05,
+      "loss": 1.0661,
+      "step": 838000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 6.47515964120522e-05,
+      "loss": 1.0238,
+      "step": 838500
+    },
+    {
+      "epoch": 23.54,
+      "learning_rate": 6.461114960394e-05,
+      "loss": 1.0643,
+      "step": 839000
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 6.447070279582778e-05,
+      "loss": 1.0537,
+      "step": 839500
+    },
+    {
+      "epoch": 23.57,
+      "learning_rate": 6.433025598771558e-05,
+      "loss": 1.0452,
+      "step": 840000
+    },
+    {
+      "epoch": 23.59,
+      "learning_rate": 6.418980917960337e-05,
+      "loss": 1.0785,
+      "step": 840500
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 6.404936237149117e-05,
+      "loss": 1.0739,
+      "step": 841000
+    },
+    {
+      "epoch": 23.61,
+      "learning_rate": 6.390891556337896e-05,
+      "loss": 1.0694,
+      "step": 841500
+    },
+    {
+      "epoch": 23.63,
+      "learning_rate": 6.376846875526674e-05,
+      "loss": 1.0649,
+      "step": 842000
+    },
+    {
+      "epoch": 23.64,
+      "learning_rate": 6.362802194715455e-05,
+      "loss": 1.0585,
+      "step": 842500
+    },
+    {
+      "epoch": 23.66,
+      "learning_rate": 6.348757513904233e-05,
+      "loss": 1.0764,
+      "step": 843000
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 6.334712833093013e-05,
+      "loss": 1.0815,
+      "step": 843500
+    },
+    {
+      "epoch": 23.68,
+      "learning_rate": 6.320668152281792e-05,
+      "loss": 1.0748,
+      "step": 844000
+    },
+    {
+      "epoch": 23.7,
+      "learning_rate": 6.306623471470572e-05,
+      "loss": 1.0502,
+      "step": 844500
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 6.29257879065935e-05,
+      "loss": 1.053,
+      "step": 845000
+    },
+    {
+      "epoch": 23.73,
+      "learning_rate": 6.278534109848129e-05,
+      "loss": 1.0697,
+      "step": 845500
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 6.264489429036909e-05,
+      "loss": 1.0811,
+      "step": 846000
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 6.250444748225688e-05,
+      "loss": 1.0412,
+      "step": 846500
+    },
+    {
+      "epoch": 23.77,
+      "learning_rate": 6.236400067414467e-05,
+      "loss": 1.0698,
+      "step": 847000
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 6.222355386603247e-05,
+      "loss": 1.0591,
+      "step": 847500
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 6.208310705792026e-05,
+      "loss": 1.0589,
+      "step": 848000
+    },
+    {
+      "epoch": 23.81,
+      "learning_rate": 6.194266024980804e-05,
+      "loss": 1.0547,
+      "step": 848500
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 6.180221344169585e-05,
+      "loss": 1.0766,
+      "step": 849000
+    },
+    {
+      "epoch": 23.84,
+      "learning_rate": 6.166176663358363e-05,
+      "loss": 1.0689,
+      "step": 849500
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 6.152131982547143e-05,
+      "loss": 1.0286,
+      "step": 850000
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 6.138087301735922e-05,
+      "loss": 1.0654,
+      "step": 850500
+    },
+    {
+      "epoch": 23.88,
+      "learning_rate": 6.124042620924702e-05,
+      "loss": 1.0763,
+      "step": 851000
+    },
+    {
+      "epoch": 23.9,
+      "learning_rate": 6.109997940113481e-05,
+      "loss": 1.0731,
+      "step": 851500
+    },
+    {
+      "epoch": 23.91,
+      "learning_rate": 6.095953259302259e-05,
+      "loss": 1.0472,
+      "step": 852000
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 6.081908578491039e-05,
+      "loss": 1.0806,
+      "step": 852500
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 6.067863897679818e-05,
+      "loss": 1.0616,
+      "step": 853000
+    },
+    {
+      "epoch": 23.95,
+      "learning_rate": 6.0538192168685975e-05,
+      "loss": 1.0509,
+      "step": 853500
+    },
+    {
+      "epoch": 23.97,
+      "learning_rate": 6.039774536057376e-05,
+      "loss": 1.0808,
+      "step": 854000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 6.0257298552461564e-05,
+      "loss": 1.0587,
+      "step": 854500
+    },
+    {
+      "epoch": 23.99,
+      "learning_rate": 6.011685174434935e-05,
+      "loss": 1.0787,
+      "step": 855000
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 5.997640493623714e-05,
+      "loss": 1.0337,
+      "step": 855500
+    },
+    {
+      "epoch": 24.02,
+      "learning_rate": 5.983595812812494e-05,
+      "loss": 1.0066,
+      "step": 856000
+    },
+    {
+      "epoch": 24.04,
+      "learning_rate": 5.969551132001273e-05,
+      "loss": 1.0121,
+      "step": 856500
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 5.9555064511900516e-05,
+      "loss": 1.0268,
+      "step": 857000
+    },
+    {
+      "epoch": 24.06,
+      "learning_rate": 5.941461770378832e-05,
+      "loss": 1.0055,
+      "step": 857500
+    },
+    {
+      "epoch": 24.08,
+      "learning_rate": 5.9274170895676105e-05,
+      "loss": 1.0119,
+      "step": 858000
+    },
+    {
+      "epoch": 24.09,
+      "learning_rate": 5.91337240875639e-05,
+      "loss": 1.0203,
+      "step": 858500
+    },
+    {
+      "epoch": 24.11,
+      "learning_rate": 5.899327727945169e-05,
+      "loss": 1.0067,
+      "step": 859000
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 5.885283047133948e-05,
+      "loss": 1.0304,
+      "step": 859500
+    },
+    {
+      "epoch": 24.13,
+      "learning_rate": 5.8712383663227276e-05,
+      "loss": 1.0314,
+      "step": 860000
+    },
+    {
+      "epoch": 24.15,
+      "learning_rate": 5.8571936855115064e-05,
+      "loss": 1.002,
+      "step": 860500
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 5.8431490047002865e-05,
+      "loss": 0.9908,
+      "step": 861000
+    },
+    {
+      "epoch": 24.18,
+      "learning_rate": 5.829104323889065e-05,
+      "loss": 1.0277,
+      "step": 861500
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 5.815059643077844e-05,
+      "loss": 1.0328,
+      "step": 862000
+    },
+    {
+      "epoch": 24.2,
+      "learning_rate": 5.801014962266624e-05,
+      "loss": 1.0052,
+      "step": 862500
+    },
+    {
+      "epoch": 24.22,
+      "learning_rate": 5.786970281455403e-05,
+      "loss": 0.9812,
+      "step": 863000
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 5.7729256006441824e-05,
+      "loss": 0.9925,
+      "step": 863500
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 5.758880919832961e-05,
+      "loss": 1.0073,
+      "step": 864000
+    },
+    {
+      "epoch": 24.26,
+      "learning_rate": 5.7448362390217406e-05,
+      "loss": 0.9962,
+      "step": 864500
+    },
+    {
+      "epoch": 24.27,
+      "learning_rate": 5.73079155821052e-05,
+      "loss": 0.9965,
+      "step": 865000
+    },
+    {
+      "epoch": 24.29,
+      "learning_rate": 5.716746877399299e-05,
+      "loss": 1.0207,
+      "step": 865500
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 5.702702196588079e-05,
+      "loss": 1.037,
+      "step": 866000
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 5.688657515776858e-05,
+      "loss": 1.031,
+      "step": 866500
+    },
+    {
+      "epoch": 24.33,
+      "learning_rate": 5.6746128349656365e-05,
+      "loss": 1.0097,
+      "step": 867000
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 5.6605681541544166e-05,
+      "loss": 1.0315,
+      "step": 867500
+    },
+    {
+      "epoch": 24.36,
+      "learning_rate": 5.6465234733431954e-05,
+      "loss": 1.0114,
+      "step": 868000
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 5.632478792531975e-05,
+      "loss": 1.0141,
+      "step": 868500
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 5.6184341117207536e-05,
+      "loss": 1.003,
+      "step": 869000
+    },
+    {
+      "epoch": 24.4,
+      "learning_rate": 5.604389430909533e-05,
+      "loss": 0.999,
+      "step": 869500
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 5.5903447500983125e-05,
+      "loss": 1.026,
+      "step": 870000
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 5.576300069287091e-05,
+      "loss": 0.9967,
+      "step": 870500
+    },
+    {
+      "epoch": 24.44,
+      "learning_rate": 5.5622553884758714e-05,
+      "loss": 1.0358,
+      "step": 871000
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 5.54821070766465e-05,
+      "loss": 1.0213,
+      "step": 871500
+    },
+    {
+      "epoch": 24.47,
+      "learning_rate": 5.534166026853429e-05,
+      "loss": 1.0055,
+      "step": 872000
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 5.5201213460422084e-05,
+      "loss": 1.0138,
+      "step": 872500
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 5.506076665230988e-05,
+      "loss": 0.9862,
+      "step": 873000
+    },
+    {
+      "epoch": 24.51,
+      "learning_rate": 5.4920319844197666e-05,
+      "loss": 1.0179,
+      "step": 873500
+    },
+    {
+      "epoch": 24.53,
+      "learning_rate": 5.477987303608546e-05,
+      "loss": 1.0121,
+      "step": 874000
+    },
+    {
+      "epoch": 24.54,
+      "learning_rate": 5.4639426227973255e-05,
+      "loss": 0.9966,
+      "step": 874500
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 5.449897941986105e-05,
+      "loss": 1.0033,
+      "step": 875000
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 5.435853261174884e-05,
+      "loss": 1.0233,
+      "step": 875500
+    },
+    {
+      "epoch": 24.58,
+      "learning_rate": 5.421808580363664e-05,
+      "loss": 1.0081,
+      "step": 876000
+    },
+    {
+      "epoch": 24.6,
+      "learning_rate": 5.4077638995524426e-05,
+      "loss": 1.0259,
+      "step": 876500
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 5.3937192187412213e-05,
+      "loss": 1.0035,
+      "step": 877000
+    },
+    {
+      "epoch": 24.63,
+      "learning_rate": 5.379674537930001e-05,
+      "loss": 1.0148,
+      "step": 877500
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 5.36562985711878e-05,
+      "loss": 1.0296,
+      "step": 878000
+    },
+    {
+      "epoch": 24.65,
+      "learning_rate": 5.351585176307559e-05,
+      "loss": 1.0188,
+      "step": 878500
+    },
+    {
+      "epoch": 24.67,
+      "learning_rate": 5.3375404954963385e-05,
+      "loss": 1.035,
+      "step": 879000
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 5.323495814685118e-05,
+      "loss": 1.0216,
+      "step": 879500
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 5.3094511338738973e-05,
+      "loss": 1.0217,
+      "step": 880000
+    },
+    {
+      "epoch": 24.71,
+      "learning_rate": 5.295406453062676e-05,
+      "loss": 1.0079,
+      "step": 880500
+    },
+    {
+      "epoch": 24.72,
+      "learning_rate": 5.281361772251456e-05,
+      "loss": 0.9966,
+      "step": 881000
+    },
+    {
+      "epoch": 24.74,
+      "learning_rate": 5.267317091440235e-05,
+      "loss": 1.0,
+      "step": 881500
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 5.253272410629014e-05,
+      "loss": 1.0185,
+      "step": 882000
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 5.2392277298177925e-05,
+      "loss": 1.0232,
+      "step": 882500
+    },
+    {
+      "epoch": 24.78,
+      "learning_rate": 5.225183049006573e-05,
+      "loss": 1.0223,
+      "step": 883000
+    },
+    {
+      "epoch": 24.79,
+      "learning_rate": 5.2111383681953514e-05,
+      "loss": 1.0066,
+      "step": 883500
+    },
+    {
+      "epoch": 24.81,
+      "learning_rate": 5.197093687384131e-05,
+      "loss": 1.0333,
+      "step": 884000
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 5.18304900657291e-05,
+      "loss": 1.0348,
+      "step": 884500
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 5.16900432576169e-05,
+      "loss": 1.0362,
+      "step": 885000
+    },
+    {
+      "epoch": 24.85,
+      "learning_rate": 5.1549596449504685e-05,
+      "loss": 1.0099,
+      "step": 885500
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 5.140914964139247e-05,
+      "loss": 1.0146,
+      "step": 886000
+    },
+    {
+      "epoch": 24.88,
+      "learning_rate": 5.1268702833280274e-05,
+      "loss": 1.0462,
+      "step": 886500
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 5.112825602516806e-05,
+      "loss": 1.0164,
+      "step": 887000
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 5.098780921705585e-05,
+      "loss": 1.0112,
+      "step": 887500
+    },
+    {
+      "epoch": 24.92,
+      "learning_rate": 5.084736240894365e-05,
+      "loss": 1.0161,
+      "step": 888000
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 5.070691560083144e-05,
+      "loss": 1.0426,
+      "step": 888500
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 5.056646879271923e-05,
+      "loss": 0.9918,
+      "step": 889000
+    },
+    {
+      "epoch": 24.96,
+      "learning_rate": 5.042602198460703e-05,
+      "loss": 1.023,
+      "step": 889500
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 5.028557517649482e-05,
+      "loss": 1.0051,
+      "step": 890000
+    },
+    {
+      "epoch": 24.99,
+      "learning_rate": 5.014512836838261e-05,
+      "loss": 1.0501,
+      "step": 890500
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 5.00046815602704e-05,
+      "loss": 1.0258,
+      "step": 891000
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 4.98642347521582e-05,
+      "loss": 0.9781,
+      "step": 891500
+    },
+    {
+      "epoch": 25.03,
+      "learning_rate": 4.9723787944045986e-05,
+      "loss": 0.9605,
+      "step": 892000
+    },
+    {
+      "epoch": 25.05,
+      "learning_rate": 4.9583341135933774e-05,
+      "loss": 0.9655,
+      "step": 892500
+    },
+    {
+      "epoch": 25.06,
+      "learning_rate": 4.9442894327821575e-05,
+      "loss": 0.964,
+      "step": 893000
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 4.930244751970936e-05,
+      "loss": 0.9666,
+      "step": 893500
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 4.916200071159716e-05,
+      "loss": 0.9809,
+      "step": 894000
+    },
+    {
+      "epoch": 25.1,
+      "learning_rate": 4.902155390348495e-05,
+      "loss": 0.9729,
+      "step": 894500
+    },
+    {
+      "epoch": 25.12,
+      "learning_rate": 4.888110709537274e-05,
+      "loss": 0.9895,
+      "step": 895000
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 4.8740660287260534e-05,
+      "loss": 0.9598,
+      "step": 895500
+    },
+    {
+      "epoch": 25.14,
+      "learning_rate": 4.860021347914832e-05,
+      "loss": 0.95,
+      "step": 896000
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 4.845976667103612e-05,
+      "loss": 0.9933,
+      "step": 896500
+    },
+    {
+      "epoch": 25.17,
+      "learning_rate": 4.831931986292391e-05,
+      "loss": 0.9686,
+      "step": 897000
+    },
+    {
+      "epoch": 25.19,
+      "learning_rate": 4.81788730548117e-05,
+      "loss": 0.9934,
+      "step": 897500
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 4.80384262466995e-05,
+      "loss": 0.9792,
+      "step": 898000
+    },
+    {
+      "epoch": 25.21,
+      "learning_rate": 4.789797943858729e-05,
+      "loss": 0.9899,
+      "step": 898500
+    },
+    {
+      "epoch": 25.23,
+      "learning_rate": 4.775753263047508e-05,
+      "loss": 0.9809,
+      "step": 899000
+    },
+    {
+      "epoch": 25.24,
+      "learning_rate": 4.7617085822362876e-05,
+      "loss": 0.9835,
+      "step": 899500
+    },
+    {
+      "epoch": 25.26,
+      "learning_rate": 4.7476639014250664e-05,
+      "loss": 0.9338,
+      "step": 900000
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 4.733619220613846e-05,
+      "loss": 0.9526,
+      "step": 900500
+    },
+    {
+      "epoch": 25.28,
+      "learning_rate": 4.7195745398026246e-05,
+      "loss": 0.9714,
+      "step": 901000
+    },
+    {
+      "epoch": 25.3,
+      "learning_rate": 4.705529858991405e-05,
+      "loss": 0.9842,
+      "step": 901500
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 4.6914851781801835e-05,
+      "loss": 0.979,
+      "step": 902000
+    },
+    {
+      "epoch": 25.33,
+      "learning_rate": 4.677440497368962e-05,
+      "loss": 0.9746,
+      "step": 902500
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 4.6633958165577424e-05,
+      "loss": 0.9716,
+      "step": 903000
+    },
+    {
+      "epoch": 25.35,
+      "learning_rate": 4.649351135746521e-05,
+      "loss": 0.9827,
+      "step": 903500
+    },
+    {
+      "epoch": 25.37,
+      "learning_rate": 4.6353064549353006e-05,
+      "loss": 0.994,
+      "step": 904000
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 4.6212617741240794e-05,
+      "loss": 0.981,
+      "step": 904500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 4.607217093312859e-05,
+      "loss": 0.9842,
+      "step": 905000
+    },
+    {
+      "epoch": 25.41,
+      "learning_rate": 4.593172412501638e-05,
+      "loss": 0.979,
+      "step": 905500
+    },
+    {
+      "epoch": 25.42,
+      "learning_rate": 4.579127731690417e-05,
+      "loss": 0.9942,
+      "step": 906000
+    },
+    {
+      "epoch": 25.44,
+      "learning_rate": 4.565083050879197e-05,
+      "loss": 0.9978,
+      "step": 906500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 4.551038370067976e-05,
+      "loss": 1.0006,
+      "step": 907000
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 4.536993689256755e-05,
+      "loss": 1.0112,
+      "step": 907500
+    },
+    {
+      "epoch": 25.48,
+      "learning_rate": 4.522949008445535e-05,
+      "loss": 0.9642,
+      "step": 908000
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 4.5089043276343136e-05,
+      "loss": 0.9905,
+      "step": 908500
+    },
+    {
+      "epoch": 25.51,
+      "learning_rate": 4.4948596468230924e-05,
+      "loss": 0.986,
+      "step": 909000
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 4.480814966011872e-05,
+      "loss": 0.9829,
+      "step": 909500
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 4.466770285200651e-05,
+      "loss": 0.9783,
+      "step": 910000
+    },
+    {
+      "epoch": 25.55,
+      "learning_rate": 4.452725604389431e-05,
+      "loss": 0.983,
+      "step": 910500
+    },
+    {
+      "epoch": 25.57,
+      "learning_rate": 4.4386809235782095e-05,
+      "loss": 0.972,
+      "step": 911000
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 4.4246362427669896e-05,
+      "loss": 0.9854,
+      "step": 911500
+    },
+    {
+      "epoch": 25.59,
+      "learning_rate": 4.4105915619557684e-05,
+      "loss": 0.981,
+      "step": 912000
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 4.396546881144547e-05,
+      "loss": 0.9816,
+      "step": 912500
+    },
+    {
+      "epoch": 25.62,
+      "learning_rate": 4.382502200333327e-05,
+      "loss": 0.9745,
+      "step": 913000
+    },
+    {
+      "epoch": 25.64,
+      "learning_rate": 4.368457519522106e-05,
+      "loss": 0.9589,
+      "step": 913500
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 4.354412838710885e-05,
+      "loss": 0.9854,
+      "step": 914000
+    },
+    {
+      "epoch": 25.66,
+      "learning_rate": 4.340368157899664e-05,
+      "loss": 0.9811,
+      "step": 914500
+    },
+    {
+      "epoch": 25.68,
+      "learning_rate": 4.326323477088444e-05,
+      "loss": 0.9934,
+      "step": 915000
+    },
+    {
+      "epoch": 25.69,
+      "learning_rate": 4.312278796277223e-05,
+      "loss": 0.9691,
+      "step": 915500
+    },
+    {
+      "epoch": 25.71,
+      "learning_rate": 4.298234115466002e-05,
+      "loss": 0.9757,
+      "step": 916000
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 4.284189434654782e-05,
+      "loss": 0.98,
+      "step": 916500
+    },
+    {
+      "epoch": 25.73,
+      "learning_rate": 4.270144753843561e-05,
+      "loss": 0.9761,
+      "step": 917000
+    },
+    {
+      "epoch": 25.75,
+      "learning_rate": 4.2561000730323396e-05,
+      "loss": 0.9808,
+      "step": 917500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 4.2420553922211184e-05,
+      "loss": 0.9752,
+      "step": 918000
+    },
+    {
+      "epoch": 25.78,
+      "learning_rate": 4.2280107114098985e-05,
+      "loss": 1.0108,
+      "step": 918500
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 4.213966030598677e-05,
+      "loss": 0.9959,
+      "step": 919000
+    },
+    {
+      "epoch": 25.8,
+      "learning_rate": 4.199921349787457e-05,
+      "loss": 0.9715,
+      "step": 919500
+    },
+    {
+      "epoch": 25.82,
+      "learning_rate": 4.185876668976236e-05,
+      "loss": 0.9801,
+      "step": 920000
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 4.1718319881650156e-05,
+      "loss": 0.9798,
+      "step": 920500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 4.1577873073537944e-05,
+      "loss": 1.0103,
+      "step": 921000
+    },
+    {
+      "epoch": 25.86,
+      "learning_rate": 4.143742626542574e-05,
+      "loss": 0.982,
+      "step": 921500
+    },
+    {
+      "epoch": 25.87,
+      "learning_rate": 4.129697945731353e-05,
+      "loss": 0.9832,
+      "step": 922000
+    },
+    {
+      "epoch": 25.89,
+      "learning_rate": 4.115653264920132e-05,
+      "loss": 0.9876,
+      "step": 922500
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 4.101608584108911e-05,
+      "loss": 0.9796,
+      "step": 923000
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 4.087563903297691e-05,
+      "loss": 0.988,
+      "step": 923500
+    },
+    {
+      "epoch": 25.93,
+      "learning_rate": 4.07351922248647e-05,
+      "loss": 0.9728,
+      "step": 924000
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 4.059474541675249e-05,
+      "loss": 0.9854,
+      "step": 924500
+    },
+    {
+      "epoch": 25.96,
+      "learning_rate": 4.0454298608640286e-05,
+      "loss": 0.9798,
+      "step": 925000
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 4.031385180052808e-05,
+      "loss": 0.9965,
+      "step": 925500
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 4.017340499241587e-05,
+      "loss": 0.978,
+      "step": 926000
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.003295818430366e-05,
+      "loss": 0.978,
+      "step": 926500
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 3.989251137619146e-05,
+      "loss": 0.9435,
+      "step": 927000
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 3.9752064568079244e-05,
+      "loss": 0.9353,
+      "step": 927500
+    },
+    {
+      "epoch": 26.04,
+      "learning_rate": 3.961161775996703e-05,
+      "loss": 0.9308,
+      "step": 928000
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 3.9471170951854833e-05,
+      "loss": 0.9444,
+      "step": 928500
+    },
+    {
+      "epoch": 26.07,
+      "learning_rate": 3.933072414374262e-05,
+      "loss": 0.9623,
+      "step": 929000
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 3.9190277335630416e-05,
+      "loss": 0.9253,
+      "step": 929500
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 3.904983052751821e-05,
+      "loss": 0.9352,
+      "step": 930000
+    },
+    {
+      "epoch": 26.11,
+      "learning_rate": 3.8909383719406e-05,
+      "loss": 0.9643,
+      "step": 930500
+    },
+    {
+      "epoch": 26.13,
+      "learning_rate": 3.876893691129379e-05,
+      "loss": 0.9441,
+      "step": 931000
+    },
+    {
+      "epoch": 26.14,
+      "learning_rate": 3.862849010318159e-05,
+      "loss": 0.9533,
+      "step": 931500
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 3.848804329506938e-05,
+      "loss": 0.9373,
+      "step": 932000
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 3.834759648695717e-05,
+      "loss": 0.9354,
+      "step": 932500
+    },
+    {
+      "epoch": 26.18,
+      "learning_rate": 3.8207149678844957e-05,
+      "loss": 0.9522,
+      "step": 933000
+    },
+    {
+      "epoch": 26.2,
+      "learning_rate": 3.806670287073276e-05,
+      "loss": 0.9477,
+      "step": 933500
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 3.7926256062620545e-05,
+      "loss": 0.9361,
+      "step": 934000
+    },
+    {
+      "epoch": 26.22,
+      "learning_rate": 3.778580925450834e-05,
+      "loss": 0.9419,
+      "step": 934500
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 3.7645362446396134e-05,
+      "loss": 0.968,
+      "step": 935000
+    },
+    {
+      "epoch": 26.25,
+      "learning_rate": 3.750491563828392e-05,
+      "loss": 0.9233,
+      "step": 935500
+    },
+    {
+      "epoch": 26.27,
+      "learning_rate": 3.7364468830171717e-05,
+      "loss": 0.9514,
+      "step": 936000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 3.722402202205951e-05,
+      "loss": 0.9475,
+      "step": 936500
+    },
+    {
+      "epoch": 26.29,
+      "learning_rate": 3.7083575213947305e-05,
+      "loss": 0.9377,
+      "step": 937000
+    },
+    {
+      "epoch": 26.31,
+      "learning_rate": 3.694312840583509e-05,
+      "loss": 0.9704,
+      "step": 937500
+    },
+    {
+      "epoch": 26.32,
+      "learning_rate": 3.680268159772289e-05,
+      "loss": 0.925,
+      "step": 938000
+    },
+    {
+      "epoch": 26.34,
+      "learning_rate": 3.6662234789610675e-05,
+      "loss": 0.9333,
+      "step": 938500
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 3.652178798149847e-05,
+      "loss": 0.9438,
+      "step": 939000
+    },
+    {
+      "epoch": 26.36,
+      "learning_rate": 3.6381341173386264e-05,
+      "loss": 0.9884,
+      "step": 939500
+    },
+    {
+      "epoch": 26.38,
+      "learning_rate": 3.624089436527405e-05,
+      "loss": 0.9539,
+      "step": 940000
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 3.6100447557161846e-05,
+      "loss": 0.9565,
+      "step": 940500
+    },
+    {
+      "epoch": 26.41,
+      "learning_rate": 3.596000074904964e-05,
+      "loss": 0.9604,
+      "step": 941000
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 3.5819553940937435e-05,
+      "loss": 0.9424,
+      "step": 941500
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 3.567910713282522e-05,
+      "loss": 0.9411,
+      "step": 942000
+    },
+    {
+      "epoch": 26.45,
+      "learning_rate": 3.553866032471302e-05,
+      "loss": 0.963,
+      "step": 942500
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 3.539821351660081e-05,
+      "loss": 0.9385,
+      "step": 943000
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 3.52577667084886e-05,
+      "loss": 0.9585,
+      "step": 943500
+    },
+    {
+      "epoch": 26.49,
+      "learning_rate": 3.5117319900376394e-05,
+      "loss": 0.9489,
+      "step": 944000
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 3.497687309226418e-05,
+      "loss": 0.9572,
+      "step": 944500
+    },
+    {
+      "epoch": 26.52,
+      "learning_rate": 3.4836426284151976e-05,
+      "loss": 0.9138,
+      "step": 945000
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 3.469597947603977e-05,
+      "loss": 0.9485,
+      "step": 945500
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 3.4555532667927565e-05,
+      "loss": 0.9519,
+      "step": 946000
+    },
+    {
+      "epoch": 26.56,
+      "learning_rate": 3.441508585981536e-05,
+      "loss": 0.9356,
+      "step": 946500
+    },
+    {
+      "epoch": 26.58,
+      "learning_rate": 3.427463905170315e-05,
+      "loss": 0.9257,
+      "step": 947000
+    },
+    {
+      "epoch": 26.59,
+      "learning_rate": 3.413419224359094e-05,
+      "loss": 0.9313,
+      "step": 947500
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 3.3993745435478736e-05,
+      "loss": 0.9591,
+      "step": 948000
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 3.3853298627366524e-05,
+      "loss": 0.9424,
+      "step": 948500
+    },
+    {
+      "epoch": 26.63,
+      "learning_rate": 3.371285181925432e-05,
+      "loss": 0.9564,
+      "step": 949000
+    },
+    {
+      "epoch": 26.65,
+      "learning_rate": 3.3572405011142106e-05,
+      "loss": 0.9548,
+      "step": 949500
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 3.34319582030299e-05,
+      "loss": 0.9363,
+      "step": 950000
+    },
+    {
+      "epoch": 26.67,
+      "learning_rate": 3.3291511394917695e-05,
+      "loss": 0.9518,
+      "step": 950500
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 3.315106458680549e-05,
+      "loss": 0.9612,
+      "step": 951000
+    },
+    {
+      "epoch": 26.7,
+      "learning_rate": 3.3010617778693284e-05,
+      "loss": 0.953,
+      "step": 951500
+    },
+    {
+      "epoch": 26.72,
+      "learning_rate": 3.287017097058107e-05,
+      "loss": 0.9675,
+      "step": 952000
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 3.2729724162468866e-05,
+      "loss": 0.9442,
+      "step": 952500
+    },
+    {
+      "epoch": 26.74,
+      "learning_rate": 3.258927735435666e-05,
+      "loss": 0.946,
+      "step": 953000
+    },
+    {
+      "epoch": 26.76,
+      "learning_rate": 3.244883054624445e-05,
+      "loss": 0.9465,
+      "step": 953500
+    },
+    {
+      "epoch": 26.77,
+      "learning_rate": 3.230838373813224e-05,
+      "loss": 0.949,
+      "step": 954000
+    },
+    {
+      "epoch": 26.79,
+      "learning_rate": 3.216793693002003e-05,
+      "loss": 0.9622,
+      "step": 954500
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 3.2027490121907825e-05,
+      "loss": 0.9415,
+      "step": 955000
+    },
+    {
+      "epoch": 26.81,
+      "learning_rate": 3.188704331379562e-05,
+      "loss": 0.9294,
+      "step": 955500
+    },
+    {
+      "epoch": 26.83,
+      "learning_rate": 3.1746596505683414e-05,
+      "loss": 0.9346,
+      "step": 956000
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 3.160614969757121e-05,
+      "loss": 0.9471,
+      "step": 956500
+    },
+    {
+      "epoch": 26.86,
+      "learning_rate": 3.1465702889458996e-05,
+      "loss": 0.9375,
+      "step": 957000
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 3.132525608134679e-05,
+      "loss": 0.9421,
+      "step": 957500
+    },
+    {
+      "epoch": 26.88,
+      "learning_rate": 3.118480927323458e-05,
+      "loss": 0.9358,
+      "step": 958000
+    },
+    {
+      "epoch": 26.9,
+      "learning_rate": 3.104436246512237e-05,
+      "loss": 0.9515,
+      "step": 958500
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 3.090391565701017e-05,
+      "loss": 0.9672,
+      "step": 959000
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 3.0763468848897955e-05,
+      "loss": 0.9548,
+      "step": 959500
+    },
+    {
+      "epoch": 26.94,
+      "learning_rate": 3.062302204078575e-05,
+      "loss": 0.9608,
+      "step": 960000
+    },
+    {
+      "epoch": 26.95,
+      "learning_rate": 3.048257523267354e-05,
+      "loss": 0.938,
+      "step": 960500
+    },
+    {
+      "epoch": 26.97,
+      "learning_rate": 3.0342128424561335e-05,
+      "loss": 0.9375,
+      "step": 961000
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 3.020168161644913e-05,
+      "loss": 0.926,
+      "step": 961500
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 3.006123480833692e-05,
+      "loss": 0.9601,
+      "step": 962000
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 2.992078800022471e-05,
+      "loss": 0.906,
+      "step": 962500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 2.9780341192112503e-05,
+      "loss": 0.9176,
+      "step": 963000
+    },
+    {
+      "epoch": 27.04,
+      "learning_rate": 2.9639894384000297e-05,
+      "loss": 0.914,
+      "step": 963500
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 2.949944757588809e-05,
+      "loss": 0.8998,
+      "step": 964000
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 2.9359000767775883e-05,
+      "loss": 0.9115,
+      "step": 964500
+    },
+    {
+      "epoch": 27.08,
+      "learning_rate": 2.9218553959663674e-05,
+      "loss": 0.9167,
+      "step": 965000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 2.9078107151551465e-05,
+      "loss": 0.9245,
+      "step": 965500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 2.893766034343926e-05,
+      "loss": 0.9306,
+      "step": 966000
+    },
+    {
+      "epoch": 27.12,
+      "learning_rate": 2.8797213535327054e-05,
+      "loss": 0.8957,
+      "step": 966500
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 2.865676672721484e-05,
+      "loss": 0.907,
+      "step": 967000
+    },
+    {
+      "epoch": 27.15,
+      "learning_rate": 2.8516319919102636e-05,
+      "loss": 0.9172,
+      "step": 967500
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 2.8375873110990427e-05,
+      "loss": 0.9171,
+      "step": 968000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 2.823542630287822e-05,
+      "loss": 0.9068,
+      "step": 968500
+    },
+    {
+      "epoch": 27.19,
+      "learning_rate": 2.8094979494766016e-05,
+      "loss": 0.8993,
+      "step": 969000
+    },
+    {
+      "epoch": 27.21,
+      "learning_rate": 2.7954532686653804e-05,
+      "loss": 0.9355,
+      "step": 969500
+    },
+    {
+      "epoch": 27.22,
+      "learning_rate": 2.7814085878541598e-05,
+      "loss": 0.9208,
+      "step": 970000
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 2.767363907042939e-05,
+      "loss": 0.9365,
+      "step": 970500
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 2.7533192262317184e-05,
+      "loss": 0.9097,
+      "step": 971000
+    },
+    {
+      "epoch": 27.26,
+      "learning_rate": 2.7392745454204978e-05,
+      "loss": 0.9277,
+      "step": 971500
+    },
+    {
+      "epoch": 27.28,
+      "learning_rate": 2.7252298646092766e-05,
+      "loss": 0.9186,
+      "step": 972000
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 2.711185183798056e-05,
+      "loss": 0.9201,
+      "step": 972500
+    },
+    {
+      "epoch": 27.3,
+      "learning_rate": 2.697140502986835e-05,
+      "loss": 0.9255,
+      "step": 973000
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 2.6830958221756146e-05,
+      "loss": 0.9401,
+      "step": 973500
+    },
+    {
+      "epoch": 27.33,
+      "learning_rate": 2.6690511413643933e-05,
+      "loss": 0.9133,
+      "step": 974000
+    },
+    {
+      "epoch": 27.35,
+      "learning_rate": 2.6550064605531728e-05,
+      "loss": 0.9175,
+      "step": 974500
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 2.6409617797419522e-05,
+      "loss": 0.9265,
+      "step": 975000
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 2.6269170989307313e-05,
+      "loss": 0.938,
+      "step": 975500
+    },
+    {
+      "epoch": 27.39,
+      "learning_rate": 2.6128724181195108e-05,
+      "loss": 0.9301,
+      "step": 976000
+    },
+    {
+      "epoch": 27.4,
+      "learning_rate": 2.5988277373082896e-05,
+      "loss": 0.8971,
+      "step": 976500
+    },
+    {
+      "epoch": 27.42,
+      "learning_rate": 2.584783056497069e-05,
+      "loss": 0.9441,
+      "step": 977000
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 2.5707383756858484e-05,
+      "loss": 0.9362,
+      "step": 977500
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 2.5566936948746276e-05,
+      "loss": 0.9083,
+      "step": 978000
+    },
+    {
+      "epoch": 27.46,
+      "learning_rate": 2.542649014063407e-05,
+      "loss": 0.9082,
+      "step": 978500
+    },
+    {
+      "epoch": 27.47,
+      "learning_rate": 2.5286043332521858e-05,
+      "loss": 0.9421,
+      "step": 979000
+    },
+    {
+      "epoch": 27.49,
+      "learning_rate": 2.5145596524409652e-05,
+      "loss": 0.8992,
+      "step": 979500
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 2.5005149716297447e-05,
+      "loss": 0.9328,
+      "step": 980000
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 2.4864702908185238e-05,
+      "loss": 0.9284,
+      "step": 980500
+    },
+    {
+      "epoch": 27.53,
+      "learning_rate": 2.4724256100073032e-05,
+      "loss": 0.9122,
+      "step": 981000
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 2.458380929196082e-05,
+      "loss": 0.8997,
+      "step": 981500
+    },
+    {
+      "epoch": 27.56,
+      "learning_rate": 2.4443362483848614e-05,
+      "loss": 0.9328,
+      "step": 982000
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 2.430291567573641e-05,
+      "loss": 0.9093,
+      "step": 982500
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 2.41624688676242e-05,
+      "loss": 0.9023,
+      "step": 983000
+    },
+    {
+      "epoch": 27.6,
+      "learning_rate": 2.4022022059511994e-05,
+      "loss": 0.9187,
+      "step": 983500
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 2.3881575251399782e-05,
+      "loss": 0.9128,
+      "step": 984000
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 2.3741128443287577e-05,
+      "loss": 0.9219,
+      "step": 984500
+    },
+    {
+      "epoch": 27.64,
+      "learning_rate": 2.360068163517537e-05,
+      "loss": 0.9225,
+      "step": 985000
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 2.3460234827063162e-05,
+      "loss": 0.9106,
+      "step": 985500
+    },
+    {
+      "epoch": 27.67,
+      "learning_rate": 2.3319788018950957e-05,
+      "loss": 0.9105,
+      "step": 986000
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 2.3179341210838744e-05,
+      "loss": 0.9277,
+      "step": 986500
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 2.303889440272654e-05,
+      "loss": 0.9037,
+      "step": 987000
+    },
+    {
+      "epoch": 27.71,
+      "learning_rate": 2.2898447594614333e-05,
+      "loss": 0.9158,
+      "step": 987500
+    },
+    {
+      "epoch": 27.73,
+      "learning_rate": 2.2758000786502124e-05,
+      "loss": 0.9387,
+      "step": 988000
+    },
+    {
+      "epoch": 27.74,
+      "learning_rate": 2.261755397838992e-05,
+      "loss": 0.9273,
+      "step": 988500
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 2.2477107170277706e-05,
+      "loss": 0.9185,
+      "step": 989000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 2.23366603621655e-05,
+      "loss": 0.91,
+      "step": 989500
+    },
+    {
+      "epoch": 27.78,
+      "learning_rate": 2.2196213554053292e-05,
+      "loss": 0.9192,
+      "step": 990000
+    },
+    {
+      "epoch": 27.8,
+      "learning_rate": 2.2055766745941086e-05,
+      "loss": 0.901,
+      "step": 990500
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 2.1915319937828877e-05,
+      "loss": 0.9072,
+      "step": 991000
+    },
+    {
+      "epoch": 27.82,
+      "learning_rate": 2.177487312971667e-05,
+      "loss": 0.9137,
+      "step": 991500
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 2.1634426321604463e-05,
+      "loss": 0.9501,
+      "step": 992000
+    },
+    {
+      "epoch": 27.85,
+      "learning_rate": 2.1493979513492254e-05,
+      "loss": 0.9243,
+      "step": 992500
+    },
+    {
+      "epoch": 27.87,
+      "learning_rate": 2.135353270538005e-05,
+      "loss": 0.9278,
+      "step": 993000
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 2.121308589726784e-05,
+      "loss": 0.8987,
+      "step": 993500
+    },
+    {
+      "epoch": 27.89,
+      "learning_rate": 2.107263908915563e-05,
+      "loss": 0.9121,
+      "step": 994000
+    },
+    {
+      "epoch": 27.91,
+      "learning_rate": 2.0932192281043425e-05,
+      "loss": 0.9219,
+      "step": 994500
+    },
+    {
+      "epoch": 27.92,
+      "learning_rate": 2.0791745472931216e-05,
+      "loss": 0.921,
+      "step": 995000
+    },
+    {
+      "epoch": 27.94,
+      "learning_rate": 2.0651298664819007e-05,
+      "loss": 0.9194,
+      "step": 995500
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 2.0510851856706802e-05,
+      "loss": 0.9205,
+      "step": 996000
+    },
+    {
+      "epoch": 27.96,
+      "learning_rate": 2.0370405048594593e-05,
+      "loss": 0.9412,
+      "step": 996500
+    },
+    {
+      "epoch": 27.98,
+      "learning_rate": 2.0229958240482387e-05,
+      "loss": 0.9151,
+      "step": 997000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 2.008951143237018e-05,
+      "loss": 0.8942,
+      "step": 997500
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 1.994906462425797e-05,
+      "loss": 0.8877,
+      "step": 998000
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 1.9808617816145764e-05,
+      "loss": 0.9026,
+      "step": 998500
+    },
+    {
+      "epoch": 28.03,
+      "learning_rate": 1.9668171008033555e-05,
+      "loss": 0.8988,
+      "step": 999000
+    },
+    {
+      "epoch": 28.05,
+      "learning_rate": 1.952772419992135e-05,
+      "loss": 0.8773,
+      "step": 999500
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 1.938727739180914e-05,
+      "loss": 0.8875,
+      "step": 1000000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 1.9246830583696932e-05,
+      "loss": 0.8946,
+      "step": 1000500
+    },
+    {
+      "epoch": 28.09,
+      "learning_rate": 1.9106383775584726e-05,
+      "loss": 0.8824,
+      "step": 1001000
+    },
+    {
+      "epoch": 28.1,
+      "learning_rate": 1.8965936967472517e-05,
+      "loss": 0.914,
+      "step": 1001500
+    },
+    {
+      "epoch": 28.12,
+      "learning_rate": 1.8825490159360312e-05,
+      "loss": 0.9184,
+      "step": 1002000
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 1.8685043351248103e-05,
+      "loss": 0.9002,
+      "step": 1002500
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 1.8544596543135894e-05,
+      "loss": 0.8737,
+      "step": 1003000
+    },
+    {
+      "epoch": 28.16,
+      "learning_rate": 1.840414973502369e-05,
+      "loss": 0.8976,
+      "step": 1003500
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 1.826370292691148e-05,
+      "loss": 0.8865,
+      "step": 1004000
+    },
+    {
+      "epoch": 28.19,
+      "learning_rate": 1.812325611879927e-05,
+      "loss": 0.8947,
+      "step": 1004500
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 1.7982809310687065e-05,
+      "loss": 0.8757,
+      "step": 1005000
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 1.7842362502574856e-05,
+      "loss": 0.8997,
+      "step": 1005500
+    },
+    {
+      "epoch": 28.23,
+      "learning_rate": 1.770191569446265e-05,
+      "loss": 0.9052,
+      "step": 1006000
+    },
+    {
+      "epoch": 28.25,
+      "learning_rate": 1.756146888635044e-05,
+      "loss": 0.8833,
+      "step": 1006500
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 1.7421022078238233e-05,
+      "loss": 0.9094,
+      "step": 1007000
+    },
+    {
+      "epoch": 28.27,
+      "learning_rate": 1.7280575270126027e-05,
+      "loss": 0.888,
+      "step": 1007500
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 1.7140128462013818e-05,
+      "loss": 0.8907,
+      "step": 1008000
+    },
+    {
+      "epoch": 28.3,
+      "learning_rate": 1.6999681653901613e-05,
+      "loss": 0.8982,
+      "step": 1008500
+    },
+    {
+      "epoch": 28.32,
+      "learning_rate": 1.6859234845789404e-05,
+      "loss": 0.8964,
+      "step": 1009000
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 1.6718788037677195e-05,
+      "loss": 0.8914,
+      "step": 1009500
+    },
+    {
+      "epoch": 28.34,
+      "learning_rate": 1.6578341229564986e-05,
+      "loss": 0.8984,
+      "step": 1010000
+    },
+    {
+      "epoch": 28.36,
+      "learning_rate": 1.643789442145278e-05,
+      "loss": 0.916,
+      "step": 1010500
+    },
+    {
+      "epoch": 28.37,
+      "learning_rate": 1.629744761334057e-05,
+      "loss": 0.8848,
+      "step": 1011000
+    },
+    {
+      "epoch": 28.39,
+      "learning_rate": 1.6157000805228366e-05,
+      "loss": 0.9012,
+      "step": 1011500
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 1.6016553997116157e-05,
+      "loss": 0.8999,
+      "step": 1012000
+    },
+    {
+      "epoch": 28.41,
+      "learning_rate": 1.5876107189003948e-05,
+      "loss": 0.8859,
+      "step": 1012500
+    },
+    {
+      "epoch": 28.43,
+      "learning_rate": 1.5735660380891743e-05,
+      "loss": 0.8975,
+      "step": 1013000
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 1.5595213572779534e-05,
+      "loss": 0.8983,
+      "step": 1013500
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 1.5454766764667328e-05,
+      "loss": 0.9061,
+      "step": 1014000
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 1.531431995655512e-05,
+      "loss": 0.8942,
+      "step": 1014500
+    },
+    {
+      "epoch": 28.48,
+      "learning_rate": 1.517387314844291e-05,
+      "loss": 0.917,
+      "step": 1015000
+    },
+    {
+      "epoch": 28.5,
+      "learning_rate": 1.5033426340330705e-05,
+      "loss": 0.8868,
+      "step": 1015500
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 1.4892979532218497e-05,
+      "loss": 0.8692,
+      "step": 1016000
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 1.4752532724106289e-05,
+      "loss": 0.8927,
+      "step": 1016500
+    },
+    {
+      "epoch": 28.54,
+      "learning_rate": 1.4612085915994081e-05,
+      "loss": 0.9105,
+      "step": 1017000
+    },
+    {
+      "epoch": 28.55,
+      "learning_rate": 1.4471639107881872e-05,
+      "loss": 0.8974,
+      "step": 1017500
+    },
+    {
+      "epoch": 28.57,
+      "learning_rate": 1.4331192299769665e-05,
+      "loss": 0.9013,
+      "step": 1018000
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 1.419074549165746e-05,
+      "loss": 0.8981,
+      "step": 1018500
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 1.405029868354525e-05,
+      "loss": 0.871,
+      "step": 1019000
+    },
+    {
+      "epoch": 28.61,
+      "learning_rate": 1.3909851875433044e-05,
+      "loss": 0.8917,
+      "step": 1019500
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 1.3769405067320835e-05,
+      "loss": 0.8872,
+      "step": 1020000
+    },
+    {
+      "epoch": 28.64,
+      "learning_rate": 1.3628958259208627e-05,
+      "loss": 0.8958,
+      "step": 1020500
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 1.3488511451096422e-05,
+      "loss": 0.8783,
+      "step": 1021000
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 1.3348064642984213e-05,
+      "loss": 0.8722,
+      "step": 1021500
+    },
+    {
+      "epoch": 28.68,
+      "learning_rate": 1.3207617834872006e-05,
+      "loss": 0.9058,
+      "step": 1022000
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 1.3067171026759797e-05,
+      "loss": 0.9316,
+      "step": 1022500
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 1.292672421864759e-05,
+      "loss": 0.8852,
+      "step": 1023000
+    },
+    {
+      "epoch": 28.72,
+      "learning_rate": 1.2786277410535382e-05,
+      "loss": 0.8857,
+      "step": 1023500
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 1.2645830602423175e-05,
+      "loss": 0.893,
+      "step": 1024000
+    },
+    {
+      "epoch": 28.75,
+      "learning_rate": 1.2505383794310968e-05,
+      "loss": 0.8962,
+      "step": 1024500
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 1.2364936986198759e-05,
+      "loss": 0.8671,
+      "step": 1025000
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 1.2224490178086552e-05,
+      "loss": 0.8975,
+      "step": 1025500
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 1.2084043369974343e-05,
+      "loss": 0.904,
+      "step": 1026000
+    },
+    {
+      "epoch": 28.81,
+      "learning_rate": 1.1943596561862137e-05,
+      "loss": 0.9048,
+      "step": 1026500
+    },
+    {
+      "epoch": 28.82,
+      "learning_rate": 1.1803149753749928e-05,
+      "loss": 0.9158,
+      "step": 1027000
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 1.1662702945637721e-05,
+      "loss": 0.8749,
+      "step": 1027500
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 1.1522256137525514e-05,
+      "loss": 0.8801,
+      "step": 1028000
+    },
+    {
+      "epoch": 28.86,
+      "learning_rate": 1.1381809329413305e-05,
+      "loss": 0.9151,
+      "step": 1028500
+    },
+    {
+      "epoch": 28.88,
+      "learning_rate": 1.12413625213011e-05,
+      "loss": 0.8973,
+      "step": 1029000
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 1.110091571318889e-05,
+      "loss": 0.8863,
+      "step": 1029500
+    },
+    {
+      "epoch": 28.9,
+      "learning_rate": 1.0960468905076683e-05,
+      "loss": 0.8966,
+      "step": 1030000
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 1.0820022096964474e-05,
+      "loss": 0.8778,
+      "step": 1030500
+    },
+    {
+      "epoch": 28.93,
+      "learning_rate": 1.0679575288852267e-05,
+      "loss": 0.8899,
+      "step": 1031000
+    },
+    {
+      "epoch": 28.95,
+      "learning_rate": 1.0539128480740062e-05,
+      "loss": 0.8851,
+      "step": 1031500
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 1.0398681672627853e-05,
+      "loss": 0.8972,
+      "step": 1032000
+    },
+    {
+      "epoch": 28.97,
+      "learning_rate": 1.0258234864515645e-05,
+      "loss": 0.8887,
+      "step": 1032500
+    },
+    {
+      "epoch": 28.99,
+      "learning_rate": 1.0117788056403437e-05,
+      "loss": 0.9048,
+      "step": 1033000
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 9.97734124829123e-06,
+      "loss": 0.8814,
+      "step": 1033500
+    },
+    {
+      "epoch": 29.02,
+      "learning_rate": 9.83689444017902e-06,
+      "loss": 0.878,
+      "step": 1034000
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 9.696447632066815e-06,
+      "loss": 0.8769,
+      "step": 1034500
+    },
+    {
+      "epoch": 29.04,
+      "learning_rate": 9.556000823954608e-06,
+      "loss": 0.8792,
+      "step": 1035000
+    },
+    {
+      "epoch": 29.06,
+      "learning_rate": 9.415554015842399e-06,
+      "loss": 0.8968,
+      "step": 1035500
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 9.275107207730191e-06,
+      "loss": 0.8679,
+      "step": 1036000
+    },
+    {
+      "epoch": 29.09,
+      "learning_rate": 9.134660399617984e-06,
+      "loss": 0.8861,
+      "step": 1036500
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 8.994213591505775e-06,
+      "loss": 0.8787,
+      "step": 1037000
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 8.85376678339357e-06,
+      "loss": 0.8827,
+      "step": 1037500
+    },
+    {
+      "epoch": 29.13,
+      "learning_rate": 8.713319975281361e-06,
+      "loss": 0.8814,
+      "step": 1038000
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 8.572873167169154e-06,
+      "loss": 0.8912,
+      "step": 1038500
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 8.432426359056946e-06,
+      "loss": 0.879,
+      "step": 1039000
+    },
+    {
+      "epoch": 29.17,
+      "learning_rate": 8.291979550944737e-06,
+      "loss": 0.8579,
+      "step": 1039500
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 8.15153274283253e-06,
+      "loss": 0.8612,
+      "step": 1040000
+    },
+    {
+      "epoch": 29.2,
+      "learning_rate": 8.011085934720323e-06,
+      "loss": 0.8734,
+      "step": 1040500
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 7.870639126608116e-06,
+      "loss": 0.8676,
+      "step": 1041000
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 7.730192318495909e-06,
+      "loss": 0.8894,
+      "step": 1041500
+    },
+    {
+      "epoch": 29.24,
+      "learning_rate": 7.5897455103837005e-06,
+      "loss": 0.9045,
+      "step": 1042000
+    },
+    {
+      "epoch": 29.26,
+      "learning_rate": 7.449298702271492e-06,
+      "loss": 0.872,
+      "step": 1042500
+    },
+    {
+      "epoch": 29.27,
+      "learning_rate": 7.308851894159285e-06,
+      "loss": 0.8897,
+      "step": 1043000
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 7.168405086047077e-06,
+      "loss": 0.867,
+      "step": 1043500
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 7.027958277934869e-06,
+      "loss": 0.8667,
+      "step": 1044000
+    },
+    {
+      "epoch": 29.31,
+      "learning_rate": 6.887511469822662e-06,
+      "loss": 0.8703,
+      "step": 1044500
+    },
+    {
+      "epoch": 29.33,
+      "learning_rate": 6.747064661710454e-06,
+      "loss": 0.8975,
+      "step": 1045000
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 6.606617853598247e-06,
+      "loss": 0.8692,
+      "step": 1045500
+    },
+    {
+      "epoch": 29.35,
+      "learning_rate": 6.466171045486039e-06,
+      "loss": 0.8732,
+      "step": 1046000
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 6.325724237373831e-06,
+      "loss": 0.8776,
+      "step": 1046500
+    },
+    {
+      "epoch": 29.38,
+      "learning_rate": 6.185277429261624e-06,
+      "loss": 0.8845,
+      "step": 1047000
+    },
+    {
+      "epoch": 29.4,
+      "learning_rate": 6.044830621149416e-06,
+      "loss": 0.8785,
+      "step": 1047500
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 5.904383813037208e-06,
+      "loss": 0.8749,
+      "step": 1048000
+    },
+    {
+      "epoch": 29.42,
+      "learning_rate": 5.7639370049250015e-06,
+      "loss": 0.8644,
+      "step": 1048500
+    },
+    {
+      "epoch": 29.44,
+      "learning_rate": 5.623490196812793e-06,
+      "loss": 0.8678,
+      "step": 1049000
+    },
+    {
+      "epoch": 29.45,
+      "learning_rate": 5.483043388700586e-06,
+      "loss": 0.8888,
+      "step": 1049500
+    },
+    {
+      "epoch": 29.47,
+      "learning_rate": 5.342596580588378e-06,
+      "loss": 0.8788,
+      "step": 1050000
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 5.20214977247617e-06,
+      "loss": 0.8708,
+      "step": 1050500
+    },
+    {
+      "epoch": 29.49,
+      "learning_rate": 5.061702964363963e-06,
+      "loss": 0.8732,
+      "step": 1051000
+    },
+    {
+      "epoch": 29.51,
+      "learning_rate": 4.921256156251755e-06,
+      "loss": 0.8776,
+      "step": 1051500
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 4.7808093481395475e-06,
+      "loss": 0.8775,
+      "step": 1052000
+    },
+    {
+      "epoch": 29.54,
+      "learning_rate": 4.64036254002734e-06,
+      "loss": 0.8747,
+      "step": 1052500
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 4.499915731915132e-06,
+      "loss": 0.8606,
+      "step": 1053000
+    },
+    {
+      "epoch": 29.56,
+      "learning_rate": 4.359468923802925e-06,
+      "loss": 0.8943,
+      "step": 1053500
+    },
+    {
+      "epoch": 29.58,
+      "learning_rate": 4.219022115690717e-06,
+      "loss": 0.8663,
+      "step": 1054000
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 4.07857530757851e-06,
+      "loss": 0.882,
+      "step": 1054500
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 3.9381284994663016e-06,
+      "loss": 0.8743,
+      "step": 1055000
+    },
+    {
+      "epoch": 29.62,
+      "learning_rate": 3.797681691354094e-06,
+      "loss": 0.8854,
+      "step": 1055500
+    },
+    {
+      "epoch": 29.63,
+      "learning_rate": 3.6572348832418867e-06,
+      "loss": 0.862,
+      "step": 1056000
+    },
+    {
+      "epoch": 29.65,
+      "learning_rate": 3.516788075129679e-06,
+      "loss": 0.8515,
+      "step": 1056500
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 3.376341267017471e-06,
+      "loss": 0.8682,
+      "step": 1057000
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 3.2358944589052637e-06,
+      "loss": 0.8791,
+      "step": 1057500
+    },
+    {
+      "epoch": 29.69,
+      "learning_rate": 3.095447650793056e-06,
+      "loss": 0.8518,
+      "step": 1058000
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 2.9550008426808484e-06,
+      "loss": 0.8799,
+      "step": 1058500
+    },
+    {
+      "epoch": 29.72,
+      "learning_rate": 2.8145540345686408e-06,
+      "loss": 0.8911,
+      "step": 1059000
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 2.674107226456433e-06,
+      "loss": 0.8737,
+      "step": 1059500
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 2.5336604183442255e-06,
+      "loss": 0.8882,
+      "step": 1060000
+    },
+    {
+      "epoch": 29.76,
+      "learning_rate": 2.3932136102320183e-06,
+      "loss": 0.8904,
+      "step": 1060500
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 2.25276680211981e-06,
+      "loss": 0.881,
+      "step": 1061000
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 2.1123199940076025e-06,
+      "loss": 0.9008,
+      "step": 1061500
+    },
+    {
+      "epoch": 29.8,
+      "learning_rate": 1.9718731858953953e-06,
+      "loss": 0.8672,
+      "step": 1062000
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 1.8314263777831874e-06,
+      "loss": 0.898,
+      "step": 1062500
+    },
+    {
+      "epoch": 29.83,
+      "learning_rate": 1.6909795696709798e-06,
+      "loss": 0.8588,
+      "step": 1063000
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 1.5505327615587721e-06,
+      "loss": 0.8846,
+      "step": 1063500
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 1.4100859534465647e-06,
+      "loss": 0.8712,
+      "step": 1064000
+    },
+    {
+      "epoch": 29.87,
+      "learning_rate": 1.2696391453343568e-06,
+      "loss": 0.872,
+      "step": 1064500
+    },
+    {
+      "epoch": 29.89,
+      "learning_rate": 1.1291923372221492e-06,
+      "loss": 0.87,
+      "step": 1065000
+    },
+    {
+      "epoch": 29.9,
+      "learning_rate": 9.887455291099415e-07,
+      "loss": 0.8602,
+      "step": 1065500
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 8.48298720997734e-07,
+      "loss": 0.8788,
+      "step": 1066000
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 7.078519128855263e-07,
+      "loss": 0.8837,
+      "step": 1066500
+    },
+    {
+      "epoch": 29.94,
+      "learning_rate": 5.674051047733188e-07,
+      "loss": 0.8723,
+      "step": 1067000
+    },
+    {
+      "epoch": 29.96,
+      "learning_rate": 4.269582966611112e-07,
+      "loss": 0.8793,
+      "step": 1067500
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 2.8651148854890354e-07,
+      "loss": 0.8865,
+      "step": 1068000
+    },
+    {
+      "epoch": 29.98,
+      "learning_rate": 1.4606468043669594e-07,
+      "loss": 0.8533,
+      "step": 1068500
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 5.617872324488305e-09,
+      "loss": 0.8603,
+      "step": 1069000
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 1069020,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 500,
+  "total_flos": 1.8687680352154907e+21,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}