diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 19.169274992015332,
+  "global_step": 30000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.389776357827476e-07,
+      "loss": 11.1023,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2779552715654952e-06,
+      "loss": 11.0812,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.916932907348243e-06,
+      "loss": 11.1009,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5559105431309904e-06,
+      "loss": 11.1185,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.194888178913738e-06,
+      "loss": 11.0872,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.833865814696486e-06,
+      "loss": 11.0512,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.472843450479234e-06,
+      "loss": 11.0023,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.111821086261981e-06,
+      "loss": 10.9751,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.750798722044729e-06,
+      "loss": 10.9535,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.389776357827476e-06,
+      "loss": 10.9472,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.028753993610224e-06,
+      "loss": 10.922,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.667731629392972e-06,
+      "loss": 10.8773,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.306709265175718e-06,
+      "loss": 10.8485,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.945686900958468e-06,
+      "loss": 10.7794,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.584664536741214e-06,
+      "loss": 10.7876,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0223642172523962e-05,
+      "loss": 10.7846,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0862619808306708e-05,
+      "loss": 10.7546,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1501597444089457e-05,
+      "loss": 10.6649,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2140575079872205e-05,
+      "loss": 10.6765,
+      "step": 190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2779552715654951e-05,
+      "loss": 10.6211,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3418530351437701e-05,
+      "loss": 10.6008,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.4057507987220449e-05,
+      "loss": 10.5362,
+      "step": 220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4696485623003195e-05,
+      "loss": 10.4825,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.5335463258785944e-05,
+      "loss": 10.449,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.597444089456869e-05,
+      "loss": 10.3695,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6613418530351437e-05,
+      "loss": 10.3128,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7252396166134186e-05,
+      "loss": 10.221,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7891373801916936e-05,
+      "loss": 10.1125,
+      "step": 280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.853035143769968e-05,
+      "loss": 9.9933,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9169329073482428e-05,
+      "loss": 9.7903,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9808306709265177e-05,
+      "loss": 9.5898,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0447284345047924e-05,
+      "loss": 9.4338,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.1086261980830673e-05,
+      "loss": 9.2167,
+      "step": 330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1725239616613416e-05,
+      "loss": 8.9299,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.2364217252396165e-05,
+      "loss": 8.5451,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3003194888178915e-05,
+      "loss": 8.377,
+      "step": 360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.364217252396166e-05,
+      "loss": 8.1605,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.428115015974441e-05,
+      "loss": 7.8136,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4920127795527157e-05,
+      "loss": 7.4985,
+      "step": 390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.5559105431309903e-05,
+      "loss": 7.1695,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6198083067092652e-05,
+      "loss": 6.9816,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6837060702875402e-05,
+      "loss": 6.4929,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7476038338658148e-05,
+      "loss": 6.3582,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.8115015974440897e-05,
+      "loss": 6.1947,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.875399361022364e-05,
+      "loss": 5.8765,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.939297124600639e-05,
+      "loss": 5.6917,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.003194888178914e-05,
+      "loss": 5.5349,
+      "step": 470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.067092651757189e-05,
+      "loss": 5.4405,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.130990415335463e-05,
+      "loss": 5.2719,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.194888178913738e-05,
+      "loss": 4.8337,
+      "step": 500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.258785942492013e-05,
+      "loss": 4.8143,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.322683706070287e-05,
+      "loss": 4.5746,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.386581469648562e-05,
+      "loss": 4.348,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.450479233226837e-05,
+      "loss": 4.1607,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.5143769968051115e-05,
+      "loss": 4.0622,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.578274760383387e-05,
+      "loss": 3.9419,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.6421725239616614e-05,
+      "loss": 3.8112,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.706070287539936e-05,
+      "loss": 3.6399,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.769968051118211e-05,
+      "loss": 3.5703,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.8338658146964856e-05,
+      "loss": 3.4017,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.8977635782747605e-05,
+      "loss": 3.2633,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.9616613418530355e-05,
+      "loss": 3.0287,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.02555910543131e-05,
+      "loss": 3.0033,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.089456869009585e-05,
+      "loss": 3.0306,
+      "step": 640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.15335463258786e-05,
+      "loss": 2.8719,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2172523961661346e-05,
+      "loss": 2.6886,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.281150159744409e-05,
+      "loss": 2.5607,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.345047923322683e-05,
+      "loss": 2.4779,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.408945686900959e-05,
+      "loss": 2.4884,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.472843450479233e-05,
+      "loss": 2.3248,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.536741214057508e-05,
+      "loss": 2.3108,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.600638977635783e-05,
+      "loss": 2.2866,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.664536741214057e-05,
+      "loss": 2.1556,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.728434504792332e-05,
+      "loss": 2.0662,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.792332268370607e-05,
+      "loss": 2.1048,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.856230031948882e-05,
+      "loss": 1.9392,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9201277955271564e-05,
+      "loss": 1.9819,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.984025559105431e-05,
+      "loss": 1.9099,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.047923322683706e-05,
+      "loss": 1.8879,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1118210862619806e-05,
+      "loss": 1.8877,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.175718849840256e-05,
+      "loss": 1.8067,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2396166134185305e-05,
+      "loss": 1.74,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3035143769968054e-05,
+      "loss": 1.705,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.3674121405750804e-05,
+      "loss": 1.5959,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.4313099041533546e-05,
+      "loss": 1.6158,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.4952076677316296e-05,
+      "loss": 1.6574,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.559105431309904e-05,
+      "loss": 1.5606,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.6230031948881795e-05,
+      "loss": 1.5595,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.686900958466454e-05,
+      "loss": 1.6025,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.750798722044728e-05,
+      "loss": 1.6317,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.814696485623004e-05,
+      "loss": 1.4737,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.878594249201278e-05,
+      "loss": 1.5243,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.942492012779553e-05,
+      "loss": 1.4923,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.006389776357828e-05,
+      "loss": 1.3972,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.070287539936102e-05,
+      "loss": 1.4433,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.134185303514378e-05,
+      "loss": 1.4677,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.198083067092652e-05,
+      "loss": 1.3918,
+      "step": 970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.261980830670926e-05,
+      "loss": 1.2902,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325878594249202e-05,
+      "loss": 1.3005,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.389776357827476e-05,
+      "loss": 1.2544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.45367412140575e-05,
+      "loss": 1.2713,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.517571884984026e-05,
+      "loss": 1.3027,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.5814696485623e-05,
+      "loss": 1.2434,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.645367412140575e-05,
+      "loss": 1.2283,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.70926517571885e-05,
+      "loss": 1.2587,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.773162939297125e-05,
+      "loss": 1.1828,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8370607028754e-05,
+      "loss": 1.1369,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.900958466453674e-05,
+      "loss": 1.1307,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.964856230031949e-05,
+      "loss": 1.1675,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.028753993610223e-05,
+      "loss": 1.1872,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.092651757188499e-05,
+      "loss": 1.1707,
+      "step": 1110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.156549520766774e-05,
+      "loss": 1.19,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.220447284345049e-05,
+      "loss": 1.1268,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.284345047923323e-05,
+      "loss": 1.0189,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.348242811501597e-05,
+      "loss": 1.0764,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.412140575079871e-05,
+      "loss": 1.073,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.476038338658147e-05,
+      "loss": 0.9837,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.539936102236423e-05,
+      "loss": 1.014,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.603833865814697e-05,
+      "loss": 1.0643,
+      "step": 1190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.667731629392971e-05,
+      "loss": 1.0439,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.731629392971245e-05,
+      "loss": 1.0419,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.795527156549521e-05,
+      "loss": 1.01,
+      "step": 1220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.859424920127795e-05,
+      "loss": 0.9829,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.923322683706071e-05,
+      "loss": 1.0113,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.987220447284345e-05,
+      "loss": 0.9818,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.05111821086262e-05,
+      "loss": 0.9299,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.115015974440895e-05,
+      "loss": 0.9891,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.17891373801917e-05,
+      "loss": 0.9429,
+      "step": 1280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.242811501597444e-05,
+      "loss": 0.8778,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.30670926517572e-05,
+      "loss": 0.8918,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.370607028753994e-05,
+      "loss": 0.8873,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.434504792332269e-05,
+      "loss": 0.9026,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.498402555910544e-05,
+      "loss": 0.9084,
+      "step": 1330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.562300319488818e-05,
+      "loss": 0.9513,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.626198083067092e-05,
+      "loss": 0.8507,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690095846645366e-05,
+      "loss": 0.9136,
+      "step": 1360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.753993610223643e-05,
+      "loss": 0.8661,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.817891373801918e-05,
+      "loss": 0.8559,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.881789137380192e-05,
+      "loss": 0.8322,
+      "step": 1390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.945686900958466e-05,
+      "loss": 0.829,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.00958466453674e-05,
+      "loss": 0.9206,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.073482428115016e-05,
+      "loss": 0.788,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.137380191693292e-05,
+      "loss": 0.8313,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.201277955271566e-05,
+      "loss": 0.8072,
+      "step": 1440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.26517571884984e-05,
+      "loss": 0.8115,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.329073482428115e-05,
+      "loss": 0.7849,
+      "step": 1460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.39297124600639e-05,
+      "loss": 0.8003,
+      "step": 1470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.456869009584664e-05,
+      "loss": 0.7835,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.52076677316294e-05,
+      "loss": 0.8598,
+      "step": 1490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.584664536741214e-05,
+      "loss": 0.746,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.648562300319489e-05,
+      "loss": 0.8485,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.712460063897764e-05,
+      "loss": 0.7976,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.776357827476038e-05,
+      "loss": 0.7023,
+      "step": 1530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.840255591054313e-05,
+      "loss": 0.7586,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.904153354632587e-05,
+      "loss": 0.771,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.968051118210863e-05,
+      "loss": 0.7551,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010031948881789138,
+      "loss": 0.7311,
+      "step": 1570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00010095846645367413,
+      "loss": 0.7774,
+      "step": 1580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010159744408945687,
+      "loss": 0.7289,
+      "step": 1590
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00010223642172523961,
+      "loss": 0.7326,
+      "step": 1600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00010287539936102237,
+      "loss": 0.7089,
+      "step": 1610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010351437699680512,
+      "loss": 0.6825,
+      "step": 1620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00010415335463258787,
+      "loss": 0.7002,
+      "step": 1630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010479233226837061,
+      "loss": 0.6959,
+      "step": 1640
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00010543130990415335,
+      "loss": 0.713,
+      "step": 1650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00010607028753993611,
+      "loss": 0.6884,
+      "step": 1660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00010670926517571885,
+      "loss": 0.6914,
+      "step": 1670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00010734824281150161,
+      "loss": 0.6628,
+      "step": 1680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00010798722044728435,
+      "loss": 0.7524,
+      "step": 1690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00010862619808306709,
+      "loss": 0.6824,
+      "step": 1700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00010926517571884985,
+      "loss": 0.6922,
+      "step": 1710
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00010990415335463259,
+      "loss": 0.6298,
+      "step": 1720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00011054313099041533,
+      "loss": 0.6545,
+      "step": 1730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00011118210862619808,
+      "loss": 0.6842,
+      "step": 1740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00011182108626198083,
+      "loss": 0.6909,
+      "step": 1750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00011246006389776359,
+      "loss": 0.687,
+      "step": 1760
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00011309904153354633,
+      "loss": 0.6096,
+      "step": 1770
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00011373801916932908,
+      "loss": 0.6599,
+      "step": 1780
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00011437699680511182,
+      "loss": 0.7088,
+      "step": 1790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00011501597444089456,
+      "loss": 0.7042,
+      "step": 1800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00011565495207667733,
+      "loss": 0.6482,
+      "step": 1810
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00011629392971246007,
+      "loss": 0.6221,
+      "step": 1820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00011693290734824282,
+      "loss": 0.6134,
+      "step": 1830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00011757188498402556,
+      "loss": 0.5959,
+      "step": 1840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001182108626198083,
+      "loss": 0.6201,
+      "step": 1850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00011884984025559106,
+      "loss": 0.6298,
+      "step": 1860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00011948881789137381,
+      "loss": 0.5908,
+      "step": 1870
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00012012779552715656,
+      "loss": 0.5862,
+      "step": 1880
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001207667731629393,
+      "loss": 0.6449,
+      "step": 1890
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00012140575079872204,
+      "loss": 0.5922,
+      "step": 1900
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001220447284345048,
+      "loss": 0.604,
+      "step": 1910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012268370607028756,
+      "loss": 0.6396,
+      "step": 1920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00012332268370607028,
+      "loss": 0.6068,
+      "step": 1930
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00012396166134185304,
+      "loss": 0.6454,
+      "step": 1940
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012460063897763577,
+      "loss": 0.5194,
+      "step": 1950
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00012523961661341853,
+      "loss": 0.5865,
+      "step": 1960
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00012587859424920128,
+      "loss": 0.5784,
+      "step": 1970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012651757188498404,
+      "loss": 0.559,
+      "step": 1980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00012715654952076677,
+      "loss": 0.5501,
+      "step": 1990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012779552715654952,
+      "loss": 0.6316,
+      "step": 2000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00012843450479233225,
+      "loss": 0.593,
+      "step": 2010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.000129073482428115,
+      "loss": 0.5946,
+      "step": 2020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00012971246006389777,
+      "loss": 0.5813,
+      "step": 2030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00013035143769968052,
+      "loss": 0.5541,
+      "step": 2040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00013099041533546328,
+      "loss": 0.5761,
+      "step": 2050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.000131629392971246,
+      "loss": 0.557,
+      "step": 2060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00013226837060702876,
+      "loss": 0.5663,
+      "step": 2070
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001329073482428115,
+      "loss": 0.5496,
+      "step": 2080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00013354632587859425,
+      "loss": 0.5291,
+      "step": 2090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.000134185303514377,
+      "loss": 0.565,
+      "step": 2100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00013482428115015973,
+      "loss": 0.5763,
+      "step": 2110
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001354632587859425,
+      "loss": 0.6175,
+      "step": 2120
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00013610223642172525,
+      "loss": 0.5378,
+      "step": 2130
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000136741214057508,
+      "loss": 0.574,
+      "step": 2140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00013738019169329073,
+      "loss": 0.561,
+      "step": 2150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001380191693290735,
+      "loss": 0.5277,
+      "step": 2160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00013865814696485625,
+      "loss": 0.5662,
+      "step": 2170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00013929712460063897,
+      "loss": 0.526,
+      "step": 2180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00013993610223642173,
+      "loss": 0.4969,
+      "step": 2190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014057507987220446,
+      "loss": 0.5275,
+      "step": 2200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00014121405750798722,
+      "loss": 0.553,
+      "step": 2210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014185303514376997,
+      "loss": 0.5598,
+      "step": 2220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00014249201277955273,
+      "loss": 0.5127,
+      "step": 2230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00014313099041533549,
+      "loss": 0.5606,
+      "step": 2240
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014376996805111821,
+      "loss": 0.5499,
+      "step": 2250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00014440894568690097,
+      "loss": 0.5118,
+      "step": 2260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001450479233226837,
+      "loss": 0.5638,
+      "step": 2270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00014568690095846646,
+      "loss": 0.5235,
+      "step": 2280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001463258785942492,
+      "loss": 0.5114,
+      "step": 2290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00014696485623003194,
+      "loss": 0.5375,
+      "step": 2300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001476038338658147,
+      "loss": 0.5737,
+      "step": 2310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00014824281150159743,
+      "loss": 0.541,
+      "step": 2320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001488817891373802,
+      "loss": 0.5112,
+      "step": 2330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00014952076677316294,
+      "loss": 0.4936,
+      "step": 2340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001501597444089457,
+      "loss": 0.515,
+      "step": 2350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015079872204472845,
+      "loss": 0.507,
+      "step": 2360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00015143769968051118,
+      "loss": 0.4934,
+      "step": 2370
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00015207667731629394,
+      "loss": 0.5081,
+      "step": 2380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00015271565495207667,
+      "loss": 0.4726,
+      "step": 2390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00015335463258785942,
+      "loss": 0.5149,
+      "step": 2400
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00015399361022364218,
+      "loss": 0.4544,
+      "step": 2410
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001546325878594249,
+      "loss": 0.5026,
+      "step": 2420
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.0001552715654952077,
+      "loss": 0.4987,
+      "step": 2430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00015591054313099042,
+      "loss": 0.4715,
+      "step": 2440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00015654952076677318,
+      "loss": 0.5137,
+      "step": 2450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.0001571884984025559,
+      "loss": 0.4659,
+      "step": 2460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015782747603833866,
+      "loss": 0.5293,
+      "step": 2470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00015846645367412142,
+      "loss": 0.4848,
+      "step": 2480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00015910543130990415,
+      "loss": 0.5464,
+      "step": 2490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001597444089456869,
+      "loss": 0.4918,
+      "step": 2500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016038338658146963,
+      "loss": 0.4821,
+      "step": 2510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001610223642172524,
+      "loss": 0.4464,
+      "step": 2520
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016166134185303515,
+      "loss": 0.5041,
+      "step": 2530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001623003194888179,
+      "loss": 0.5254,
+      "step": 2540
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016293929712460066,
+      "loss": 0.4722,
+      "step": 2550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001635782747603834,
+      "loss": 0.5127,
+      "step": 2560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016421725239616614,
+      "loss": 0.4978,
+      "step": 2570
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016485623003194887,
+      "loss": 0.5174,
+      "step": 2580
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016549520766773163,
+      "loss": 0.528,
+      "step": 2590
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001661341853035144,
+      "loss": 0.5612,
+      "step": 2600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016677316293929712,
+      "loss": 0.4953,
+      "step": 2610
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016741214057507987,
+      "loss": 0.475,
+      "step": 2620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016805111821086263,
+      "loss": 0.476,
+      "step": 2630
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016869009584664538,
+      "loss": 0.4799,
+      "step": 2640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016932907348242811,
+      "loss": 0.4838,
+      "step": 2650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016996805111821087,
+      "loss": 0.5215,
+      "step": 2660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00017060702875399363,
+      "loss": 0.4927,
+      "step": 2670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00017124600638977636,
+      "loss": 0.4973,
+      "step": 2680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001718849840255591,
+      "loss": 0.4703,
+      "step": 2690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00017252396166134184,
+      "loss": 0.5073,
+      "step": 2700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001731629392971246,
+      "loss": 0.4445,
+      "step": 2710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00017380191693290733,
+      "loss": 0.463,
+      "step": 2720
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001744408945686901,
+      "loss": 0.4289,
+      "step": 2730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00017507987220447287,
+      "loss": 0.4378,
+      "step": 2740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001757188498402556,
+      "loss": 0.4408,
+      "step": 2750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00017635782747603835,
+      "loss": 0.518,
+      "step": 2760
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00017699680511182108,
+      "loss": 0.4583,
+      "step": 2770
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00017763578274760384,
+      "loss": 0.4415,
+      "step": 2780
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.0001782747603833866,
+      "loss": 0.4543,
+      "step": 2790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00017891373801916932,
+      "loss": 0.4353,
+      "step": 2800
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00017955271565495208,
+      "loss": 0.5315,
+      "step": 2810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.0001801916932907348,
+      "loss": 0.468,
+      "step": 2820
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001808306709265176,
+      "loss": 0.4915,
+      "step": 2830
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00018146964856230032,
+      "loss": 0.4684,
+      "step": 2840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00018210862619808308,
+      "loss": 0.4367,
+      "step": 2850
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00018274760383386583,
+      "loss": 0.3751,
+      "step": 2860
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00018338658146964856,
+      "loss": 0.4347,
+      "step": 2870
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00018402555910543132,
+      "loss": 0.4611,
+      "step": 2880
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00018466453674121405,
+      "loss": 0.4425,
+      "step": 2890
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001853035143769968,
+      "loss": 0.4483,
+      "step": 2900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00018594249201277953,
+      "loss": 0.4603,
+      "step": 2910
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001865814696485623,
+      "loss": 0.4369,
+      "step": 2920
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00018722044728434507,
+      "loss": 0.4521,
+      "step": 2930
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001878594249201278,
+      "loss": 0.4239,
+      "step": 2940
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00018849840255591056,
+      "loss": 0.3795,
+      "step": 2950
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001891373801916933,
+      "loss": 0.477,
+      "step": 2960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00018977635782747604,
+      "loss": 0.4273,
+      "step": 2970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001904153354632588,
+      "loss": 0.4786,
+      "step": 2980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00019105431309904153,
+      "loss": 0.4803,
+      "step": 2990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00019169329073482429,
+      "loss": 0.4497,
+      "step": 3000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00019233226837060702,
+      "loss": 0.4749,
+      "step": 3010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00019297124600638977,
+      "loss": 0.4231,
+      "step": 3020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00019361022364217253,
+      "loss": 0.4618,
+      "step": 3030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00019424920127795528,
+      "loss": 0.4265,
+      "step": 3040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00019488817891373804,
+      "loss": 0.39,
+      "step": 3050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00019552715654952077,
+      "loss": 0.4346,
+      "step": 3060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00019616613418530353,
+      "loss": 0.4079,
+      "step": 3070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00019680511182108626,
+      "loss": 0.4351,
+      "step": 3080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000197444089456869,
+      "loss": 0.4402,
+      "step": 3090
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00019808306709265174,
+      "loss": 0.4294,
+      "step": 3100
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001987220447284345,
+      "loss": 0.4079,
+      "step": 3110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00019936102236421725,
+      "loss": 0.398,
+      "step": 3120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0002,
+      "loss": 0.3759,
+      "step": 3130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00020063897763578277,
+      "loss": 0.408,
+      "step": 3140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0002012779552715655,
+      "loss": 0.4261,
+      "step": 3150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00020191693290734825,
+      "loss": 0.3951,
+      "step": 3160
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000202555910543131,
+      "loss": 0.3488,
+      "step": 3170
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00020319488817891374,
+      "loss": 0.3529,
+      "step": 3180
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.0002038338658146965,
+      "loss": 0.3806,
+      "step": 3190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00020447284345047922,
+      "loss": 0.3641,
+      "step": 3200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00020511182108626198,
+      "loss": 0.4043,
+      "step": 3210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00020575079872204473,
+      "loss": 0.3598,
+      "step": 3220
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0002063897763578275,
+      "loss": 0.3858,
+      "step": 3230
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00020702875399361025,
+      "loss": 0.3666,
+      "step": 3240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00020766773162939298,
+      "loss": 0.3702,
+      "step": 3250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00020830670926517573,
+      "loss": 0.3782,
+      "step": 3260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00020894568690095846,
+      "loss": 0.3782,
+      "step": 3270
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00020958466453674122,
+      "loss": 0.3474,
+      "step": 3280
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00021022364217252395,
+      "loss": 0.3645,
+      "step": 3290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0002108626198083067,
+      "loss": 0.4034,
+      "step": 3300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00021150159744408946,
+      "loss": 0.3892,
+      "step": 3310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00021214057507987222,
+      "loss": 0.3843,
+      "step": 3320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00021277955271565497,
+      "loss": 0.3667,
+      "step": 3330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0002134185303514377,
+      "loss": 0.3954,
+      "step": 3340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00021405750798722046,
+      "loss": 0.3733,
+      "step": 3350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00021469648562300321,
+      "loss": 0.3801,
+      "step": 3360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00021533546325878594,
+      "loss": 0.3813,
+      "step": 3370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.0002159744408945687,
+      "loss": 0.382,
+      "step": 3380
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00021661341853035143,
+      "loss": 0.3959,
+      "step": 3390
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00021725239616613419,
+      "loss": 0.338,
+      "step": 3400
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00021789137380191691,
+      "loss": 0.3884,
+      "step": 3410
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.0002185303514376997,
+      "loss": 0.3583,
+      "step": 3420
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00021916932907348245,
+      "loss": 0.3224,
+      "step": 3430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00021980830670926518,
+      "loss": 0.3587,
+      "step": 3440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00022044728434504794,
+      "loss": 0.3493,
+      "step": 3450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00022108626198083067,
+      "loss": 0.392,
+      "step": 3460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00022172523961661343,
+      "loss": 0.3555,
+      "step": 3470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00022236421725239615,
+      "loss": 0.4006,
+      "step": 3480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0002230031948881789,
+      "loss": 0.3817,
+      "step": 3490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00022364217252396167,
+      "loss": 0.3957,
+      "step": 3500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0002242811501597444,
+      "loss": 0.3538,
+      "step": 3510
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00022492012779552718,
+      "loss": 0.3811,
+      "step": 3520
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0002255591054313099,
+      "loss": 0.3615,
+      "step": 3530
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00022619808306709267,
+      "loss": 0.3491,
+      "step": 3540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00022683706070287542,
+      "loss": 0.3904,
+      "step": 3550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00022747603833865815,
+      "loss": 0.3692,
+      "step": 3560
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0002281150159744409,
+      "loss": 0.3764,
+      "step": 3570
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00022875399361022364,
+      "loss": 0.3189,
+      "step": 3580
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0002293929712460064,
+      "loss": 0.3437,
+      "step": 3590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00023003194888178912,
+      "loss": 0.38,
+      "step": 3600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00023067092651757188,
+      "loss": 0.3694,
+      "step": 3610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00023130990415335466,
+      "loss": 0.3297,
+      "step": 3620
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.0002319488817891374,
+      "loss": 0.391,
+      "step": 3630
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00023258785942492015,
+      "loss": 0.3465,
+      "step": 3640
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00023322683706070288,
+      "loss": 0.3879,
+      "step": 3650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00023386581469648563,
+      "loss": 0.3551,
+      "step": 3660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00023450479233226836,
+      "loss": 0.3383,
+      "step": 3670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00023514376996805112,
+      "loss": 0.3284,
+      "step": 3680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00023578274760383387,
+      "loss": 0.3362,
+      "step": 3690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0002364217252396166,
+      "loss": 0.349,
+      "step": 3700
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00023706070287539936,
+      "loss": 0.357,
+      "step": 3710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00023769968051118212,
+      "loss": 0.3993,
+      "step": 3720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00023833865814696487,
+      "loss": 0.3611,
+      "step": 3730
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00023897763578274763,
+      "loss": 0.3989,
+      "step": 3740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00023961661341853036,
+      "loss": 0.3792,
+      "step": 3750
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00024025559105431311,
+      "loss": 0.3434,
+      "step": 3760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00024089456869009584,
+      "loss": 0.3669,
+      "step": 3770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.0002415335463258786,
+      "loss": 0.3858,
+      "step": 3780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00024217252396166133,
+      "loss": 0.3016,
+      "step": 3790
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024281150159744408,
+      "loss": 0.3253,
+      "step": 3800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00024345047923322684,
+      "loss": 0.3811,
+      "step": 3810
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.0002440894568690096,
+      "loss": 0.338,
+      "step": 3820
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002447284345047923,
+      "loss": 0.3214,
+      "step": 3830
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0002453674121405751,
+      "loss": 0.3696,
+      "step": 3840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00024600638977635784,
+      "loss": 0.3624,
+      "step": 3850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024664536741214057,
+      "loss": 0.3538,
+      "step": 3860
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00024728434504792335,
+      "loss": 0.372,
+      "step": 3870
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0002479233226837061,
+      "loss": 0.3884,
+      "step": 3880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0002485623003194888,
+      "loss": 0.325,
+      "step": 3890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00024920127795527154,
+      "loss": 0.3567,
+      "step": 3900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0002498402555910543,
+      "loss": 0.379,
+      "step": 3910
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00025047923322683705,
+      "loss": 0.3542,
+      "step": 3920
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00025111821086261984,
+      "loss": 0.3739,
+      "step": 3930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00025175718849840256,
+      "loss": 0.3657,
+      "step": 3940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0002523961661341853,
+      "loss": 0.3499,
+      "step": 3950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0002530351437699681,
+      "loss": 0.3428,
+      "step": 3960
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.0002536741214057508,
+      "loss": 0.3369,
+      "step": 3970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00025431309904153354,
+      "loss": 0.4055,
+      "step": 3980
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00025495207667731626,
+      "loss": 0.365,
+      "step": 3990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00025559105431309905,
+      "loss": 0.3533,
+      "step": 4000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.0002562300319488818,
+      "loss": 0.3223,
+      "step": 4010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002568690095846645,
+      "loss": 0.3176,
+      "step": 4020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.0002575079872204473,
+      "loss": 0.3098,
+      "step": 4030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00025814696485623,
+      "loss": 0.3792,
+      "step": 4040
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025878594249201275,
+      "loss": 0.357,
+      "step": 4050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00025942492012779553,
+      "loss": 0.3706,
+      "step": 4060
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0002600638977635783,
+      "loss": 0.346,
+      "step": 4070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00026070287539936104,
+      "loss": 0.341,
+      "step": 4080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0002613418530351438,
+      "loss": 0.3352,
+      "step": 4090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00026198083067092656,
+      "loss": 0.3219,
+      "step": 4100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0002626198083067093,
+      "loss": 0.3638,
+      "step": 4110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.000263258785942492,
+      "loss": 0.3356,
+      "step": 4120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0002638977635782748,
+      "loss": 0.3309,
+      "step": 4130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00026453674121405753,
+      "loss": 0.3649,
+      "step": 4140
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00026517571884984026,
+      "loss": 0.3334,
+      "step": 4150
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.000265814696485623,
+      "loss": 0.3279,
+      "step": 4160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00026645367412140577,
+      "loss": 0.3603,
+      "step": 4170
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.0002670926517571885,
+      "loss": 0.3207,
+      "step": 4180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00026773162939297123,
+      "loss": 0.2887,
+      "step": 4190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.000268370607028754,
+      "loss": 0.343,
+      "step": 4200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00026900958466453674,
+      "loss": 0.3347,
+      "step": 4210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00026964856230031947,
+      "loss": 0.3013,
+      "step": 4220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.0002702875399361022,
+      "loss": 0.3681,
+      "step": 4230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.000270926517571885,
+      "loss": 0.313,
+      "step": 4240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002715654952076677,
+      "loss": 0.3769,
+      "step": 4250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0002722044728434505,
+      "loss": 0.346,
+      "step": 4260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0002728434504792333,
+      "loss": 0.321,
+      "step": 4270
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000273482428115016,
+      "loss": 0.3783,
+      "step": 4280
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00027412140575079874,
+      "loss": 0.3621,
+      "step": 4290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00027476038338658147,
+      "loss": 0.3509,
+      "step": 4300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00027539936102236425,
+      "loss": 0.3836,
+      "step": 4310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000276038338658147,
+      "loss": 0.3595,
+      "step": 4320
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002766773162939297,
+      "loss": 0.3375,
+      "step": 4330
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0002773162939297125,
+      "loss": 0.3343,
+      "step": 4340
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0002779552715654952,
+      "loss": 0.3306,
+      "step": 4350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00027859424920127795,
+      "loss": 0.3474,
+      "step": 4360
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.0002792332268370607,
+      "loss": 0.3696,
+      "step": 4370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00027987220447284346,
+      "loss": 0.35,
+      "step": 4380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.0002805111821086262,
+      "loss": 0.3887,
+      "step": 4390
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.0002811501597444089,
+      "loss": 0.3449,
+      "step": 4400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0002817891373801917,
+      "loss": 0.369,
+      "step": 4410
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00028242811501597443,
+      "loss": 0.373,
+      "step": 4420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00028306709265175716,
+      "loss": 0.3299,
+      "step": 4430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00028370607028753995,
+      "loss": 0.3233,
+      "step": 4440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0002843450479233227,
+      "loss": 0.3629,
+      "step": 4450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00028498402555910546,
+      "loss": 0.3576,
+      "step": 4460
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0002856230031948882,
+      "loss": 0.3053,
+      "step": 4470
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00028626198083067097,
+      "loss": 0.3185,
+      "step": 4480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0002869009584664537,
+      "loss": 0.3423,
+      "step": 4490
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00028753993610223643,
+      "loss": 0.3727,
+      "step": 4500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0002881789137380192,
+      "loss": 0.3121,
+      "step": 4510
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00028881789137380194,
+      "loss": 0.322,
+      "step": 4520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00028945686900958467,
+      "loss": 0.3581,
+      "step": 4530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0002900958466453674,
+      "loss": 0.339,
+      "step": 4540
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002907348242811502,
+      "loss": 0.3226,
+      "step": 4550
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0002913738019169329,
+      "loss": 0.2947,
+      "step": 4560
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00029201277955271564,
+      "loss": 0.3572,
+      "step": 4570
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.0002926517571884984,
+      "loss": 0.3029,
+      "step": 4580
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00029329073482428115,
+      "loss": 0.3149,
+      "step": 4590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.0002939297124600639,
+      "loss": 0.3521,
+      "step": 4600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002945686900958466,
+      "loss": 0.3207,
+      "step": 4610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0002952076677316294,
+      "loss": 0.2926,
+      "step": 4620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0002958466453674121,
+      "loss": 0.3318,
+      "step": 4630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00029648562300319485,
+      "loss": 0.2907,
+      "step": 4640
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00029712460063897764,
+      "loss": 0.3876,
+      "step": 4650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0002977635782747604,
+      "loss": 0.3375,
+      "step": 4660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00029840255591054315,
+      "loss": 0.3327,
+      "step": 4670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0002990415335463259,
+      "loss": 0.2991,
+      "step": 4680
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00029968051118210866,
+      "loss": 0.3027,
+      "step": 4690
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0003003194888178914,
+      "loss": 0.3189,
+      "step": 4700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0003009584664536741,
+      "loss": 0.3073,
+      "step": 4710
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0003015974440894569,
+      "loss": 0.2792,
+      "step": 4720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00030223642172523963,
+      "loss": 0.2776,
+      "step": 4730
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00030287539936102236,
+      "loss": 0.3064,
+      "step": 4740
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003035143769968051,
+      "loss": 0.2736,
+      "step": 4750
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0003041533546325879,
+      "loss": 0.2906,
+      "step": 4760
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0003047923322683706,
+      "loss": 0.2766,
+      "step": 4770
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00030543130990415333,
+      "loss": 0.3055,
+      "step": 4780
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.0003060702875399361,
+      "loss": 0.2889,
+      "step": 4790
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00030670926517571885,
+      "loss": 0.2952,
+      "step": 4800
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.0003073482428115016,
+      "loss": 0.3135,
+      "step": 4810
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00030798722044728436,
+      "loss": 0.3029,
+      "step": 4820
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003086261980830671,
+      "loss": 0.2786,
+      "step": 4830
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0003092651757188498,
+      "loss": 0.2854,
+      "step": 4840
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00030990415335463255,
+      "loss": 0.2851,
+      "step": 4850
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003105431309904154,
+      "loss": 0.2715,
+      "step": 4860
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0003111821086261981,
+      "loss": 0.2966,
+      "step": 4870
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00031182108626198084,
+      "loss": 0.2524,
+      "step": 4880
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0003124600638977636,
+      "loss": 0.3158,
+      "step": 4890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00031309904153354636,
+      "loss": 0.2955,
+      "step": 4900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003137380191693291,
+      "loss": 0.3291,
+      "step": 4910
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0003143769968051118,
+      "loss": 0.3231,
+      "step": 4920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0003150159744408946,
+      "loss": 0.3137,
+      "step": 4930
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.0003156549520766773,
+      "loss": 0.2828,
+      "step": 4940
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00031629392971246006,
+      "loss": 0.2963,
+      "step": 4950
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00031693290734824284,
+      "loss": 0.281,
+      "step": 4960
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00031757188498402557,
+      "loss": 0.2507,
+      "step": 4970
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.0003182108626198083,
+      "loss": 0.2772,
+      "step": 4980
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.000318849840255591,
+      "loss": 0.2989,
+      "step": 4990
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.0003194888178913738,
+      "loss": 0.2848,
+      "step": 5000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00032012779552715654,
+      "loss": 0.2925,
+      "step": 5010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00032076677316293927,
+      "loss": 0.2756,
+      "step": 5020
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00032140575079872205,
+      "loss": 0.2793,
+      "step": 5030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0003220447284345048,
+      "loss": 0.3096,
+      "step": 5040
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003226837060702875,
+      "loss": 0.3056,
+      "step": 5050
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0003233226837060703,
+      "loss": 0.2636,
+      "step": 5060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0003239616613418531,
+      "loss": 0.362,
+      "step": 5070
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0003246006389776358,
+      "loss": 0.294,
+      "step": 5080
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00032523961661341854,
+      "loss": 0.2913,
+      "step": 5090
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0003258785942492013,
+      "loss": 0.3018,
+      "step": 5100
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00032651757188498405,
+      "loss": 0.3096,
+      "step": 5110
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0003271565495207668,
+      "loss": 0.2857,
+      "step": 5120
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0003277955271565495,
+      "loss": 0.2586,
+      "step": 5130
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0003284345047923323,
+      "loss": 0.2943,
+      "step": 5140
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.000329073482428115,
+      "loss": 0.2719,
+      "step": 5150
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00032971246006389775,
+      "loss": 0.2682,
+      "step": 5160
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00033035143769968053,
+      "loss": 0.2883,
+      "step": 5170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00033099041533546326,
+      "loss": 0.286,
+      "step": 5180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.000331629392971246,
+      "loss": 0.3188,
+      "step": 5190
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.0003322683706070288,
+      "loss": 0.2683,
+      "step": 5200
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0003329073482428115,
+      "loss": 0.2782,
+      "step": 5210
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00033354632587859423,
+      "loss": 0.2968,
+      "step": 5220
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00033418530351437696,
+      "loss": 0.2765,
+      "step": 5230
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00033482428115015974,
+      "loss": 0.2481,
+      "step": 5240
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00033546325878594253,
+      "loss": 0.301,
+      "step": 5250
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00033610223642172526,
+      "loss": 0.2576,
+      "step": 5260
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00033674121405750804,
+      "loss": 0.3013,
+      "step": 5270
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00033738019169329077,
+      "loss": 0.3062,
+      "step": 5280
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0003380191693290735,
+      "loss": 0.2969,
+      "step": 5290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00033865814696485623,
+      "loss": 0.3039,
+      "step": 5300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.000339297124600639,
+      "loss": 0.2954,
+      "step": 5310
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00033993610223642174,
+      "loss": 0.2835,
+      "step": 5320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00034057507987220447,
+      "loss": 0.3215,
+      "step": 5330
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00034121405750798725,
+      "loss": 0.2902,
+      "step": 5340
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00034185303514377,
+      "loss": 0.2957,
+      "step": 5350
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.0003424920127795527,
+      "loss": 0.2844,
+      "step": 5360
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00034313099041533544,
+      "loss": 0.2502,
+      "step": 5370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0003437699680511182,
+      "loss": 0.2757,
+      "step": 5380
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00034440894568690095,
+      "loss": 0.3069,
+      "step": 5390
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.0003450479233226837,
+      "loss": 0.2948,
+      "step": 5400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00034568690095846647,
+      "loss": 0.2886,
+      "step": 5410
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0003463258785942492,
+      "loss": 0.2625,
+      "step": 5420
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0003469648562300319,
+      "loss": 0.3224,
+      "step": 5430
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00034760383386581465,
+      "loss": 0.3151,
+      "step": 5440
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0003482428115015975,
+      "loss": 0.3082,
+      "step": 5450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0003488817891373802,
+      "loss": 0.2851,
+      "step": 5460
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00034952076677316295,
+      "loss": 0.2495,
+      "step": 5470
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00035015974440894573,
+      "loss": 0.3161,
+      "step": 5480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00035079872204472846,
+      "loss": 0.3085,
+      "step": 5490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0003514376996805112,
+      "loss": 0.2806,
+      "step": 5500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0003520766773162939,
+      "loss": 0.2427,
+      "step": 5510
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.0003527156549520767,
+      "loss": 0.2567,
+      "step": 5520
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00035335463258785943,
+      "loss": 0.2581,
+      "step": 5530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00035399361022364216,
+      "loss": 0.3021,
+      "step": 5540
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00035463258785942495,
+      "loss": 0.2633,
+      "step": 5550
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.0003552715654952077,
+      "loss": 0.2753,
+      "step": 5560
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.0003559105431309904,
+      "loss": 0.2755,
+      "step": 5570
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003565495207667732,
+      "loss": 0.297,
+      "step": 5580
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.0003571884984025559,
+      "loss": 0.2739,
+      "step": 5590
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00035782747603833865,
+      "loss": 0.2994,
+      "step": 5600
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0003584664536741214,
+      "loss": 0.3026,
+      "step": 5610
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00035910543130990416,
+      "loss": 0.2724,
+      "step": 5620
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003597444089456869,
+      "loss": 0.2863,
+      "step": 5630
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0003603833865814696,
+      "loss": 0.3245,
+      "step": 5640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00036102236421725245,
+      "loss": 0.2751,
+      "step": 5650
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003616613418530352,
+      "loss": 0.2783,
+      "step": 5660
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0003623003194888179,
+      "loss": 0.3159,
+      "step": 5670
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00036293929712460064,
+      "loss": 0.3649,
+      "step": 5680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0003635782747603834,
+      "loss": 0.2486,
+      "step": 5690
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00036421725239616615,
+      "loss": 0.319,
+      "step": 5700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0003648562300319489,
+      "loss": 0.2897,
+      "step": 5710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00036549520766773167,
+      "loss": 0.2648,
+      "step": 5720
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.0003661341853035144,
+      "loss": 0.3063,
+      "step": 5730
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0003667731629392971,
+      "loss": 0.2459,
+      "step": 5740
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00036741214057507985,
+      "loss": 0.3034,
+      "step": 5750
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00036805111821086264,
+      "loss": 0.2805,
+      "step": 5760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00036869009584664537,
+      "loss": 0.2748,
+      "step": 5770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.0003693290734824281,
+      "loss": 0.2738,
+      "step": 5780
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.0003699680511182109,
+      "loss": 0.2739,
+      "step": 5790
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0003706070287539936,
+      "loss": 0.3178,
+      "step": 5800
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00037124600638977634,
+      "loss": 0.2705,
+      "step": 5810
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00037188498402555907,
+      "loss": 0.3111,
+      "step": 5820
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00037252396166134185,
+      "loss": 0.2781,
+      "step": 5830
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0003731629392971246,
+      "loss": 0.2899,
+      "step": 5840
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00037380191693290736,
+      "loss": 0.316,
+      "step": 5850
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00037444089456869015,
+      "loss": 0.3097,
+      "step": 5860
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0003750798722044729,
+      "loss": 0.2693,
+      "step": 5870
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0003757188498402556,
+      "loss": 0.2846,
+      "step": 5880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00037635782747603833,
+      "loss": 0.2892,
+      "step": 5890
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0003769968051118211,
+      "loss": 0.313,
+      "step": 5900
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00037763578274760385,
+      "loss": 0.2979,
+      "step": 5910
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0003782747603833866,
+      "loss": 0.2734,
+      "step": 5920
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00037891373801916936,
+      "loss": 0.2706,
+      "step": 5930
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003795527156549521,
+      "loss": 0.2963,
+      "step": 5940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.0003801916932907348,
+      "loss": 0.2532,
+      "step": 5950
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.0003808306709265176,
+      "loss": 0.3183,
+      "step": 5960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00038146964856230033,
+      "loss": 0.3032,
+      "step": 5970
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00038210862619808306,
+      "loss": 0.3129,
+      "step": 5980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0003827476038338658,
+      "loss": 0.267,
+      "step": 5990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00038338658146964857,
+      "loss": 0.2766,
+      "step": 6000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0003840255591054313,
+      "loss": 0.2861,
+      "step": 6010
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00038466453674121403,
+      "loss": 0.2807,
+      "step": 6020
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0003853035143769968,
+      "loss": 0.3118,
+      "step": 6030
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00038594249201277954,
+      "loss": 0.286,
+      "step": 6040
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0003865814696485623,
+      "loss": 0.2676,
+      "step": 6050
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00038722044728434506,
+      "loss": 0.3028,
+      "step": 6060
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00038785942492012784,
+      "loss": 0.2564,
+      "step": 6070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00038849840255591057,
+      "loss": 0.2974,
+      "step": 6080
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0003891373801916933,
+      "loss": 0.2727,
+      "step": 6090
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003897763578274761,
+      "loss": 0.2673,
+      "step": 6100
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.0003904153354632588,
+      "loss": 0.27,
+      "step": 6110
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00039105431309904154,
+      "loss": 0.2758,
+      "step": 6120
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00039169329073482427,
+      "loss": 0.2891,
+      "step": 6130
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00039233226837060705,
+      "loss": 0.2904,
+      "step": 6140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.0003929712460063898,
+      "loss": 0.3009,
+      "step": 6150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003936102236421725,
+      "loss": 0.314,
+      "step": 6160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.0003942492012779553,
+      "loss": 0.289,
+      "step": 6170
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.000394888178913738,
+      "loss": 0.3163,
+      "step": 6180
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00039552715654952075,
+      "loss": 0.2605,
+      "step": 6190
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0003961661341853035,
+      "loss": 0.2758,
+      "step": 6200
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00039680511182108626,
+      "loss": 0.3064,
+      "step": 6210
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.000397444089456869,
+      "loss": 0.2991,
+      "step": 6220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0003980830670926517,
+      "loss": 0.307,
+      "step": 6230
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003987220447284345,
+      "loss": 0.285,
+      "step": 6240
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0003993610223642173,
+      "loss": 0.2797,
+      "step": 6250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0004,
+      "loss": 0.2737,
+      "step": 6260
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040063897763578275,
+      "loss": 0.2392,
+      "step": 6270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00040127795527156553,
+      "loss": 0.2275,
+      "step": 6280
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00040191693290734826,
+      "loss": 0.2617,
+      "step": 6290
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.000402555910543131,
+      "loss": 0.2528,
+      "step": 6300
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.0004031948881789138,
+      "loss": 0.2262,
+      "step": 6310
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.0004038338658146965,
+      "loss": 0.259,
+      "step": 6320
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00040447284345047923,
+      "loss": 0.2519,
+      "step": 6330
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.000405111821086262,
+      "loss": 0.2259,
+      "step": 6340
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00040575079872204474,
+      "loss": 0.2386,
+      "step": 6350
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0004063897763578275,
+      "loss": 0.2217,
+      "step": 6360
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.0004070287539936102,
+      "loss": 0.2309,
+      "step": 6370
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.000407667731629393,
+      "loss": 0.211,
+      "step": 6380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.0004083067092651757,
+      "loss": 0.2459,
+      "step": 6390
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00040894568690095844,
+      "loss": 0.2603,
+      "step": 6400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00040958466453674123,
+      "loss": 0.2343,
+      "step": 6410
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00041022364217252396,
+      "loss": 0.2562,
+      "step": 6420
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0004108626198083067,
+      "loss": 0.2148,
+      "step": 6430
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00041150159744408947,
+      "loss": 0.2349,
+      "step": 6440
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00041214057507987225,
+      "loss": 0.2975,
+      "step": 6450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.000412779552715655,
+      "loss": 0.2541,
+      "step": 6460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0004134185303514377,
+      "loss": 0.2555,
+      "step": 6470
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0004140575079872205,
+      "loss": 0.2818,
+      "step": 6480
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0004146964856230032,
+      "loss": 0.2577,
+      "step": 6490
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00041533546325878595,
+      "loss": 0.2433,
+      "step": 6500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.0004159744408945687,
+      "loss": 0.2231,
+      "step": 6510
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00041661341853035147,
+      "loss": 0.2841,
+      "step": 6520
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.0004172523961661342,
+      "loss": 0.2048,
+      "step": 6530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.0004178913738019169,
+      "loss": 0.2484,
+      "step": 6540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0004185303514376997,
+      "loss": 0.2542,
+      "step": 6550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00041916932907348244,
+      "loss": 0.2293,
+      "step": 6560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00041980830670926517,
+      "loss": 0.2545,
+      "step": 6570
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.0004204472843450479,
+      "loss": 0.2565,
+      "step": 6580
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.0004210862619808307,
+      "loss": 0.2441,
+      "step": 6590
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0004217252396166134,
+      "loss": 0.2623,
+      "step": 6600
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00042236421725239614,
+      "loss": 0.2542,
+      "step": 6610
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0004230031948881789,
+      "loss": 0.2368,
+      "step": 6620
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00042364217252396165,
+      "loss": 0.2683,
+      "step": 6630
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00042428115015974443,
+      "loss": 0.2468,
+      "step": 6640
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00042492012779552716,
+      "loss": 0.226,
+      "step": 6650
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00042555910543130995,
+      "loss": 0.2362,
+      "step": 6660
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0004261980830670927,
+      "loss": 0.2606,
+      "step": 6670
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0004268370607028754,
+      "loss": 0.2436,
+      "step": 6680
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0004274760383386582,
+      "loss": 0.2501,
+      "step": 6690
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.0004281150159744409,
+      "loss": 0.2367,
+      "step": 6700
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00042875399361022365,
+      "loss": 0.248,
+      "step": 6710
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00042939297124600643,
+      "loss": 0.2475,
+      "step": 6720
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00043003194888178916,
+      "loss": 0.2544,
+      "step": 6730
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0004306709265175719,
+      "loss": 0.2127,
+      "step": 6740
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.0004313099041533546,
+      "loss": 0.2495,
+      "step": 6750
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0004319488817891374,
+      "loss": 0.2434,
+      "step": 6760
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00043258785942492013,
+      "loss": 0.2641,
+      "step": 6770
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00043322683706070286,
+      "loss": 0.2549,
+      "step": 6780
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00043386581469648564,
+      "loss": 0.2748,
+      "step": 6790
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00043450479233226837,
+      "loss": 0.2991,
+      "step": 6800
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0004351437699680511,
+      "loss": 0.2217,
+      "step": 6810
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00043578274760383383,
+      "loss": 0.2307,
+      "step": 6820
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0004364217252396166,
+      "loss": 0.234,
+      "step": 6830
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0004370607028753994,
+      "loss": 0.2637,
+      "step": 6840
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0004376996805111821,
+      "loss": 0.2285,
+      "step": 6850
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0004383386581469649,
+      "loss": 0.2585,
+      "step": 6860
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00043897763578274764,
+      "loss": 0.2358,
+      "step": 6870
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00043961661341853037,
+      "loss": 0.246,
+      "step": 6880
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.0004402555910543131,
+      "loss": 0.2672,
+      "step": 6890
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0004408945686900959,
+      "loss": 0.2606,
+      "step": 6900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.0004415335463258786,
+      "loss": 0.2397,
+      "step": 6910
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00044217252396166134,
+      "loss": 0.2539,
+      "step": 6920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.0004428115015974441,
+      "loss": 0.2348,
+      "step": 6930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00044345047923322685,
+      "loss": 0.2845,
+      "step": 6940
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.0004440894568690096,
+      "loss": 0.2402,
+      "step": 6950
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0004447284345047923,
+      "loss": 0.2678,
+      "step": 6960
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0004453674121405751,
+      "loss": 0.2653,
+      "step": 6970
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0004460063897763578,
+      "loss": 0.2813,
+      "step": 6980
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00044664536741214055,
+      "loss": 0.262,
+      "step": 6990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00044728434504792333,
+      "loss": 0.2247,
+      "step": 7000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00044792332268370606,
+      "loss": 0.2547,
+      "step": 7010
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0004485623003194888,
+      "loss": 0.2627,
+      "step": 7020
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0004492012779552715,
+      "loss": 0.277,
+      "step": 7030
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00044984025559105436,
+      "loss": 0.2467,
+      "step": 7040
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0004504792332268371,
+      "loss": 0.2437,
+      "step": 7050
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0004511182108626198,
+      "loss": 0.2605,
+      "step": 7060
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.0004517571884984026,
+      "loss": 0.2649,
+      "step": 7070
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00045239616613418533,
+      "loss": 0.283,
+      "step": 7080
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00045303514376996806,
+      "loss": 0.2337,
+      "step": 7090
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00045367412140575084,
+      "loss": 0.2756,
+      "step": 7100
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00045431309904153357,
+      "loss": 0.2699,
+      "step": 7110
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0004549520766773163,
+      "loss": 0.2649,
+      "step": 7120
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00045559105431309903,
+      "loss": 0.2685,
+      "step": 7130
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.0004562300319488818,
+      "loss": 0.2451,
+      "step": 7140
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00045686900958466454,
+      "loss": 0.2614,
+      "step": 7150
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00045750798722044727,
+      "loss": 0.2484,
+      "step": 7160
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00045814696485623006,
+      "loss": 0.2642,
+      "step": 7170
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0004587859424920128,
+      "loss": 0.2413,
+      "step": 7180
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.0004594249201277955,
+      "loss": 0.2324,
+      "step": 7190
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00046006389776357824,
+      "loss": 0.2632,
+      "step": 7200
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.000460702875399361,
+      "loss": 0.2472,
+      "step": 7210
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00046134185303514376,
+      "loss": 0.276,
+      "step": 7220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0004619808306709265,
+      "loss": 0.2647,
+      "step": 7230
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0004626198083067093,
+      "loss": 0.2409,
+      "step": 7240
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00046325878594249205,
+      "loss": 0.2568,
+      "step": 7250
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0004638977635782748,
+      "loss": 0.2856,
+      "step": 7260
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0004645367412140575,
+      "loss": 0.249,
+      "step": 7270
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.0004651757188498403,
+      "loss": 0.2779,
+      "step": 7280
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.000465814696485623,
+      "loss": 0.2317,
+      "step": 7290
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00046645367412140575,
+      "loss": 0.2889,
+      "step": 7300
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00046709265175718854,
+      "loss": 0.2711,
+      "step": 7310
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00046773162939297126,
+      "loss": 0.2467,
+      "step": 7320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.000468370607028754,
+      "loss": 0.2422,
+      "step": 7330
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.0004690095846645367,
+      "loss": 0.2018,
+      "step": 7340
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.0004696485623003195,
+      "loss": 0.2285,
+      "step": 7350
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00047028753993610224,
+      "loss": 0.2277,
+      "step": 7360
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00047092651757188496,
+      "loss": 0.2734,
+      "step": 7370
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00047156549520766775,
+      "loss": 0.2438,
+      "step": 7380
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0004722044728434505,
+      "loss": 0.2769,
+      "step": 7390
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0004728434504792332,
+      "loss": 0.2448,
+      "step": 7400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00047348242811501594,
+      "loss": 0.2272,
+      "step": 7410
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0004741214057507987,
+      "loss": 0.2935,
+      "step": 7420
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0004747603833865815,
+      "loss": 0.2244,
+      "step": 7430
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00047539936102236423,
+      "loss": 0.2567,
+      "step": 7440
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.000476038338658147,
+      "loss": 0.2725,
+      "step": 7450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00047667731629392974,
+      "loss": 0.2912,
+      "step": 7460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0004773162939297125,
+      "loss": 0.2618,
+      "step": 7470
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00047795527156549526,
+      "loss": 0.2646,
+      "step": 7480
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.000478594249201278,
+      "loss": 0.2548,
+      "step": 7490
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.0004792332268370607,
+      "loss": 0.2626,
+      "step": 7500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00047987220447284344,
+      "loss": 0.2671,
+      "step": 7510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00048051118210862623,
+      "loss": 0.2704,
+      "step": 7520
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00048115015974440896,
+      "loss": 0.2659,
+      "step": 7530
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.0004817891373801917,
+      "loss": 0.2804,
+      "step": 7540
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00048242811501597447,
+      "loss": 0.2503,
+      "step": 7550
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.0004830670926517572,
+      "loss": 0.2489,
+      "step": 7560
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00048370607028753993,
+      "loss": 0.2837,
+      "step": 7570
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00048434504792332266,
+      "loss": 0.2364,
+      "step": 7580
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00048498402555910544,
+      "loss": 0.2508,
+      "step": 7590
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00048562300319488817,
+      "loss": 0.2279,
+      "step": 7600
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0004862619808306709,
+      "loss": 0.2174,
+      "step": 7610
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0004869009584664537,
+      "loss": 0.2837,
+      "step": 7620
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00048753993610223647,
+      "loss": 0.2414,
+      "step": 7630
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0004881789137380192,
+      "loss": 0.2452,
+      "step": 7640
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0004888178913738019,
+      "loss": 0.2854,
+      "step": 7650
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0004894568690095847,
+      "loss": 0.2593,
+      "step": 7660
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0004900958466453675,
+      "loss": 0.2507,
+      "step": 7670
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.0004907348242811502,
+      "loss": 0.2407,
+      "step": 7680
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.000491373801916933,
+      "loss": 0.2345,
+      "step": 7690
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.0004920127795527157,
+      "loss": 0.2786,
+      "step": 7700
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0004926517571884984,
+      "loss": 0.2639,
+      "step": 7710
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.0004932907348242811,
+      "loss": 0.2819,
+      "step": 7720
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0004939297124600639,
+      "loss": 0.2602,
+      "step": 7730
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0004945686900958467,
+      "loss": 0.2647,
+      "step": 7740
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.0004952076677316294,
+      "loss": 0.2126,
+      "step": 7750
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0004958466453674122,
+      "loss": 0.2694,
+      "step": 7760
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.0004964856230031949,
+      "loss": 0.2193,
+      "step": 7770
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.0004971246006389776,
+      "loss": 0.2219,
+      "step": 7780
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0004977635782747603,
+      "loss": 0.2674,
+      "step": 7790
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0004984025559105431,
+      "loss": 0.2368,
+      "step": 7800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0004990415335463259,
+      "loss": 0.2356,
+      "step": 7810
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0004996805111821086,
+      "loss": 0.2548,
+      "step": 7820
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0005003194888178914,
+      "loss": 0.277,
+      "step": 7830
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.0005009584664536741,
+      "loss": 0.2186,
+      "step": 7840
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0005015974440894568,
+      "loss": 0.2285,
+      "step": 7850
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 0.0005022364217252397,
+      "loss": 0.2198,
+      "step": 7860
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.0005028753993610223,
+      "loss": 0.1782,
+      "step": 7870
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0005035143769968051,
+      "loss": 0.2177,
+      "step": 7880
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.0005041533546325879,
+      "loss": 0.2345,
+      "step": 7890
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0005047923322683706,
+      "loss": 0.2268,
+      "step": 7900
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.0005054313099041533,
+      "loss": 0.2024,
+      "step": 7910
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.0005060702875399362,
+      "loss": 0.2183,
+      "step": 7920
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005067092651757189,
+      "loss": 0.2126,
+      "step": 7930
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 0.0005073482428115016,
+      "loss": 0.241,
+      "step": 7940
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0005079872204472845,
+      "loss": 0.2137,
+      "step": 7950
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0005086261980830671,
+      "loss": 0.2156,
+      "step": 7960
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.0005092651757188499,
+      "loss": 0.2081,
+      "step": 7970
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.0005099041533546325,
+      "loss": 0.2235,
+      "step": 7980
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005105431309904154,
+      "loss": 0.2023,
+      "step": 7990
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 0.0005111821086261981,
+      "loss": 0.2001,
+      "step": 8000
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0005118210862619808,
+      "loss": 0.1899,
+      "step": 8010
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0005124600638977636,
+      "loss": 0.2328,
+      "step": 8020
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.0005130990415335464,
+      "loss": 0.222,
+      "step": 8030
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.000513738019169329,
+      "loss": 0.2191,
+      "step": 8040
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.0005143769968051119,
+      "loss": 0.2012,
+      "step": 8050
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.0005150159744408946,
+      "loss": 0.2093,
+      "step": 8060
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.0005156549520766773,
+      "loss": 0.192,
+      "step": 8070
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 0.00051629392971246,
+      "loss": 0.2442,
+      "step": 8080
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.0005169329073482429,
+      "loss": 0.2149,
+      "step": 8090
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005175718849840255,
+      "loss": 0.2278,
+      "step": 8100
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.0005182108626198083,
+      "loss": 0.2626,
+      "step": 8110
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0005188498402555911,
+      "loss": 0.2234,
+      "step": 8120
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.0005194888178913738,
+      "loss": 0.2177,
+      "step": 8130
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 0.0005201277955271566,
+      "loss": 0.264,
+      "step": 8140
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005207667731629393,
+      "loss": 0.2129,
+      "step": 8150
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0005214057507987221,
+      "loss": 0.2431,
+      "step": 8160
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.0005220447284345048,
+      "loss": 0.2204,
+      "step": 8170
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0005226837060702875,
+      "loss": 0.2081,
+      "step": 8180
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.0005233226837060703,
+      "loss": 0.207,
+      "step": 8190
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0005239616613418531,
+      "loss": 0.2233,
+      "step": 8200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005246006389776357,
+      "loss": 0.2488,
+      "step": 8210
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 0.0005252396166134186,
+      "loss": 0.207,
+      "step": 8220
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.0005258785942492013,
+      "loss": 0.2447,
+      "step": 8230
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.000526517571884984,
+      "loss": 0.2295,
+      "step": 8240
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.0005271565495207668,
+      "loss": 0.2539,
+      "step": 8250
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0005277955271565496,
+      "loss": 0.243,
+      "step": 8260
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.0005284345047923322,
+      "loss": 0.2116,
+      "step": 8270
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0005290734824281151,
+      "loss": 0.2204,
+      "step": 8280
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0005297124600638977,
+      "loss": 0.2041,
+      "step": 8290
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 0.0005303514376996805,
+      "loss": 0.2402,
+      "step": 8300
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.0005309904153354632,
+      "loss": 0.2276,
+      "step": 8310
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.000531629392971246,
+      "loss": 0.2305,
+      "step": 8320
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.0005322683706070288,
+      "loss": 0.2257,
+      "step": 8330
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.0005329073482428115,
+      "loss": 0.2062,
+      "step": 8340
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.0005335463258785943,
+      "loss": 0.2013,
+      "step": 8350
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 0.000534185303514377,
+      "loss": 0.2429,
+      "step": 8360
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0005348242811501598,
+      "loss": 0.1853,
+      "step": 8370
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.0005354632587859425,
+      "loss": 0.2314,
+      "step": 8380
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.0005361022364217253,
+      "loss": 0.2033,
+      "step": 8390
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.000536741214057508,
+      "loss": 0.2227,
+      "step": 8400
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0005373801916932908,
+      "loss": 0.2243,
+      "step": 8410
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.0005380191693290735,
+      "loss": 0.2322,
+      "step": 8420
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0005386581469648563,
+      "loss": 0.2103,
+      "step": 8430
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 0.0005392971246006389,
+      "loss": 0.2361,
+      "step": 8440
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.0005399361022364218,
+      "loss": 0.2518,
+      "step": 8450
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0005405750798722044,
+      "loss": 0.2079,
+      "step": 8460
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.0005412140575079872,
+      "loss": 0.2353,
+      "step": 8470
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.00054185303514377,
+      "loss": 0.2471,
+      "step": 8480
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0005424920127795527,
+      "loss": 0.2506,
+      "step": 8490
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 0.0005431309904153354,
+      "loss": 0.2277,
+      "step": 8500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.0005437699680511183,
+      "loss": 0.2754,
+      "step": 8510
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.000544408945686901,
+      "loss": 0.2325,
+      "step": 8520
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.0005450479233226837,
+      "loss": 0.2329,
+      "step": 8530
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0005456869009584666,
+      "loss": 0.2542,
+      "step": 8540
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.0005463258785942492,
+      "loss": 0.2054,
+      "step": 8550
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.000546964856230032,
+      "loss": 0.2109,
+      "step": 8560
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0005476038338658147,
+      "loss": 0.2119,
+      "step": 8570
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 0.0005482428115015975,
+      "loss": 0.2293,
+      "step": 8580
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.0005488817891373802,
+      "loss": 0.2151,
+      "step": 8590
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0005495207667731629,
+      "loss": 0.2247,
+      "step": 8600
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.0005501597444089457,
+      "loss": 0.2787,
+      "step": 8610
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005507987220447285,
+      "loss": 0.2135,
+      "step": 8620
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0005514376996805111,
+      "loss": 0.2429,
+      "step": 8630
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 0.000552076677316294,
+      "loss": 0.224,
+      "step": 8640
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0005527156549520767,
+      "loss": 0.2559,
+      "step": 8650
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.0005533546325878594,
+      "loss": 0.2006,
+      "step": 8660
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.0005539936102236421,
+      "loss": 0.2098,
+      "step": 8670
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.000554632587859425,
+      "loss": 0.2211,
+      "step": 8680
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.0005552715654952076,
+      "loss": 0.2151,
+      "step": 8690
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0005559105431309904,
+      "loss": 0.2487,
+      "step": 8700
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0005565495207667732,
+      "loss": 0.1934,
+      "step": 8710
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 0.0005571884984025559,
+      "loss": 0.237,
+      "step": 8720
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005578274760383387,
+      "loss": 0.2356,
+      "step": 8730
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.0005584664536741214,
+      "loss": 0.242,
+      "step": 8740
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.0005591054313099042,
+      "loss": 0.2187,
+      "step": 8750
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0005597444089456869,
+      "loss": 0.2434,
+      "step": 8760
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0005603833865814697,
+      "loss": 0.2641,
+      "step": 8770
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 0.0005610223642172524,
+      "loss": 0.2223,
+      "step": 8780
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0005616613418530352,
+      "loss": 0.2159,
+      "step": 8790
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.0005623003194888178,
+      "loss": 0.233,
+      "step": 8800
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.0005629392971246007,
+      "loss": 0.2623,
+      "step": 8810
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0005635782747603834,
+      "loss": 0.2424,
+      "step": 8820
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.0005642172523961661,
+      "loss": 0.236,
+      "step": 8830
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005648562300319489,
+      "loss": 0.2372,
+      "step": 8840
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0005654952076677317,
+      "loss": 0.2165,
+      "step": 8850
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 0.0005661341853035143,
+      "loss": 0.2702,
+      "step": 8860
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0005667731629392972,
+      "loss": 0.2253,
+      "step": 8870
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.0005674121405750799,
+      "loss": 0.2361,
+      "step": 8880
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0005680511182108626,
+      "loss": 0.2314,
+      "step": 8890
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005686900958466453,
+      "loss": 0.2188,
+      "step": 8900
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.0005693290734824281,
+      "loss": 0.2121,
+      "step": 8910
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 0.0005699680511182109,
+      "loss": 0.208,
+      "step": 8920
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0005706070287539936,
+      "loss": 0.2587,
+      "step": 8930
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.0005712460063897764,
+      "loss": 0.248,
+      "step": 8940
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.0005718849840255591,
+      "loss": 0.2321,
+      "step": 8950
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0005725239616613419,
+      "loss": 0.2464,
+      "step": 8960
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0005731629392971246,
+      "loss": 0.2344,
+      "step": 8970
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0005738019169329074,
+      "loss": 0.2346,
+      "step": 8980
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.0005744408945686901,
+      "loss": 0.2296,
+      "step": 8990
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 0.0005750798722044729,
+      "loss": 0.2556,
+      "step": 9000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005757188498402556,
+      "loss": 0.2424,
+      "step": 9010
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.0005763578274760384,
+      "loss": 0.2561,
+      "step": 9020
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.000576996805111821,
+      "loss": 0.2339,
+      "step": 9030
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0005776357827476039,
+      "loss": 0.2646,
+      "step": 9040
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.0005782747603833865,
+      "loss": 0.2297,
+      "step": 9050
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.0005789137380191693,
+      "loss": 0.1751,
+      "step": 9060
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005795527156549521,
+      "loss": 0.2185,
+      "step": 9070
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 0.0005801916932907348,
+      "loss": 0.2366,
+      "step": 9080
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0005808306709265175,
+      "loss": 0.2377,
+      "step": 9090
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.0005814696485623004,
+      "loss": 0.2242,
+      "step": 9100
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.000582108626198083,
+      "loss": 0.2487,
+      "step": 9110
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005827476038338658,
+      "loss": 0.2498,
+      "step": 9120
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.0005833865814696487,
+      "loss": 0.21,
+      "step": 9130
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 0.0005840255591054313,
+      "loss": 0.2414,
+      "step": 9140
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0005846645367412141,
+      "loss": 0.2372,
+      "step": 9150
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.0005853035143769969,
+      "loss": 0.2066,
+      "step": 9160
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0005859424920127796,
+      "loss": 0.2338,
+      "step": 9170
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005865814696485623,
+      "loss": 0.2372,
+      "step": 9180
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0005872204472843451,
+      "loss": 0.2528,
+      "step": 9190
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0005878594249201278,
+      "loss": 0.2418,
+      "step": 9200
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.0005884984025559106,
+      "loss": 0.2089,
+      "step": 9210
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 0.0005891373801916932,
+      "loss": 0.2637,
+      "step": 9220
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005897763578274761,
+      "loss": 0.262,
+      "step": 9230
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.0005904153354632588,
+      "loss": 0.2451,
+      "step": 9240
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.0005910543130990415,
+      "loss": 0.2283,
+      "step": 9250
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0005916932907348243,
+      "loss": 0.227,
+      "step": 9260
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.0005923322683706071,
+      "loss": 0.2483,
+      "step": 9270
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 0.0005929712460063897,
+      "loss": 0.2338,
+      "step": 9280
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005936102236421725,
+      "loss": 0.2519,
+      "step": 9290
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.0005942492012779553,
+      "loss": 0.2135,
+      "step": 9300
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.000594888178913738,
+      "loss": 0.2144,
+      "step": 9310
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.0005955271565495208,
+      "loss": 0.2324,
+      "step": 9320
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.0005961661341853036,
+      "loss": 0.2121,
+      "step": 9330
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.0005968051118210863,
+      "loss": 0.2149,
+      "step": 9340
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.000597444089456869,
+      "loss": 0.2414,
+      "step": 9350
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 0.0005980830670926518,
+      "loss": 0.2348,
+      "step": 9360
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0005987220447284345,
+      "loss": 0.2711,
+      "step": 9370
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.0005993610223642173,
+      "loss": 0.2151,
+      "step": 9380
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.0006,
+      "loss": 0.2554,
+      "step": 9390
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0006006389776357828,
+      "loss": 0.1843,
+      "step": 9400
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.0006012779552715655,
+      "loss": 0.2518,
+      "step": 9410
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 0.0006019169329073482,
+      "loss": 0.2017,
+      "step": 9420
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.000602555910543131,
+      "loss": 0.2016,
+      "step": 9430
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.0006031948881789138,
+      "loss": 0.1955,
+      "step": 9440
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0006038338658146964,
+      "loss": 0.2106,
+      "step": 9450
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.0006044728434504793,
+      "loss": 0.2085,
+      "step": 9460
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.000605111821086262,
+      "loss": 0.2208,
+      "step": 9470
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0006057507987220447,
+      "loss": 0.2107,
+      "step": 9480
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.0006063897763578275,
+      "loss": 0.2016,
+      "step": 9490
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 0.0006070287539936102,
+      "loss": 0.2203,
+      "step": 9500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.000607667731629393,
+      "loss": 0.1773,
+      "step": 9510
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.0006083067092651758,
+      "loss": 0.2162,
+      "step": 9520
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0006089456869009585,
+      "loss": 0.1805,
+      "step": 9530
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.0006095846645367412,
+      "loss": 0.275,
+      "step": 9540
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.000610223642172524,
+      "loss": 0.2398,
+      "step": 9550
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0006108626198083067,
+      "loss": 0.1857,
+      "step": 9560
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 0.0006115015974440895,
+      "loss": 0.2261,
+      "step": 9570
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.0006121405750798722,
+      "loss": 0.1816,
+      "step": 9580
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.000612779552715655,
+      "loss": 0.212,
+      "step": 9590
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.0006134185303514377,
+      "loss": 0.2027,
+      "step": 9600
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.0006140575079872205,
+      "loss": 0.2054,
+      "step": 9610
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0006146964856230032,
+      "loss": 0.2073,
+      "step": 9620
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.000615335463258786,
+      "loss": 0.199,
+      "step": 9630
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 0.0006159744408945687,
+      "loss": 0.2037,
+      "step": 9640
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0006166134185303514,
+      "loss": 0.1966,
+      "step": 9650
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.0006172523961661342,
+      "loss": 0.2098,
+      "step": 9660
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.0006178913738019169,
+      "loss": 0.2208,
+      "step": 9670
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0006185303514376996,
+      "loss": 0.1932,
+      "step": 9680
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.0006191693290734825,
+      "loss": 0.2283,
+      "step": 9690
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0006198083067092651,
+      "loss": 0.2247,
+      "step": 9700
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 0.0006204472843450479,
+      "loss": 0.2131,
+      "step": 9710
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.0006210862619808308,
+      "loss": 0.1977,
+      "step": 9720
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0006217252396166134,
+      "loss": 0.2284,
+      "step": 9730
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.0006223642172523962,
+      "loss": 0.2088,
+      "step": 9740
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.000623003194888179,
+      "loss": 0.238,
+      "step": 9750
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0006236421725239617,
+      "loss": 0.198,
+      "step": 9760
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.0006242811501597444,
+      "loss": 0.2132,
+      "step": 9770
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 0.0006249201277955273,
+      "loss": 0.2107,
+      "step": 9780
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0006255591054313099,
+      "loss": 0.1984,
+      "step": 9790
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0006261980830670927,
+      "loss": 0.2316,
+      "step": 9800
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0006268370607028753,
+      "loss": 0.2404,
+      "step": 9810
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.0006274760383386582,
+      "loss": 0.227,
+      "step": 9820
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.0006281150159744409,
+      "loss": 0.1707,
+      "step": 9830
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0006287539936102236,
+      "loss": 0.2225,
+      "step": 9840
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.0006293929712460064,
+      "loss": 0.2663,
+      "step": 9850
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 0.0006300319488817892,
+      "loss": 0.2101,
+      "step": 9860
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0006306709265175718,
+      "loss": 0.2007,
+      "step": 9870
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.0006313099041533547,
+      "loss": 0.2094,
+      "step": 9880
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.0006319488817891374,
+      "loss": 0.2393,
+      "step": 9890
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.0006325878594249201,
+      "loss": 0.2459,
+      "step": 9900
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.000633226837060703,
+      "loss": 0.195,
+      "step": 9910
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0006338658146964857,
+      "loss": 0.1721,
+      "step": 9920
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 0.0006345047923322684,
+      "loss": 0.2254,
+      "step": 9930
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.0006351437699680511,
+      "loss": 0.2252,
+      "step": 9940
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.000635782747603834,
+      "loss": 0.1812,
+      "step": 9950
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.0006364217252396166,
+      "loss": 0.2133,
+      "step": 9960
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.0006370607028753994,
+      "loss": 0.2453,
+      "step": 9970
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.000637699680511182,
+      "loss": 0.2257,
+      "step": 9980
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.0006383386581469649,
+      "loss": 0.2017,
+      "step": 9990
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 0.0006389776357827476,
+      "loss": 0.2141,
+      "step": 10000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0006396166134185303,
+      "loss": 0.221,
+      "step": 10010
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0006402555910543131,
+      "loss": 0.2082,
+      "step": 10020
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.0006408945686900959,
+      "loss": 0.2305,
+      "step": 10030
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0006415335463258785,
+      "loss": 0.1989,
+      "step": 10040
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.0006421725239616614,
+      "loss": 0.2114,
+      "step": 10050
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0006428115015974441,
+      "loss": 0.244,
+      "step": 10060
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 0.0006434504792332268,
+      "loss": 0.2128,
+      "step": 10070
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.0006440894568690096,
+      "loss": 0.2411,
+      "step": 10080
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.0006447284345047924,
+      "loss": 0.2089,
+      "step": 10090
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.000645367412140575,
+      "loss": 0.2048,
+      "step": 10100
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.0006460063897763579,
+      "loss": 0.2081,
+      "step": 10110
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0006466453674121406,
+      "loss": 0.2285,
+      "step": 10120
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.0006472843450479233,
+      "loss": 0.217,
+      "step": 10130
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 0.0006479233226837062,
+      "loss": 0.2066,
+      "step": 10140
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0006485623003194888,
+      "loss": 0.2274,
+      "step": 10150
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.0006492012779552716,
+      "loss": 0.2116,
+      "step": 10160
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0006498402555910543,
+      "loss": 0.2406,
+      "step": 10170
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.0006504792332268371,
+      "loss": 0.2094,
+      "step": 10180
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.0006511182108626198,
+      "loss": 0.2406,
+      "step": 10190
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0006517571884984026,
+      "loss": 0.2134,
+      "step": 10200
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 0.0006523961661341853,
+      "loss": 0.2337,
+      "step": 10210
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.0006530351437699681,
+      "loss": 0.2169,
+      "step": 10220
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0006536741214057508,
+      "loss": 0.1976,
+      "step": 10230
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.0006543130990415336,
+      "loss": 0.2305,
+      "step": 10240
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.0006549520766773163,
+      "loss": 0.2315,
+      "step": 10250
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.000655591054313099,
+      "loss": 0.2116,
+      "step": 10260
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.0006562300319488817,
+      "loss": 0.2238,
+      "step": 10270
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0006568690095846646,
+      "loss": 0.1958,
+      "step": 10280
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 0.0006575079872204472,
+      "loss": 0.2142,
+      "step": 10290
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00065814696485623,
+      "loss": 0.2262,
+      "step": 10300
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0006587859424920129,
+      "loss": 0.2009,
+      "step": 10310
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.0006594249201277955,
+      "loss": 0.2405,
+      "step": 10320
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.0006600638977635783,
+      "loss": 0.239,
+      "step": 10330
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0006607028753993611,
+      "loss": 0.2234,
+      "step": 10340
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 0.0006613418530351438,
+      "loss": 0.2276,
+      "step": 10350
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0006619808306709265,
+      "loss": 0.2201,
+      "step": 10360
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.0006626198083067094,
+      "loss": 0.2277,
+      "step": 10370
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.000663258785942492,
+      "loss": 0.2399,
+      "step": 10380
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.0006638977635782748,
+      "loss": 0.2097,
+      "step": 10390
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0006645367412140575,
+      "loss": 0.2211,
+      "step": 10400
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.0006651757188498403,
+      "loss": 0.2249,
+      "step": 10410
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.000665814696485623,
+      "loss": 0.223,
+      "step": 10420
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 0.0006664536741214057,
+      "loss": 0.2492,
+      "step": 10430
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.0006670926517571885,
+      "loss": 0.1998,
+      "step": 10440
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0006677316293929713,
+      "loss": 0.2508,
+      "step": 10450
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.0006683706070287539,
+      "loss": 0.2545,
+      "step": 10460
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.0006690095846645368,
+      "loss": 0.2027,
+      "step": 10470
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0006696485623003195,
+      "loss": 0.2344,
+      "step": 10480
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 0.0006702875399361022,
+      "loss": 0.2268,
+      "step": 10490
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.0006709265175718851,
+      "loss": 0.2414,
+      "step": 10500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0006715654952076678,
+      "loss": 0.2263,
+      "step": 10510
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.0006722044728434505,
+      "loss": 0.1974,
+      "step": 10520
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0006728434504792332,
+      "loss": 0.2196,
+      "step": 10530
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.0006734824281150161,
+      "loss": 0.1958,
+      "step": 10540
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0006741214057507987,
+      "loss": 0.2002,
+      "step": 10550
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0006747603833865815,
+      "loss": 0.2186,
+      "step": 10560
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 0.0006753993610223642,
+      "loss": 0.2254,
+      "step": 10570
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.000676038338658147,
+      "loss": 0.1954,
+      "step": 10580
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0006766773162939297,
+      "loss": 0.2413,
+      "step": 10590
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.0006773162939297125,
+      "loss": 0.2373,
+      "step": 10600
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.0006779552715654952,
+      "loss": 0.2389,
+      "step": 10610
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.000678594249201278,
+      "loss": 0.203,
+      "step": 10620
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.0006792332268370606,
+      "loss": 0.2357,
+      "step": 10630
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0006798722044728435,
+      "loss": 0.2449,
+      "step": 10640
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 0.0006805111821086262,
+      "loss": 0.2056,
+      "step": 10650
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.0006811501597444089,
+      "loss": 0.2524,
+      "step": 10660
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0006817891373801917,
+      "loss": 0.2157,
+      "step": 10670
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.0006824281150159745,
+      "loss": 0.1798,
+      "step": 10680
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.0006830670926517571,
+      "loss": 0.1969,
+      "step": 10690
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.00068370607028754,
+      "loss": 0.2223,
+      "step": 10700
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 0.0006843450479233228,
+      "loss": 0.2193,
+      "step": 10710
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.0006849840255591054,
+      "loss": 0.2223,
+      "step": 10720
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0006856230031948883,
+      "loss": 0.2113,
+      "step": 10730
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.0006862619808306709,
+      "loss": 0.2488,
+      "step": 10740
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.0006869009584664537,
+      "loss": 0.2308,
+      "step": 10750
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0006875399361022364,
+      "loss": 0.2283,
+      "step": 10760
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0006881789137380192,
+      "loss": 0.2672,
+      "step": 10770
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0006888178913738019,
+      "loss": 0.2538,
+      "step": 10780
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 0.0006894568690095847,
+      "loss": 0.2907,
+      "step": 10790
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.0006900958466453674,
+      "loss": 0.1976,
+      "step": 10800
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0006907348242811502,
+      "loss": 0.2619,
+      "step": 10810
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.0006913738019169329,
+      "loss": 0.1975,
+      "step": 10820
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0006920127795527157,
+      "loss": 0.1838,
+      "step": 10830
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0006926517571884984,
+      "loss": 0.2242,
+      "step": 10840
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 0.0006932907348242812,
+      "loss": 0.2057,
+      "step": 10850
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.0006939297124600638,
+      "loss": 0.2116,
+      "step": 10860
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0006945686900958467,
+      "loss": 0.2289,
+      "step": 10870
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.0006952076677316293,
+      "loss": 0.202,
+      "step": 10880
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.0006958466453674121,
+      "loss": 0.2221,
+      "step": 10890
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.000696485623003195,
+      "loss": 0.237,
+      "step": 10900
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.0006971246006389776,
+      "loss": 0.3199,
+      "step": 10910
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0006977635782747604,
+      "loss": 0.2029,
+      "step": 10920
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 0.0006984025559105432,
+      "loss": 0.223,
+      "step": 10930
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.0006990415335463259,
+      "loss": 0.216,
+      "step": 10940
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0006996805111821086,
+      "loss": 0.1952,
+      "step": 10950
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.0007003194888178915,
+      "loss": 0.2173,
+      "step": 10960
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0007009584664536741,
+      "loss": 0.1895,
+      "step": 10970
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0007015974440894569,
+      "loss": 0.1721,
+      "step": 10980
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 0.0007022364217252397,
+      "loss": 0.1872,
+      "step": 10990
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.0007028753993610224,
+      "loss": 0.1846,
+      "step": 11000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0007035143769968051,
+      "loss": 0.207,
+      "step": 11010
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.0007041533546325878,
+      "loss": 0.2032,
+      "step": 11020
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0007047923322683706,
+      "loss": 0.1662,
+      "step": 11030
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.0007054313099041534,
+      "loss": 0.167,
+      "step": 11040
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.000706070287539936,
+      "loss": 0.2351,
+      "step": 11050
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0007067092651757189,
+      "loss": 0.1653,
+      "step": 11060
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 0.0007073482428115016,
+      "loss": 0.1826,
+      "step": 11070
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.0007079872204472843,
+      "loss": 0.2162,
+      "step": 11080
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.000708626198083067,
+      "loss": 0.1853,
+      "step": 11090
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.0007092651757188499,
+      "loss": 0.1865,
+      "step": 11100
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.0007099041533546326,
+      "loss": 0.1868,
+      "step": 11110
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0007105431309904153,
+      "loss": 0.1628,
+      "step": 11120
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 0.0007111821086261982,
+      "loss": 0.1975,
+      "step": 11130
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0007118210862619808,
+      "loss": 0.1853,
+      "step": 11140
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.0007124600638977636,
+      "loss": 0.2046,
+      "step": 11150
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.0007130990415335464,
+      "loss": 0.1966,
+      "step": 11160
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0007137380191693291,
+      "loss": 0.2022,
+      "step": 11170
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0007143769968051118,
+      "loss": 0.1635,
+      "step": 11180
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.0007150159744408946,
+      "loss": 0.2223,
+      "step": 11190
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0007156549520766773,
+      "loss": 0.2094,
+      "step": 11200
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 0.0007162939297124601,
+      "loss": 0.1938,
+      "step": 11210
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.0007169329073482428,
+      "loss": 0.2053,
+      "step": 11220
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0007175718849840256,
+      "loss": 0.1981,
+      "step": 11230
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.0007182108626198083,
+      "loss": 0.1969,
+      "step": 11240
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.000718849840255591,
+      "loss": 0.1912,
+      "step": 11250
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.0007194888178913738,
+      "loss": 0.1875,
+      "step": 11260
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 0.0007201277955271566,
+      "loss": 0.1944,
+      "step": 11270
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0007207667731629392,
+      "loss": 0.2254,
+      "step": 11280
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.0007214057507987221,
+      "loss": 0.2204,
+      "step": 11290
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.0007220447284345049,
+      "loss": 0.2089,
+      "step": 11300
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0007226837060702875,
+      "loss": 0.2098,
+      "step": 11310
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0007233226837060704,
+      "loss": 0.2107,
+      "step": 11320
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.000723961661341853,
+      "loss": 0.2035,
+      "step": 11330
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0007246006389776358,
+      "loss": 0.2098,
+      "step": 11340
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 0.0007252396166134186,
+      "loss": 0.2332,
+      "step": 11350
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.0007258785942492013,
+      "loss": 0.1978,
+      "step": 11360
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.000726517571884984,
+      "loss": 0.1873,
+      "step": 11370
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.0007271565495207669,
+      "loss": 0.1933,
+      "step": 11380
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0007277955271565495,
+      "loss": 0.2329,
+      "step": 11390
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0007284345047923323,
+      "loss": 0.1914,
+      "step": 11400
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.000729073482428115,
+      "loss": 0.1788,
+      "step": 11410
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0007297124600638978,
+      "loss": 0.2233,
+      "step": 11420
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 0.0007303514376996805,
+      "loss": 0.2045,
+      "step": 11430
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.0007309904153354633,
+      "loss": 0.1945,
+      "step": 11440
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.000731629392971246,
+      "loss": 0.2035,
+      "step": 11450
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.0007322683706070288,
+      "loss": 0.1888,
+      "step": 11460
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.0007329073482428114,
+      "loss": 0.2102,
+      "step": 11470
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0007335463258785943,
+      "loss": 0.2422,
+      "step": 11480
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 0.0007341853035143771,
+      "loss": 0.2067,
+      "step": 11490
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0007348242811501597,
+      "loss": 0.2258,
+      "step": 11500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.0007354632587859425,
+      "loss": 0.1953,
+      "step": 11510
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.0007361022364217253,
+      "loss": 0.2008,
+      "step": 11520
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.000736741214057508,
+      "loss": 0.1799,
+      "step": 11530
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0007373801916932907,
+      "loss": 0.1961,
+      "step": 11540
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.0007380191693290736,
+      "loss": 0.2147,
+      "step": 11550
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.0007386581469648562,
+      "loss": 0.2173,
+      "step": 11560
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 0.000739297124600639,
+      "loss": 0.2015,
+      "step": 11570
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.0007399361022364218,
+      "loss": 0.2245,
+      "step": 11580
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0007405750798722045,
+      "loss": 0.2001,
+      "step": 11590
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0007412140575079872,
+      "loss": 0.2022,
+      "step": 11600
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0007418530351437701,
+      "loss": 0.2094,
+      "step": 11610
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.0007424920127795527,
+      "loss": 0.2429,
+      "step": 11620
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 0.0007431309904153355,
+      "loss": 0.174,
+      "step": 11630
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.0007437699680511181,
+      "loss": 0.2271,
+      "step": 11640
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.000744408945686901,
+      "loss": 0.1932,
+      "step": 11650
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.0007450479233226837,
+      "loss": 0.1983,
+      "step": 11660
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0007456869009584664,
+      "loss": 0.1996,
+      "step": 11670
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.0007463258785942492,
+      "loss": 0.2084,
+      "step": 11680
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.000746964856230032,
+      "loss": 0.2157,
+      "step": 11690
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0007476038338658147,
+      "loss": 0.2366,
+      "step": 11700
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 0.0007482428115015975,
+      "loss": 0.2104,
+      "step": 11710
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.0007488817891373803,
+      "loss": 0.2164,
+      "step": 11720
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0007495207667731629,
+      "loss": 0.1951,
+      "step": 11730
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0007501597444089458,
+      "loss": 0.2207,
+      "step": 11740
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0007507987220447285,
+      "loss": 0.1786,
+      "step": 11750
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 0.0007514376996805112,
+      "loss": 0.2087,
+      "step": 11760
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 0.0007520766773162939,
+      "loss": 0.2059,
+      "step": 11770
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0007527156549520767,
+      "loss": 0.2304,
+      "step": 11780
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 0.0007533546325878594,
+      "loss": 0.2173,
+      "step": 11790
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 0.0007539936102236422,
+      "loss": 0.2221,
+      "step": 11800
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0007546325878594249,
+      "loss": 0.203,
+      "step": 11810
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 0.0007552715654952077,
+      "loss": 0.2087,
+      "step": 11820
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 0.0007559105431309904,
+      "loss": 0.2292,
+      "step": 11830
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0007565495207667732,
+      "loss": 0.2116,
+      "step": 11840
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 0.0007571884984025559,
+      "loss": 0.2227,
+      "step": 11850
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0007578274760383387,
+      "loss": 0.207,
+      "step": 11860
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 0.0007584664536741213,
+      "loss": 0.2051,
+      "step": 11870
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 0.0007591054313099042,
+      "loss": 0.2409,
+      "step": 11880
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.000759744408945687,
+      "loss": 0.1849,
+      "step": 11890
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 0.0007603833865814696,
+      "loss": 0.2098,
+      "step": 11900
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 0.0007610223642172525,
+      "loss": 0.2225,
+      "step": 11910
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0007616613418530352,
+      "loss": 0.1973,
+      "step": 11920
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 0.0007623003194888179,
+      "loss": 0.239,
+      "step": 11930
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 0.0007629392971246007,
+      "loss": 0.2186,
+      "step": 11940
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0007635782747603834,
+      "loss": 0.1849,
+      "step": 11950
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 0.0007642172523961661,
+      "loss": 0.2316,
+      "step": 11960
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.000764856230031949,
+      "loss": 0.2223,
+      "step": 11970
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 0.0007654952076677316,
+      "loss": 0.2457,
+      "step": 11980
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 0.0007661341853035144,
+      "loss": 0.2006,
+      "step": 11990
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0007667731629392971,
+      "loss": 0.2142,
+      "step": 12000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 0.0007674121405750799,
+      "loss": 0.1964,
+      "step": 12010
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 0.0007680511182108626,
+      "loss": 0.2215,
+      "step": 12020
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0007686900958466454,
+      "loss": 0.2158,
+      "step": 12030
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0007693290734824281,
+      "loss": 0.2071,
+      "step": 12040
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 0.0007699680511182109,
+      "loss": 0.221,
+      "step": 12050
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0007706070287539936,
+      "loss": 0.2322,
+      "step": 12060
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 0.0007712460063897764,
+      "loss": 0.2176,
+      "step": 12070
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 0.0007718849840255591,
+      "loss": 0.1703,
+      "step": 12080
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0007725239616613418,
+      "loss": 0.2264,
+      "step": 12090
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 0.0007731629392971247,
+      "loss": 0.2123,
+      "step": 12100
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0007738019169329074,
+      "loss": 0.2096,
+      "step": 12110
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 0.0007744408945686901,
+      "loss": 0.228,
+      "step": 12120
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 0.0007750798722044728,
+      "loss": 0.2193,
+      "step": 12130
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0007757188498402557,
+      "loss": 0.2062,
+      "step": 12140
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 0.0007763578274760383,
+      "loss": 0.1992,
+      "step": 12150
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 0.0007769968051118211,
+      "loss": 0.1922,
+      "step": 12160
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0007776357827476039,
+      "loss": 0.1991,
+      "step": 12170
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 0.0007782747603833866,
+      "loss": 0.2427,
+      "step": 12180
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 0.0007789137380191693,
+      "loss": 0.2242,
+      "step": 12190
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0007795527156549522,
+      "loss": 0.2559,
+      "step": 12200
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 0.0007801916932907348,
+      "loss": 0.2238,
+      "step": 12210
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0007808306709265176,
+      "loss": 0.2055,
+      "step": 12220
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 0.0007814696485623002,
+      "loss": 0.2139,
+      "step": 12230
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 0.0007821086261980831,
+      "loss": 0.2198,
+      "step": 12240
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0007827476038338658,
+      "loss": 0.2401,
+      "step": 12250
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 0.0007833865814696485,
+      "loss": 0.2253,
+      "step": 12260
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 0.0007840255591054313,
+      "loss": 0.2157,
+      "step": 12270
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0007846645367412141,
+      "loss": 0.2274,
+      "step": 12280
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 0.0007853035143769968,
+      "loss": 0.2281,
+      "step": 12290
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 0.0007859424920127796,
+      "loss": 0.1965,
+      "step": 12300
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.0007865814696485624,
+      "loss": 0.2179,
+      "step": 12310
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 0.000787220447284345,
+      "loss": 0.1826,
+      "step": 12320
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0007878594249201279,
+      "loss": 0.172,
+      "step": 12330
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 0.0007884984025559106,
+      "loss": 0.2562,
+      "step": 12340
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 0.0007891373801916933,
+      "loss": 0.2094,
+      "step": 12350
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.000789776357827476,
+      "loss": 0.1897,
+      "step": 12360
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 0.0007904153354632589,
+      "loss": 0.2108,
+      "step": 12370
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 0.0007910543130990415,
+      "loss": 0.21,
+      "step": 12380
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.0007916932907348243,
+      "loss": 0.2316,
+      "step": 12390
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 0.000792332268370607,
+      "loss": 0.1809,
+      "step": 12400
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 0.0007929712460063898,
+      "loss": 0.2491,
+      "step": 12410
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0007936102236421725,
+      "loss": 0.1887,
+      "step": 12420
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 0.0007942492012779553,
+      "loss": 0.2278,
+      "step": 12430
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.000794888178913738,
+      "loss": 0.2504,
+      "step": 12440
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 0.0007955271565495208,
+      "loss": 0.2111,
+      "step": 12450
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 0.0007961661341853034,
+      "loss": 0.2649,
+      "step": 12460
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.0007968051118210863,
+      "loss": 0.2035,
+      "step": 12470
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 0.000797444089456869,
+      "loss": 0.2132,
+      "step": 12480
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 0.0007980830670926517,
+      "loss": 0.2869,
+      "step": 12490
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0007987220447284346,
+      "loss": 0.261,
+      "step": 12500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 0.0007993610223642173,
+      "loss": 0.1863,
+      "step": 12510
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 0.0008,
+      "loss": 0.2178,
+      "step": 12520
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0008006389776357828,
+      "loss": 0.1807,
+      "step": 12530
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 0.0008012779552715655,
+      "loss": 0.1813,
+      "step": 12540
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 0.0008019169329073482,
+      "loss": 0.1791,
+      "step": 12550
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0008025559105431311,
+      "loss": 0.1472,
+      "step": 12560
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 0.0008031948881789137,
+      "loss": 0.18,
+      "step": 12570
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0008038338658146965,
+      "loss": 0.1829,
+      "step": 12580
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 0.0008044728434504793,
+      "loss": 0.1877,
+      "step": 12590
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 0.000805111821086262,
+      "loss": 0.2126,
+      "step": 12600
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.0008057507987220447,
+      "loss": 0.1921,
+      "step": 12610
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 0.0008063897763578275,
+      "loss": 0.1952,
+      "step": 12620
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 0.0008070287539936102,
+      "loss": 0.1775,
+      "step": 12630
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.000807667731629393,
+      "loss": 0.2038,
+      "step": 12640
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 0.0008083067092651757,
+      "loss": 0.2054,
+      "step": 12650
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 0.0008089456869009585,
+      "loss": 0.2258,
+      "step": 12660
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.0008095846645367412,
+      "loss": 0.2226,
+      "step": 12670
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 0.000810223642172524,
+      "loss": 0.1902,
+      "step": 12680
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0008108626198083068,
+      "loss": 0.1996,
+      "step": 12690
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 0.0008115015974440895,
+      "loss": 0.1904,
+      "step": 12700
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 0.0008121405750798722,
+      "loss": 0.1486,
+      "step": 12710
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.000812779552715655,
+      "loss": 0.1905,
+      "step": 12720
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 0.0008134185303514378,
+      "loss": 0.1883,
+      "step": 12730
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 0.0008140575079872204,
+      "loss": 0.1863,
+      "step": 12740
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.0008146964856230032,
+      "loss": 0.1934,
+      "step": 12750
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 0.000815335463258786,
+      "loss": 0.1778,
+      "step": 12760
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 0.0008159744408945687,
+      "loss": 0.1796,
+      "step": 12770
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0008166134185303514,
+      "loss": 0.1902,
+      "step": 12780
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 0.0008172523961661343,
+      "loss": 0.1913,
+      "step": 12790
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 0.0008178913738019169,
+      "loss": 0.1887,
+      "step": 12800
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0008185303514376997,
+      "loss": 0.1658,
+      "step": 12810
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 0.0008191693290734825,
+      "loss": 0.169,
+      "step": 12820
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.0008198083067092652,
+      "loss": 0.1946,
+      "step": 12830
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 0.0008204472843450479,
+      "loss": 0.1913,
+      "step": 12840
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 0.0008210862619808306,
+      "loss": 0.2135,
+      "step": 12850
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0008217252396166134,
+      "loss": 0.1913,
+      "step": 12860
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 0.0008223642172523962,
+      "loss": 0.1958,
+      "step": 12870
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 0.0008230031948881789,
+      "loss": 0.2013,
+      "step": 12880
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0008236421725239617,
+      "loss": 0.2246,
+      "step": 12890
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 0.0008242811501597445,
+      "loss": 0.2016,
+      "step": 12900
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 0.0008249201277955271,
+      "loss": 0.1932,
+      "step": 12910
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.00082555910543131,
+      "loss": 0.1766,
+      "step": 12920
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 0.0008261980830670927,
+      "loss": 0.1959,
+      "step": 12930
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0008268370607028754,
+      "loss": 0.1978,
+      "step": 12940
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 0.0008274760383386582,
+      "loss": 0.2187,
+      "step": 12950
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 0.000828115015974441,
+      "loss": 0.1971,
+      "step": 12960
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0008287539936102236,
+      "loss": 0.1839,
+      "step": 12970
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 0.0008293929712460064,
+      "loss": 0.2237,
+      "step": 12980
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 0.0008300319488817891,
+      "loss": 0.1553,
+      "step": 12990
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0008306709265175719,
+      "loss": 0.193,
+      "step": 13000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 0.0008313099041533546,
+      "loss": 0.2027,
+      "step": 13010
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 0.0008319488817891374,
+      "loss": 0.1927,
+      "step": 13020
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0008325878594249201,
+      "loss": 0.2067,
+      "step": 13030
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 0.0008332268370607029,
+      "loss": 0.2119,
+      "step": 13040
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0008338658146964856,
+      "loss": 0.2125,
+      "step": 13050
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 0.0008345047923322684,
+      "loss": 0.2148,
+      "step": 13060
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 0.0008351437699680511,
+      "loss": 0.1722,
+      "step": 13070
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.0008357827476038338,
+      "loss": 0.212,
+      "step": 13080
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 0.0008364217252396167,
+      "loss": 0.2077,
+      "step": 13090
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 0.0008370607028753994,
+      "loss": 0.2223,
+      "step": 13100
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0008376996805111821,
+      "loss": 0.2055,
+      "step": 13110
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 0.0008383386581469649,
+      "loss": 0.1995,
+      "step": 13120
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 0.0008389776357827477,
+      "loss": 0.176,
+      "step": 13130
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.0008396166134185303,
+      "loss": 0.2049,
+      "step": 13140
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 0.0008402555910543132,
+      "loss": 0.2177,
+      "step": 13150
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 0.0008408945686900958,
+      "loss": 0.1961,
+      "step": 13160
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0008415335463258786,
+      "loss": 0.2094,
+      "step": 13170
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 0.0008421725239616614,
+      "loss": 0.2121,
+      "step": 13180
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0008428115015974441,
+      "loss": 0.1996,
+      "step": 13190
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 0.0008434504792332268,
+      "loss": 0.1836,
+      "step": 13200
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 0.0008440894568690097,
+      "loss": 0.1804,
+      "step": 13210
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0008447284345047923,
+      "loss": 0.2604,
+      "step": 13220
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 0.0008453674121405751,
+      "loss": 0.2071,
+      "step": 13230
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.0008460063897763578,
+      "loss": 0.2113,
+      "step": 13240
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0008466453674121406,
+      "loss": 0.1961,
+      "step": 13250
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 0.0008472843450479233,
+      "loss": 0.2149,
+      "step": 13260
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 0.0008479233226837061,
+      "loss": 0.2112,
+      "step": 13270
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.0008485623003194889,
+      "loss": 0.2003,
+      "step": 13280
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 0.0008492012779552716,
+      "loss": 0.2098,
+      "step": 13290
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.0008498402555910543,
+      "loss": 0.1986,
+      "step": 13300
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 0.000850479233226837,
+      "loss": 0.2055,
+      "step": 13310
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 0.0008511182108626199,
+      "loss": 0.169,
+      "step": 13320
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0008517571884984025,
+      "loss": 0.2417,
+      "step": 13330
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 0.0008523961661341853,
+      "loss": 0.2163,
+      "step": 13340
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 0.0008530351437699681,
+      "loss": 0.1996,
+      "step": 13350
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0008536741214057508,
+      "loss": 0.1736,
+      "step": 13360
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 0.0008543130990415335,
+      "loss": 0.204,
+      "step": 13370
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 0.0008549520766773164,
+      "loss": 0.2189,
+      "step": 13380
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.000855591054313099,
+      "loss": 0.1797,
+      "step": 13390
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 0.0008562300319488818,
+      "loss": 0.1723,
+      "step": 13400
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0008568690095846646,
+      "loss": 0.2189,
+      "step": 13410
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 0.0008575079872204473,
+      "loss": 0.2004,
+      "step": 13420
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 0.00085814696485623,
+      "loss": 0.2102,
+      "step": 13430
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0008587859424920129,
+      "loss": 0.173,
+      "step": 13440
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 0.0008594249201277955,
+      "loss": 0.2067,
+      "step": 13450
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 0.0008600638977635783,
+      "loss": 0.2229,
+      "step": 13460
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.0008607028753993609,
+      "loss": 0.2101,
+      "step": 13470
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 0.0008613418530351438,
+      "loss": 0.1975,
+      "step": 13480
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 0.0008619808306709266,
+      "loss": 0.1821,
+      "step": 13490
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0008626198083067092,
+      "loss": 0.244,
+      "step": 13500
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 0.0008632587859424921,
+      "loss": 0.1859,
+      "step": 13510
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 0.0008638977635782748,
+      "loss": 0.178,
+      "step": 13520
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.0008645367412140575,
+      "loss": 0.2166,
+      "step": 13530
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 0.0008651757188498403,
+      "loss": 0.166,
+      "step": 13540
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0008658146964856231,
+      "loss": 0.1821,
+      "step": 13550
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 0.0008664536741214057,
+      "loss": 0.1787,
+      "step": 13560
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 0.0008670926517571886,
+      "loss": 0.1851,
+      "step": 13570
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.0008677316293929713,
+      "loss": 0.181,
+      "step": 13580
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 0.000868370607028754,
+      "loss": 0.1952,
+      "step": 13590
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 0.0008690095846645367,
+      "loss": 0.1873,
+      "step": 13600
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0008696485623003195,
+      "loss": 0.212,
+      "step": 13610
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 0.0008702875399361022,
+      "loss": 0.2072,
+      "step": 13620
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 0.000870926517571885,
+      "loss": 0.2092,
+      "step": 13630
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0008715654952076677,
+      "loss": 0.2119,
+      "step": 13640
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 0.0008722044728434505,
+      "loss": 0.2139,
+      "step": 13650
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.0008728434504792332,
+      "loss": 0.1895,
+      "step": 13660
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 0.000873482428115016,
+      "loss": 0.2178,
+      "step": 13670
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 0.0008741214057507988,
+      "loss": 0.229,
+      "step": 13680
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0008747603833865815,
+      "loss": 0.2104,
+      "step": 13690
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 0.0008753993610223643,
+      "loss": 0.2215,
+      "step": 13700
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 0.000876038338658147,
+      "loss": 0.1943,
+      "step": 13710
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.0008766773162939298,
+      "loss": 0.2193,
+      "step": 13720
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 0.0008773162939297124,
+      "loss": 0.2294,
+      "step": 13730
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 0.0008779552715654953,
+      "loss": 0.1998,
+      "step": 13740
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.0008785942492012779,
+      "loss": 0.2152,
+      "step": 13750
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 0.0008792332268370607,
+      "loss": 0.1832,
+      "step": 13760
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0008798722044728435,
+      "loss": 0.2159,
+      "step": 13770
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 0.0008805111821086262,
+      "loss": 0.2204,
+      "step": 13780
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 0.0008811501597444089,
+      "loss": 0.224,
+      "step": 13790
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.0008817891373801918,
+      "loss": 0.2238,
+      "step": 13800
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 0.0008824281150159744,
+      "loss": 0.2188,
+      "step": 13810
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 0.0008830670926517572,
+      "loss": 0.2244,
+      "step": 13820
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.00088370607028754,
+      "loss": 0.193,
+      "step": 13830
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 0.0008843450479233227,
+      "loss": 0.2003,
+      "step": 13840
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 0.0008849840255591054,
+      "loss": 0.2043,
+      "step": 13850
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.0008856230031948882,
+      "loss": 0.2051,
+      "step": 13860
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 0.000886261980830671,
+      "loss": 0.183,
+      "step": 13870
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 0.0008869009584664537,
+      "loss": 0.2322,
+      "step": 13880
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0008875399361022365,
+      "loss": 0.2282,
+      "step": 13890
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 0.0008881789137380192,
+      "loss": 0.196,
+      "step": 13900
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.000888817891373802,
+      "loss": 0.1948,
+      "step": 13910
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 0.0008894568690095846,
+      "loss": 0.2252,
+      "step": 13920
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 0.0008900958466453675,
+      "loss": 0.2003,
+      "step": 13930
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.0008907348242811502,
+      "loss": 0.2181,
+      "step": 13940
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 0.0008913738019169329,
+      "loss": 0.2073,
+      "step": 13950
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 0.0008920127795527156,
+      "loss": 0.2436,
+      "step": 13960
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.0008926517571884985,
+      "loss": 0.1896,
+      "step": 13970
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 0.0008932907348242811,
+      "loss": 0.201,
+      "step": 13980
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 0.0008939297124600639,
+      "loss": 0.2023,
+      "step": 13990
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0008945686900958467,
+      "loss": 0.2138,
+      "step": 14000
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 0.0008952076677316294,
+      "loss": 0.2247,
+      "step": 14010
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.0008958466453674121,
+      "loss": 0.2329,
+      "step": 14020
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 0.000896485623003195,
+      "loss": 0.1843,
+      "step": 14030
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 0.0008971246006389776,
+      "loss": 0.1895,
+      "step": 14040
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.0008977635782747604,
+      "loss": 0.1814,
+      "step": 14050
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 0.000898402555910543,
+      "loss": 0.2078,
+      "step": 14060
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 0.0008990415335463259,
+      "loss": 0.2324,
+      "step": 14070
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0008996805111821087,
+      "loss": 0.2346,
+      "step": 14080
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 0.0009003194888178913,
+      "loss": 0.2024,
+      "step": 14090
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 0.0009009584664536742,
+      "loss": 0.174,
+      "step": 14100
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0009015974440894569,
+      "loss": 0.1736,
+      "step": 14110
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 0.0009022364217252396,
+      "loss": 0.2153,
+      "step": 14120
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 0.0009028753993610224,
+      "loss": 0.1848,
+      "step": 14130
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0009035143769968052,
+      "loss": 0.1719,
+      "step": 14140
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 0.0009041533546325878,
+      "loss": 0.1904,
+      "step": 14150
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.0009047923322683707,
+      "loss": 0.1692,
+      "step": 14160
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 0.0009054313099041534,
+      "loss": 0.1367,
+      "step": 14170
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 0.0009060702875399361,
+      "loss": 0.1853,
+      "step": 14180
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0009067092651757188,
+      "loss": 0.1945,
+      "step": 14190
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 0.0009073482428115017,
+      "loss": 0.1827,
+      "step": 14200
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 0.0009079872204472843,
+      "loss": 0.1927,
+      "step": 14210
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0009086261980830671,
+      "loss": 0.1513,
+      "step": 14220
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 0.0009092651757188498,
+      "loss": 0.2149,
+      "step": 14230
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 0.0009099041533546326,
+      "loss": 0.1609,
+      "step": 14240
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.0009105431309904153,
+      "loss": 0.1814,
+      "step": 14250
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 0.0009111821086261981,
+      "loss": 0.192,
+      "step": 14260
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.0009118210862619809,
+      "loss": 0.2028,
+      "step": 14270
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 0.0009124600638977636,
+      "loss": 0.1814,
+      "step": 14280
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 0.0009130990415335464,
+      "loss": 0.1721,
+      "step": 14290
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0009137380191693291,
+      "loss": 0.1894,
+      "step": 14300
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 0.0009143769968051119,
+      "loss": 0.1716,
+      "step": 14310
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 0.0009150159744408945,
+      "loss": 0.1957,
+      "step": 14320
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0009156549520766774,
+      "loss": 0.1671,
+      "step": 14330
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 0.0009162939297124601,
+      "loss": 0.1852,
+      "step": 14340
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 0.0009169329073482428,
+      "loss": 0.2002,
+      "step": 14350
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0009175718849840256,
+      "loss": 0.1912,
+      "step": 14360
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 0.0009182108626198083,
+      "loss": 0.1929,
+      "step": 14370
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.000918849840255591,
+      "loss": 0.1772,
+      "step": 14380
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 0.0009194888178913739,
+      "loss": 0.1767,
+      "step": 14390
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 0.0009201277955271565,
+      "loss": 0.1999,
+      "step": 14400
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.0009207667731629393,
+      "loss": 0.2022,
+      "step": 14410
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 0.000921405750798722,
+      "loss": 0.2222,
+      "step": 14420
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 0.0009220447284345048,
+      "loss": 0.1796,
+      "step": 14430
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0009226837060702875,
+      "loss": 0.197,
+      "step": 14440
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.0009233226837060703,
+      "loss": 0.1699,
+      "step": 14450
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 0.000923961661341853,
+      "loss": 0.1662,
+      "step": 14460
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0009246006389776358,
+      "loss": 0.1596,
+      "step": 14470
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 0.0009252396166134186,
+      "loss": 0.1826,
+      "step": 14480
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 0.0009258785942492013,
+      "loss": 0.1801,
+      "step": 14490
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.0009265175718849841,
+      "loss": 0.217,
+      "step": 14500
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 0.0009271565495207667,
+      "loss": 0.1575,
+      "step": 14510
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0009277955271565496,
+      "loss": 0.2204,
+      "step": 14520
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 0.0009284345047923323,
+      "loss": 0.1904,
+      "step": 14530
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 0.000929073482428115,
+      "loss": 0.1612,
+      "step": 14540
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.0009297124600638978,
+      "loss": 0.2108,
+      "step": 14550
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 0.0009303514376996806,
+      "loss": 0.2171,
+      "step": 14560
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 0.0009309904153354632,
+      "loss": 0.1911,
+      "step": 14570
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.000931629392971246,
+      "loss": 0.2062,
+      "step": 14580
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 0.0009322683706070288,
+      "loss": 0.2073,
+      "step": 14590
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 0.0009329073482428115,
+      "loss": 0.2118,
+      "step": 14600
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0009335463258785942,
+      "loss": 0.1814,
+      "step": 14610
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 0.0009341853035143771,
+      "loss": 0.1654,
+      "step": 14620
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0009348242811501597,
+      "loss": 0.1794,
+      "step": 14630
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 0.0009354632587859425,
+      "loss": 0.199,
+      "step": 14640
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 0.0009361022364217253,
+      "loss": 0.1785,
+      "step": 14650
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.000936741214057508,
+      "loss": 0.222,
+      "step": 14660
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 0.0009373801916932908,
+      "loss": 0.2213,
+      "step": 14670
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 0.0009380191693290734,
+      "loss": 0.1864,
+      "step": 14680
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.0009386581469648563,
+      "loss": 0.1973,
+      "step": 14690
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 0.000939297124600639,
+      "loss": 0.2027,
+      "step": 14700
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 0.0009399361022364217,
+      "loss": 0.2115,
+      "step": 14710
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.0009405750798722045,
+      "loss": 0.207,
+      "step": 14720
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 0.0009412140575079873,
+      "loss": 0.1878,
+      "step": 14730
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.0009418530351437699,
+      "loss": 0.1794,
+      "step": 14740
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 0.0009424920127795528,
+      "loss": 0.2063,
+      "step": 14750
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 0.0009431309904153355,
+      "loss": 0.2194,
+      "step": 14760
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.0009437699680511182,
+      "loss": 0.1841,
+      "step": 14770
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 0.000944408945686901,
+      "loss": 0.2375,
+      "step": 14780
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 0.0009450479233226838,
+      "loss": 0.1869,
+      "step": 14790
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.0009456869009584664,
+      "loss": 0.1953,
+      "step": 14800
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 0.0009463258785942493,
+      "loss": 0.202,
+      "step": 14810
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 0.0009469648562300319,
+      "loss": 0.2095,
+      "step": 14820
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.0009476038338658147,
+      "loss": 0.1894,
+      "step": 14830
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 0.0009482428115015974,
+      "loss": 0.2249,
+      "step": 14840
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 0.0009488817891373802,
+      "loss": 0.1649,
+      "step": 14850
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.000949520766773163,
+      "loss": 0.1917,
+      "step": 14860
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 0.0009501597444089457,
+      "loss": 0.1856,
+      "step": 14870
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.0009507987220447285,
+      "loss": 0.2134,
+      "step": 14880
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 0.0009514376996805112,
+      "loss": 0.2183,
+      "step": 14890
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 0.000952076677316294,
+      "loss": 0.2113,
+      "step": 14900
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.0009527156549520767,
+      "loss": 0.1836,
+      "step": 14910
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 0.0009533546325878595,
+      "loss": 0.1806,
+      "step": 14920
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 0.0009539936102236422,
+      "loss": 0.2051,
+      "step": 14930
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.000954632587859425,
+      "loss": 0.1987,
+      "step": 14940
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 0.0009552715654952077,
+      "loss": 0.2183,
+      "step": 14950
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 0.0009559105431309905,
+      "loss": 0.188,
+      "step": 14960
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.0009565495207667731,
+      "loss": 0.2088,
+      "step": 14970
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 0.000957188498402556,
+      "loss": 0.2153,
+      "step": 14980
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.0009578274760383386,
+      "loss": 0.2069,
+      "step": 14990
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 0.0009584664536741214,
+      "loss": 0.1698,
+      "step": 15000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 0.0009591054313099042,
+      "loss": 0.205,
+      "step": 15010
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.0009597444089456869,
+      "loss": 0.1753,
+      "step": 15020
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 0.0009603833865814696,
+      "loss": 0.1908,
+      "step": 15030
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 0.0009610223642172525,
+      "loss": 0.1935,
+      "step": 15040
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.0009616613418530351,
+      "loss": 0.2243,
+      "step": 15050
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 0.0009623003194888179,
+      "loss": 0.2018,
+      "step": 15060
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 0.0009629392971246008,
+      "loss": 0.2062,
+      "step": 15070
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0009635782747603834,
+      "loss": 0.2033,
+      "step": 15080
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 0.0009642172523961662,
+      "loss": 0.1993,
+      "step": 15090
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0009648562300319489,
+      "loss": 0.1946,
+      "step": 15100
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 0.0009654952076677317,
+      "loss": 0.2113,
+      "step": 15110
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 0.0009661341853035144,
+      "loss": 0.2004,
+      "step": 15120
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.0009667731629392971,
+      "loss": 0.21,
+      "step": 15130
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 0.0009674121405750799,
+      "loss": 0.1794,
+      "step": 15140
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 0.0009680511182108627,
+      "loss": 0.2395,
+      "step": 15150
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.0009686900958466453,
+      "loss": 0.2506,
+      "step": 15160
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 0.0009693290734824282,
+      "loss": 0.1916,
+      "step": 15170
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 0.0009699680511182109,
+      "loss": 0.1933,
+      "step": 15180
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.0009706070287539936,
+      "loss": 0.2162,
+      "step": 15190
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 0.0009712460063897763,
+      "loss": 0.221,
+      "step": 15200
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 0.0009718849840255592,
+      "loss": 0.1934,
+      "step": 15210
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.0009725239616613418,
+      "loss": 0.1856,
+      "step": 15220
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 0.0009731629392971246,
+      "loss": 0.1951,
+      "step": 15230
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0009738019169329074,
+      "loss": 0.1962,
+      "step": 15240
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 0.0009744408945686901,
+      "loss": 0.1947,
+      "step": 15250
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 0.0009750798722044729,
+      "loss": 0.217,
+      "step": 15260
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0009757188498402556,
+      "loss": 0.2092,
+      "step": 15270
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 0.0009763578274760384,
+      "loss": 0.1695,
+      "step": 15280
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 0.0009769968051118211,
+      "loss": 0.1863,
+      "step": 15290
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.0009776357827476038,
+      "loss": 0.1947,
+      "step": 15300
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 0.0009782747603833866,
+      "loss": 0.1878,
+      "step": 15310
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 0.0009789137380191693,
+      "loss": 0.2087,
+      "step": 15320
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.000979552715654952,
+      "loss": 0.183,
+      "step": 15330
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 0.000980191693290735,
+      "loss": 0.1836,
+      "step": 15340
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0009808306709265175,
+      "loss": 0.2152,
+      "step": 15350
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 0.0009814696485623004,
+      "loss": 0.2007,
+      "step": 15360
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 0.000982108626198083,
+      "loss": 0.2202,
+      "step": 15370
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.000982747603833866,
+      "loss": 0.2071,
+      "step": 15380
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 0.0009833865814696486,
+      "loss": 0.212,
+      "step": 15390
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 0.0009840255591054314,
+      "loss": 0.217,
+      "step": 15400
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.000984664536741214,
+      "loss": 0.1713,
+      "step": 15410
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 0.0009853035143769968,
+      "loss": 0.1993,
+      "step": 15420
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 0.0009859424920127795,
+      "loss": 0.1823,
+      "step": 15430
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.0009865814696485623,
+      "loss": 0.179,
+      "step": 15440
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 0.000987220447284345,
+      "loss": 0.1938,
+      "step": 15450
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.0009878594249201277,
+      "loss": 0.2068,
+      "step": 15460
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 0.0009884984025559107,
+      "loss": 0.1709,
+      "step": 15470
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 0.0009891373801916934,
+      "loss": 0.2332,
+      "step": 15480
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0009897763578274761,
+      "loss": 0.193,
+      "step": 15490
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 0.0009904153354632589,
+      "loss": 0.212,
+      "step": 15500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 0.0009910543130990416,
+      "loss": 0.2095,
+      "step": 15510
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.0009916932907348243,
+      "loss": 0.1849,
+      "step": 15520
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 0.000992332268370607,
+      "loss": 0.212,
+      "step": 15530
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 0.0009929712460063898,
+      "loss": 0.1945,
+      "step": 15540
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0009936102236421725,
+      "loss": 0.2293,
+      "step": 15550
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 0.0009942492012779552,
+      "loss": 0.2218,
+      "step": 15560
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.000994888178913738,
+      "loss": 0.2198,
+      "step": 15570
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 0.0009955271565495207,
+      "loss": 0.2099,
+      "step": 15580
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 0.0009961661341853036,
+      "loss": 0.2159,
+      "step": 15590
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.0009968051118210862,
+      "loss": 0.2077,
+      "step": 15600
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 0.000997444089456869,
+      "loss": 0.2324,
+      "step": 15610
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 0.0009980830670926518,
+      "loss": 0.2459,
+      "step": 15620
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0009987220447284346,
+      "loss": 0.2167,
+      "step": 15630
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0009993610223642173,
+      "loss": 0.1978,
+      "step": 15640
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.001,
+      "loss": 0.1983,
+      "step": 15650
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.000999929002484913,
+      "loss": 0.1927,
+      "step": 15660
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 0.000999858004969826,
+      "loss": 0.1822,
+      "step": 15670
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 0.0009997870074547391,
+      "loss": 0.1596,
+      "step": 15680
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.000999716009939652,
+      "loss": 0.163,
+      "step": 15690
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 0.0009996450124245652,
+      "loss": 0.1961,
+      "step": 15700
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.0009995740149094782,
+      "loss": 0.1907,
+      "step": 15710
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 0.0009995030173943912,
+      "loss": 0.1704,
+      "step": 15720
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 0.0009994320198793041,
+      "loss": 0.1994,
+      "step": 15730
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0009993610223642173,
+      "loss": 0.1503,
+      "step": 15740
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 0.0009992900248491302,
+      "loss": 0.151,
+      "step": 15750
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 0.0009992190273340434,
+      "loss": 0.1875,
+      "step": 15760
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0009991480298189564,
+      "loss": 0.1754,
+      "step": 15770
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 0.0009990770323038693,
+      "loss": 0.1899,
+      "step": 15780
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 0.0009990060347887823,
+      "loss": 0.1809,
+      "step": 15790
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.0009989350372736955,
+      "loss": 0.1691,
+      "step": 15800
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 0.0009988640397586084,
+      "loss": 0.1784,
+      "step": 15810
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0009987930422435216,
+      "loss": 0.1872,
+      "step": 15820
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 0.0009987220447284346,
+      "loss": 0.1759,
+      "step": 15830
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 0.0009986510472133475,
+      "loss": 0.1688,
+      "step": 15840
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0009985800496982605,
+      "loss": 0.2019,
+      "step": 15850
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 0.0009985090521831737,
+      "loss": 0.1546,
+      "step": 15860
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 0.0009984380546680866,
+      "loss": 0.207,
+      "step": 15870
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.0009983670571529998,
+      "loss": 0.1832,
+      "step": 15880
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 0.0009982960596379127,
+      "loss": 0.1724,
+      "step": 15890
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 0.0009982250621228257,
+      "loss": 0.1905,
+      "step": 15900
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0009981540646077387,
+      "loss": 0.1705,
+      "step": 15910
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 0.0009980830670926518,
+      "loss": 0.1848,
+      "step": 15920
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 0.0009980120695775648,
+      "loss": 0.1958,
+      "step": 15930
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.000997941072062478,
+      "loss": 0.1769,
+      "step": 15940
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 0.000997870074547391,
+      "loss": 0.1765,
+      "step": 15950
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.0009977990770323039,
+      "loss": 0.2068,
+      "step": 15960
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 0.0009977280795172168,
+      "loss": 0.146,
+      "step": 15970
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 0.0009976570820021298,
+      "loss": 0.1541,
+      "step": 15980
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.000997586084487043,
+      "loss": 0.188,
+      "step": 15990
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 0.000997515086971956,
+      "loss": 0.1822,
+      "step": 16000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 0.000997444089456869,
+      "loss": 0.1641,
+      "step": 16010
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.000997373091941782,
+      "loss": 0.1893,
+      "step": 16020
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 0.000997302094426695,
+      "loss": 0.191,
+      "step": 16030
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 0.000997231096911608,
+      "loss": 0.2111,
+      "step": 16040
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.0009971600993965212,
+      "loss": 0.186,
+      "step": 16050
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 0.0009970891018814341,
+      "loss": 0.1952,
+      "step": 16060
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0009970181043663473,
+      "loss": 0.1963,
+      "step": 16070
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 0.0009969471068512602,
+      "loss": 0.183,
+      "step": 16080
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 0.0009968761093361732,
+      "loss": 0.1845,
+      "step": 16090
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0009968051118210862,
+      "loss": 0.2002,
+      "step": 16100
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 0.0009967341143059993,
+      "loss": 0.1953,
+      "step": 16110
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 0.0009966631167909123,
+      "loss": 0.1778,
+      "step": 16120
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0009965921192758255,
+      "loss": 0.2073,
+      "step": 16130
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 0.0009965211217607384,
+      "loss": 0.1755,
+      "step": 16140
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 0.0009964501242456514,
+      "loss": 0.2085,
+      "step": 16150
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.0009963791267305643,
+      "loss": 0.1978,
+      "step": 16160
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 0.0009963081292154775,
+      "loss": 0.2101,
+      "step": 16170
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0009962371317003905,
+      "loss": 0.2027,
+      "step": 16180
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 0.0009961661341853036,
+      "loss": 0.1905,
+      "step": 16190
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 0.0009960951366702166,
+      "loss": 0.175,
+      "step": 16200
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.0009960241391551296,
+      "loss": 0.1816,
+      "step": 16210
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 0.0009959531416400425,
+      "loss": 0.1917,
+      "step": 16220
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 0.0009958821441249557,
+      "loss": 0.1956,
+      "step": 16230
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.0009958111466098687,
+      "loss": 0.2037,
+      "step": 16240
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 0.0009957401490947818,
+      "loss": 0.2194,
+      "step": 16250
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 0.0009956691515796948,
+      "loss": 0.1633,
+      "step": 16260
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.0009955981540646077,
+      "loss": 0.1803,
+      "step": 16270
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 0.0009955271565495207,
+      "loss": 0.1885,
+      "step": 16280
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 0.0009954561590344337,
+      "loss": 0.2004,
+      "step": 16290
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0009953851615193468,
+      "loss": 0.1961,
+      "step": 16300
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 0.0009953141640042598,
+      "loss": 0.1827,
+      "step": 16310
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.000995243166489173,
+      "loss": 0.1775,
+      "step": 16320
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 0.000995172168974086,
+      "loss": 0.1905,
+      "step": 16330
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 0.0009951011714589989,
+      "loss": 0.2072,
+      "step": 16340
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.0009950301739439118,
+      "loss": 0.1725,
+      "step": 16350
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 0.000994959176428825,
+      "loss": 0.2023,
+      "step": 16360
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 0.000994888178913738,
+      "loss": 0.1921,
+      "step": 16370
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.0009948171813986511,
+      "loss": 0.1988,
+      "step": 16380
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 0.000994746183883564,
+      "loss": 0.1888,
+      "step": 16390
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 0.000994675186368477,
+      "loss": 0.2047,
+      "step": 16400
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.00099460418885339,
+      "loss": 0.179,
+      "step": 16410
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 0.0009945331913383032,
+      "loss": 0.1867,
+      "step": 16420
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.0009944621938232162,
+      "loss": 0.1868,
+      "step": 16430
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 0.0009943911963081293,
+      "loss": 0.212,
+      "step": 16440
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 0.0009943201987930423,
+      "loss": 0.1937,
+      "step": 16450
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0009942492012779552,
+      "loss": 0.1915,
+      "step": 16460
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 0.0009941782037628682,
+      "loss": 0.1712,
+      "step": 16470
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 0.0009941072062477814,
+      "loss": 0.2135,
+      "step": 16480
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.0009940362087326943,
+      "loss": 0.1716,
+      "step": 16490
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 0.0009939652112176075,
+      "loss": 0.193,
+      "step": 16500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 0.0009938942137025205,
+      "loss": 0.2042,
+      "step": 16510
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0009938232161874334,
+      "loss": 0.2068,
+      "step": 16520
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 0.0009937522186723464,
+      "loss": 0.1977,
+      "step": 16530
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.0009936812211572596,
+      "loss": 0.1968,
+      "step": 16540
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 0.0009936102236421725,
+      "loss": 0.2021,
+      "step": 16550
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 0.0009935392261270857,
+      "loss": 0.2083,
+      "step": 16560
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0009934682286119986,
+      "loss": 0.1932,
+      "step": 16570
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 0.0009933972310969116,
+      "loss": 0.2191,
+      "step": 16580
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 0.0009933262335818246,
+      "loss": 0.2102,
+      "step": 16590
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0009932552360667377,
+      "loss": 0.1698,
+      "step": 16600
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 0.0009931842385516507,
+      "loss": 0.2065,
+      "step": 16610
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 0.0009931132410365637,
+      "loss": 0.1976,
+      "step": 16620
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0009930422435214768,
+      "loss": 0.2074,
+      "step": 16630
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 0.0009929712460063898,
+      "loss": 0.22,
+      "step": 16640
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 0.0009929002484913027,
+      "loss": 0.1612,
+      "step": 16650
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.000992829250976216,
+      "loss": 0.2161,
+      "step": 16660
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 0.0009927582534611289,
+      "loss": 0.1898,
+      "step": 16670
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.0009926872559460418,
+      "loss": 0.155,
+      "step": 16680
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 0.000992616258430955,
+      "loss": 0.1943,
+      "step": 16690
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 0.000992545260915868,
+      "loss": 0.1812,
+      "step": 16700
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.000992474263400781,
+      "loss": 0.19,
+      "step": 16710
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 0.000992403265885694,
+      "loss": 0.2032,
+      "step": 16720
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 0.000992332268370607,
+      "loss": 0.1948,
+      "step": 16730
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.00099226127085552,
+      "loss": 0.2,
+      "step": 16740
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 0.0009921902733404332,
+      "loss": 0.1899,
+      "step": 16750
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 0.0009921192758253461,
+      "loss": 0.1937,
+      "step": 16760
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.000992048278310259,
+      "loss": 0.1854,
+      "step": 16770
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 0.0009919772807951723,
+      "loss": 0.2145,
+      "step": 16780
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0009919062832800852,
+      "loss": 0.2008,
+      "step": 16790
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 0.0009918352857649982,
+      "loss": 0.1859,
+      "step": 16800
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 0.0009917642882499114,
+      "loss": 0.1952,
+      "step": 16810
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.0009916932907348243,
+      "loss": 0.1662,
+      "step": 16820
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 0.0009916222932197373,
+      "loss": 0.1971,
+      "step": 16830
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 0.0009915512957046502,
+      "loss": 0.176,
+      "step": 16840
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0009914802981895634,
+      "loss": 0.1621,
+      "step": 16850
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.0009914093006744764,
+      "loss": 0.2204,
+      "step": 16860
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 0.0009913383031593895,
+      "loss": 0.1798,
+      "step": 16870
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0009912673056443025,
+      "loss": 0.1984,
+      "step": 16880
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 0.0009911963081292155,
+      "loss": 0.1789,
+      "step": 16890
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.0009911253106141284,
+      "loss": 0.1897,
+      "step": 16900
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 0.0009910543130990416,
+      "loss": 0.1888,
+      "step": 16910
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 0.0009909833155839546,
+      "loss": 0.1828,
+      "step": 16920
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0009909123180688675,
+      "loss": 0.1798,
+      "step": 16930
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 0.0009908413205537807,
+      "loss": 0.1729,
+      "step": 16940
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 0.0009907703230386936,
+      "loss": 0.2113,
+      "step": 16950
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 0.0009906993255236066,
+      "loss": 0.2041,
+      "step": 16960
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 0.0009906283280085198,
+      "loss": 0.1625,
+      "step": 16970
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 0.0009905573304934327,
+      "loss": 0.1726,
+      "step": 16980
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.0009904863329783457,
+      "loss": 0.1875,
+      "step": 16990
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 0.0009904153354632589,
+      "loss": 0.2126,
+      "step": 17000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 0.0009903443379481718,
+      "loss": 0.1882,
+      "step": 17010
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.0009902733404330848,
+      "loss": 0.2062,
+      "step": 17020
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 0.000990202342917998,
+      "loss": 0.1705,
+      "step": 17030
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.000990131345402911,
+      "loss": 0.2103,
+      "step": 17040
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 0.0009900603478878239,
+      "loss": 0.1766,
+      "step": 17050
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 0.000989989350372737,
+      "loss": 0.189,
+      "step": 17060
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.00098991835285765,
+      "loss": 0.1797,
+      "step": 17070
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 0.000989847355342563,
+      "loss": 0.2013,
+      "step": 17080
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 0.0009897763578274761,
+      "loss": 0.2034,
+      "step": 17090
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.000989705360312389,
+      "loss": 0.1947,
+      "step": 17100
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 0.000989634362797302,
+      "loss": 0.2248,
+      "step": 17110
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 0.0009895633652822152,
+      "loss": 0.2021,
+      "step": 17120
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.0009894923677671282,
+      "loss": 0.1932,
+      "step": 17130
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 0.0009894213702520411,
+      "loss": 0.1998,
+      "step": 17140
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0009893503727369543,
+      "loss": 0.2081,
+      "step": 17150
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 0.0009892793752218673,
+      "loss": 0.2047,
+      "step": 17160
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 0.0009892083777067802,
+      "loss": 0.1864,
+      "step": 17170
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0009891373801916934,
+      "loss": 0.2137,
+      "step": 17180
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 0.0009890663826766064,
+      "loss": 0.1858,
+      "step": 17190
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 0.0009889953851615193,
+      "loss": 0.2036,
+      "step": 17200
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.0009889243876464325,
+      "loss": 0.1979,
+      "step": 17210
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 0.0009888533901313455,
+      "loss": 0.1788,
+      "step": 17220
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 0.0009887823926162584,
+      "loss": 0.1777,
+      "step": 17230
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 0.0009887113951011714,
+      "loss": 0.1835,
+      "step": 17240
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 0.0009886403975860845,
+      "loss": 0.1546,
+      "step": 17250
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 0.0009885694000709975,
+      "loss": 0.1461,
+      "step": 17260
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.0009884984025559107,
+      "loss": 0.1716,
+      "step": 17270
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 0.0009884274050408236,
+      "loss": 0.1556,
+      "step": 17280
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 0.0009883564075257366,
+      "loss": 0.1664,
+      "step": 17290
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 0.0009882854100106496,
+      "loss": 0.1738,
+      "step": 17300
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 0.0009882144124955627,
+      "loss": 0.1797,
+      "step": 17310
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.0009881434149804757,
+      "loss": 0.2018,
+      "step": 17320
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 0.0009880724174653889,
+      "loss": 0.1683,
+      "step": 17330
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 0.0009880014199503018,
+      "loss": 0.1928,
+      "step": 17340
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.0009879304224352148,
+      "loss": 0.1737,
+      "step": 17350
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 0.0009878594249201277,
+      "loss": 0.1715,
+      "step": 17360
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 0.000987788427405041,
+      "loss": 0.1691,
+      "step": 17370
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.0009877174298899539,
+      "loss": 0.1462,
+      "step": 17380
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 0.000987646432374867,
+      "loss": 0.1835,
+      "step": 17390
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.00098757543485978,
+      "loss": 0.1515,
+      "step": 17400
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 0.000987504437344693,
+      "loss": 0.1819,
+      "step": 17410
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 0.000987433439829606,
+      "loss": 0.137,
+      "step": 17420
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.000987362442314519,
+      "loss": 0.1742,
+      "step": 17430
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 0.000987291444799432,
+      "loss": 0.147,
+      "step": 17440
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 0.000987220447284345,
+      "loss": 0.1645,
+      "step": 17450
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.0009871494497692582,
+      "loss": 0.1775,
+      "step": 17460
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 0.0009870784522541711,
+      "loss": 0.1766,
+      "step": 17470
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 0.000987007454739084,
+      "loss": 0.19,
+      "step": 17480
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.0009869364572239973,
+      "loss": 0.1521,
+      "step": 17490
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 0.0009868654597089102,
+      "loss": 0.1355,
+      "step": 17500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.0009867944621938232,
+      "loss": 0.169,
+      "step": 17510
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 0.0009867234646787364,
+      "loss": 0.1842,
+      "step": 17520
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 0.0009866524671636493,
+      "loss": 0.1588,
+      "step": 17530
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0009865814696485623,
+      "loss": 0.1587,
+      "step": 17540
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 0.0009865104721334752,
+      "loss": 0.1792,
+      "step": 17550
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 0.0009864394746183884,
+      "loss": 0.1832,
+      "step": 17560
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.0009863684771033014,
+      "loss": 0.145,
+      "step": 17570
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 0.0009862974795882145,
+      "loss": 0.1764,
+      "step": 17580
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 0.0009862264820731275,
+      "loss": 0.1706,
+      "step": 17590
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0009861554845580405,
+      "loss": 0.1737,
+      "step": 17600
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 0.0009860844870429534,
+      "loss": 0.1445,
+      "step": 17610
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 0.0009860134895278666,
+      "loss": 0.1704,
+      "step": 17620
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.0009859424920127795,
+      "loss": 0.1714,
+      "step": 17630
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 0.0009858714944976927,
+      "loss": 0.1495,
+      "step": 17640
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0009858004969826057,
+      "loss": 0.1698,
+      "step": 17650
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 0.0009857294994675186,
+      "loss": 0.1824,
+      "step": 17660
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 0.0009856585019524316,
+      "loss": 0.1515,
+      "step": 17670
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0009855875044373448,
+      "loss": 0.169,
+      "step": 17680
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 0.0009855165069222577,
+      "loss": 0.1814,
+      "step": 17690
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 0.000985445509407171,
+      "loss": 0.1548,
+      "step": 17700
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0009853745118920839,
+      "loss": 0.1615,
+      "step": 17710
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 0.0009853035143769968,
+      "loss": 0.1975,
+      "step": 17720
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 0.0009852325168619098,
+      "loss": 0.1402,
+      "step": 17730
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.000985161519346823,
+      "loss": 0.1733,
+      "step": 17740
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 0.000985090521831736,
+      "loss": 0.1782,
+      "step": 17750
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.000985019524316649,
+      "loss": 0.1664,
+      "step": 17760
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 0.000984948526801562,
+      "loss": 0.2117,
+      "step": 17770
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 0.000984877529286475,
+      "loss": 0.1698,
+      "step": 17780
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.000984806531771388,
+      "loss": 0.1607,
+      "step": 17790
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 0.0009847355342563011,
+      "loss": 0.171,
+      "step": 17800
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 0.000984664536741214,
+      "loss": 0.1969,
+      "step": 17810
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0009845935392261273,
+      "loss": 0.1615,
+      "step": 17820
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 0.0009845225417110402,
+      "loss": 0.1649,
+      "step": 17830
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 0.0009844515441959532,
+      "loss": 0.1773,
+      "step": 17840
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.0009843805466808661,
+      "loss": 0.19,
+      "step": 17850
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 0.000984309549165779,
+      "loss": 0.1501,
+      "step": 17860
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.0009842385516506923,
+      "loss": 0.1397,
+      "step": 17870
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 0.0009841675541356052,
+      "loss": 0.1641,
+      "step": 17880
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 0.0009840965566205184,
+      "loss": 0.161,
+      "step": 17890
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.0009840255591054314,
+      "loss": 0.1751,
+      "step": 17900
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 0.0009839545615903443,
+      "loss": 0.1774,
+      "step": 17910
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 0.0009838835640752573,
+      "loss": 0.2226,
+      "step": 17920
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0009838125665601704,
+      "loss": 0.1531,
+      "step": 17930
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 0.0009837415690450834,
+      "loss": 0.1807,
+      "step": 17940
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 0.0009836705715299966,
+      "loss": 0.1769,
+      "step": 17950
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.0009835995740149095,
+      "loss": 0.1727,
+      "step": 17960
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 0.0009835285764998225,
+      "loss": 0.1738,
+      "step": 17970
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 0.0009834575789847355,
+      "loss": 0.1354,
+      "step": 17980
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0009833865814696486,
+      "loss": 0.1979,
+      "step": 17990
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 0.0009833155839545616,
+      "loss": 0.1774,
+      "step": 18000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0009832445864394748,
+      "loss": 0.1869,
+      "step": 18010
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 0.0009831735889243877,
+      "loss": 0.1486,
+      "step": 18020
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 0.0009831025914093007,
+      "loss": 0.2041,
+      "step": 18030
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.0009830315938942136,
+      "loss": 0.188,
+      "step": 18040
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 0.0009829605963791268,
+      "loss": 0.1349,
+      "step": 18050
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0009828895988640398,
+      "loss": 0.1786,
+      "step": 18060
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.000982818601348953,
+      "loss": 0.1886,
+      "step": 18070
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 0.000982747603833866,
+      "loss": 0.1876,
+      "step": 18080
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 0.0009826766063187789,
+      "loss": 0.189,
+      "step": 18090
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.0009826056088036918,
+      "loss": 0.1677,
+      "step": 18100
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 0.000982534611288605,
+      "loss": 0.2143,
+      "step": 18110
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.000982463613773518,
+      "loss": 0.1722,
+      "step": 18120
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 0.0009823926162584311,
+      "loss": 0.1762,
+      "step": 18130
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 0.000982321618743344,
+      "loss": 0.1833,
+      "step": 18140
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.000982250621228257,
+      "loss": 0.1845,
+      "step": 18150
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 0.00098217962371317,
+      "loss": 0.1567,
+      "step": 18160
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 0.000982108626198083,
+      "loss": 0.1642,
+      "step": 18170
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.0009820376286829961,
+      "loss": 0.1576,
+      "step": 18180
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 0.000981966631167909,
+      "loss": 0.1377,
+      "step": 18190
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 0.0009818956336528223,
+      "loss": 0.1921,
+      "step": 18200
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.0009818246361377352,
+      "loss": 0.2037,
+      "step": 18210
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 0.0009817536386226482,
+      "loss": 0.1988,
+      "step": 18220
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0009816826411075611,
+      "loss": 0.1576,
+      "step": 18230
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 0.0009816116435924743,
+      "loss": 0.1727,
+      "step": 18240
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 0.0009815406460773873,
+      "loss": 0.1925,
+      "step": 18250
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0009814696485623004,
+      "loss": 0.1942,
+      "step": 18260
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 0.0009813986510472134,
+      "loss": 0.1506,
+      "step": 18270
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 0.0009813276535321264,
+      "loss": 0.1973,
+      "step": 18280
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.0009812566560170393,
+      "loss": 0.1641,
+      "step": 18290
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 0.0009811856585019525,
+      "loss": 0.1719,
+      "step": 18300
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 0.0009811146609868654,
+      "loss": 0.1681,
+      "step": 18310
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.0009810436634717786,
+      "loss": 0.1644,
+      "step": 18320
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 0.0009809726659566916,
+      "loss": 0.153,
+      "step": 18330
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 0.0009809016684416045,
+      "loss": 0.1978,
+      "step": 18340
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.0009808306709265175,
+      "loss": 0.1808,
+      "step": 18350
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 0.0009807596734114307,
+      "loss": 0.1967,
+      "step": 18360
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0009806886758963436,
+      "loss": 0.1699,
+      "step": 18370
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 0.0009806176783812568,
+      "loss": 0.1866,
+      "step": 18380
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 0.0009805466808661698,
+      "loss": 0.1813,
+      "step": 18390
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0009804756833510827,
+      "loss": 0.1742,
+      "step": 18400
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 0.0009804046858359957,
+      "loss": 0.2112,
+      "step": 18410
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 0.0009803336883209088,
+      "loss": 0.1966,
+      "step": 18420
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.0009802626908058218,
+      "loss": 0.1639,
+      "step": 18430
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 0.000980191693290735,
+      "loss": 0.1614,
+      "step": 18440
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 0.000980120695775648,
+      "loss": 0.1992,
+      "step": 18450
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.000980049698260561,
+      "loss": 0.1965,
+      "step": 18460
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 0.0009799787007454739,
+      "loss": 0.1885,
+      "step": 18470
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0009799077032303868,
+      "loss": 0.1509,
+      "step": 18480
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 0.0009798367057153,
+      "loss": 0.1952,
+      "step": 18490
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 0.000979765708200213,
+      "loss": 0.1801,
+      "step": 18500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.0009796947106851261,
+      "loss": 0.1521,
+      "step": 18510
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 0.000979623713170039,
+      "loss": 0.1632,
+      "step": 18520
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 0.000979552715654952,
+      "loss": 0.1945,
+      "step": 18530
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.000979481718139865,
+      "loss": 0.1883,
+      "step": 18540
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 0.0009794107206247782,
+      "loss": 0.1832,
+      "step": 18550
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 0.0009793397231096911,
+      "loss": 0.1933,
+      "step": 18560
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.0009792687255946043,
+      "loss": 0.1785,
+      "step": 18570
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 0.0009791977280795173,
+      "loss": 0.1788,
+      "step": 18580
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.0009791267305644302,
+      "loss": 0.18,
+      "step": 18590
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 0.0009790557330493432,
+      "loss": 0.1882,
+      "step": 18600
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 0.0009789847355342563,
+      "loss": 0.1823,
+      "step": 18610
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0009789137380191693,
+      "loss": 0.1843,
+      "step": 18620
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 0.0009788427405040825,
+      "loss": 0.1725,
+      "step": 18630
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 0.0009787717429889954,
+      "loss": 0.1905,
+      "step": 18640
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.0009787007454739084,
+      "loss": 0.1904,
+      "step": 18650
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 0.0009786297479588214,
+      "loss": 0.1537,
+      "step": 18660
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 0.0009785587504437345,
+      "loss": 0.1519,
+      "step": 18670
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0009784877529286475,
+      "loss": 0.1831,
+      "step": 18680
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 0.0009784167554135607,
+      "loss": 0.1562,
+      "step": 18690
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.0009783457578984736,
+      "loss": 0.2007,
+      "step": 18700
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 0.0009782747603833866,
+      "loss": 0.1683,
+      "step": 18710
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 0.0009782037628682995,
+      "loss": 0.1927,
+      "step": 18720
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0009781327653532127,
+      "loss": 0.178,
+      "step": 18730
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 0.0009780617678381257,
+      "loss": 0.1629,
+      "step": 18740
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 0.0009779907703230388,
+      "loss": 0.177,
+      "step": 18750
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 0.0009779197728079518,
+      "loss": 0.1758,
+      "step": 18760
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 0.0009778487752928648,
+      "loss": 0.2114,
+      "step": 18770
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 0.0009777777777777777,
+      "loss": 0.1582,
+      "step": 18780
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0009777067802626907,
+      "loss": 0.1586,
+      "step": 18790
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 0.0009776357827476038,
+      "loss": 0.1523,
+      "step": 18800
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 0.0009775647852325168,
+      "loss": 0.1614,
+      "step": 18810
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.00097749378771743,
+      "loss": 0.169,
+      "step": 18820
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 0.000977422790202343,
+      "loss": 0.1339,
+      "step": 18830
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.000977351792687256,
+      "loss": 0.1313,
+      "step": 18840
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 0.0009772807951721689,
+      "loss": 0.1481,
+      "step": 18850
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 0.000977209797657082,
+      "loss": 0.1294,
+      "step": 18860
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.000977138800141995,
+      "loss": 0.1555,
+      "step": 18870
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 0.0009770678026269082,
+      "loss": 0.1555,
+      "step": 18880
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 0.0009769968051118211,
+      "loss": 0.1688,
+      "step": 18890
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.000976925807596734,
+      "loss": 0.1738,
+      "step": 18900
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 0.000976854810081647,
+      "loss": 0.1463,
+      "step": 18910
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 0.0009767838125665602,
+      "loss": 0.1241,
+      "step": 18920
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.0009767128150514732,
+      "loss": 0.1401,
+      "step": 18930
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 0.0009766418175363863,
+      "loss": 0.1685,
+      "step": 18940
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.0009765708200212993,
+      "loss": 0.1761,
+      "step": 18950
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 0.0009764998225062123,
+      "loss": 0.1738,
+      "step": 18960
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 0.0009764288249911253,
+      "loss": 0.1195,
+      "step": 18970
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.0009763578274760384,
+      "loss": 0.1416,
+      "step": 18980
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 0.0009762868299609515,
+      "loss": 0.1504,
+      "step": 18990
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 0.0009762158324458644,
+      "loss": 0.1515,
+      "step": 19000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0009761448349307775,
+      "loss": 0.1773,
+      "step": 19010
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 0.0009760738374156904,
+      "loss": 0.1548,
+      "step": 19020
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 0.0009760028399006035,
+      "loss": 0.1606,
+      "step": 19030
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.0009759318423855166,
+      "loss": 0.1607,
+      "step": 19040
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 0.0009758608448704296,
+      "loss": 0.1473,
+      "step": 19050
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 0.0009757898473553426,
+      "loss": 0.1539,
+      "step": 19060
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0009757188498402556,
+      "loss": 0.1577,
+      "step": 19070
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 0.0009756478523251686,
+      "loss": 0.1825,
+      "step": 19080
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.0009755768548100816,
+      "loss": 0.1687,
+      "step": 19090
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 0.0009755058572949948,
+      "loss": 0.188,
+      "step": 19100
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 0.0009754348597799077,
+      "loss": 0.1424,
+      "step": 19110
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0009753638622648208,
+      "loss": 0.1478,
+      "step": 19120
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 0.0009752928647497337,
+      "loss": 0.1601,
+      "step": 19130
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 0.0009752218672346468,
+      "loss": 0.1706,
+      "step": 19140
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0009751508697195598,
+      "loss": 0.1565,
+      "step": 19150
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 0.0009750798722044729,
+      "loss": 0.1258,
+      "step": 19160
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 0.0009750088746893859,
+      "loss": 0.1468,
+      "step": 19170
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.000974937877174299,
+      "loss": 0.1865,
+      "step": 19180
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 0.0009748668796592119,
+      "loss": 0.1551,
+      "step": 19190
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.000974795882144125,
+      "loss": 0.1688,
+      "step": 19200
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 0.0009747248846290379,
+      "loss": 0.16,
+      "step": 19210
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 0.000974653887113951,
+      "loss": 0.1686,
+      "step": 19220
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.0009745828895988641,
+      "loss": 0.1644,
+      "step": 19230
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 0.0009745118920837771,
+      "loss": 0.1605,
+      "step": 19240
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 0.0009744408945686901,
+      "loss": 0.1506,
+      "step": 19250
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0009743698970536032,
+      "loss": 0.1436,
+      "step": 19260
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.0009742988995385161,
+      "loss": 0.1685,
+      "step": 19270
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 0.0009742279020234292,
+      "loss": 0.1505,
+      "step": 19280
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0009741569045083423,
+      "loss": 0.1609,
+      "step": 19290
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 0.0009740859069932553,
+      "loss": 0.1803,
+      "step": 19300
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0009740149094781683,
+      "loss": 0.1806,
+      "step": 19310
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 0.0009739439119630813,
+      "loss": 0.1733,
+      "step": 19320
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 0.0009738729144479943,
+      "loss": 0.1618,
+      "step": 19330
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.0009738019169329074,
+      "loss": 0.157,
+      "step": 19340
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 0.0009737309194178204,
+      "loss": 0.1374,
+      "step": 19350
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 0.0009736599219027335,
+      "loss": 0.1689,
+      "step": 19360
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0009735889243876465,
+      "loss": 0.1813,
+      "step": 19370
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 0.0009735179268725594,
+      "loss": 0.1313,
+      "step": 19380
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 0.0009734469293574725,
+      "loss": 0.16,
+      "step": 19390
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0009733759318423854,
+      "loss": 0.1317,
+      "step": 19400
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 0.0009733049343272986,
+      "loss": 0.13,
+      "step": 19410
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 0.0009732339368122116,
+      "loss": 0.15,
+      "step": 19420
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0009731629392971246,
+      "loss": 0.1567,
+      "step": 19430
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 0.0009730919417820376,
+      "loss": 0.1477,
+      "step": 19440
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.0009730209442669507,
+      "loss": 0.1409,
+      "step": 19450
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 0.0009729499467518636,
+      "loss": 0.1793,
+      "step": 19460
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 0.0009728789492367768,
+      "loss": 0.1322,
+      "step": 19470
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0009728079517216898,
+      "loss": 0.1645,
+      "step": 19480
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 0.0009727369542066028,
+      "loss": 0.195,
+      "step": 19490
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 0.0009726659566915158,
+      "loss": 0.1419,
+      "step": 19500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.0009725949591764288,
+      "loss": 0.1422,
+      "step": 19510
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 0.0009725239616613418,
+      "loss": 0.1796,
+      "step": 19520
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 0.000972452964146255,
+      "loss": 0.1615,
+      "step": 19530
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.0009723819666311679,
+      "loss": 0.173,
+      "step": 19540
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 0.000972310969116081,
+      "loss": 0.1442,
+      "step": 19550
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.000972239971600994,
+      "loss": 0.1648,
+      "step": 19560
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 0.000972168974085907,
+      "loss": 0.1543,
+      "step": 19570
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 0.00097209797657082,
+      "loss": 0.1507,
+      "step": 19580
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0009720269790557332,
+      "loss": 0.1571,
+      "step": 19590
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 0.0009719559815406461,
+      "loss": 0.1697,
+      "step": 19600
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 0.0009718849840255592,
+      "loss": 0.1622,
+      "step": 19610
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.0009718139865104721,
+      "loss": 0.1548,
+      "step": 19620
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 0.0009717429889953852,
+      "loss": 0.169,
+      "step": 19630
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 0.0009716719914802982,
+      "loss": 0.1644,
+      "step": 19640
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.0009716009939652113,
+      "loss": 0.1387,
+      "step": 19650
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 0.0009715299964501243,
+      "loss": 0.177,
+      "step": 19660
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0009714589989350374,
+      "loss": 0.1627,
+      "step": 19670
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 0.0009713880014199503,
+      "loss": 0.1451,
+      "step": 19680
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 0.0009713170039048633,
+      "loss": 0.1827,
+      "step": 19690
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0009712460063897763,
+      "loss": 0.1593,
+      "step": 19700
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 0.0009711750088746894,
+      "loss": 0.1521,
+      "step": 19710
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 0.0009711040113596025,
+      "loss": 0.1759,
+      "step": 19720
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0009710330138445154,
+      "loss": 0.1868,
+      "step": 19730
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 0.0009709620163294285,
+      "loss": 0.1327,
+      "step": 19740
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 0.0009708910188143415,
+      "loss": 0.1349,
+      "step": 19750
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0009708200212992545,
+      "loss": 0.1645,
+      "step": 19760
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 0.0009707490237841676,
+      "loss": 0.1315,
+      "step": 19770
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 0.0009706780262690807,
+      "loss": 0.184,
+      "step": 19780
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.0009706070287539936,
+      "loss": 0.1557,
+      "step": 19790
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 0.0009705360312389067,
+      "loss": 0.1622,
+      "step": 19800
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0009704650337238196,
+      "loss": 0.1598,
+      "step": 19810
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 0.0009703940362087327,
+      "loss": 0.1914,
+      "step": 19820
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 0.0009703230386936457,
+      "loss": 0.1757,
+      "step": 19830
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0009702520411785588,
+      "loss": 0.1613,
+      "step": 19840
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 0.0009701810436634718,
+      "loss": 0.175,
+      "step": 19850
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 0.0009701100461483849,
+      "loss": 0.1778,
+      "step": 19860
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0009700390486332978,
+      "loss": 0.14,
+      "step": 19870
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 0.0009699680511182109,
+      "loss": 0.148,
+      "step": 19880
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 0.0009698970536031238,
+      "loss": 0.1899,
+      "step": 19890
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.000969826056088037,
+      "loss": 0.143,
+      "step": 19900
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 0.00096975505857295,
+      "loss": 0.1668,
+      "step": 19910
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.000969684061057863,
+      "loss": 0.1607,
+      "step": 19920
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 0.000969613063542776,
+      "loss": 0.1357,
+      "step": 19930
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 0.0009695420660276891,
+      "loss": 0.18,
+      "step": 19940
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.000969471068512602,
+      "loss": 0.1766,
+      "step": 19950
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 0.0009694000709975152,
+      "loss": 0.1852,
+      "step": 19960
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 0.0009693290734824282,
+      "loss": 0.1969,
+      "step": 19970
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0009692580759673412,
+      "loss": 0.1483,
+      "step": 19980
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 0.0009691870784522542,
+      "loss": 0.1823,
+      "step": 19990
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 0.0009691160809371671,
+      "loss": 0.1495,
+      "step": 20000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.0009690450834220802,
+      "loss": 0.144,
+      "step": 20010
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 0.0009689740859069933,
+      "loss": 0.1633,
+      "step": 20020
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0009689030883919063,
+      "loss": 0.1736,
+      "step": 20030
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 0.0009688320908768193,
+      "loss": 0.1848,
+      "step": 20040
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 0.0009687610933617324,
+      "loss": 0.1701,
+      "step": 20050
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.0009686900958466453,
+      "loss": 0.1647,
+      "step": 20060
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 0.0009686190983315584,
+      "loss": 0.1801,
+      "step": 20070
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 0.0009685481008164714,
+      "loss": 0.1595,
+      "step": 20080
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.0009684771033013845,
+      "loss": 0.1618,
+      "step": 20090
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 0.0009684061057862975,
+      "loss": 0.1559,
+      "step": 20100
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 0.0009683351082712105,
+      "loss": 0.173,
+      "step": 20110
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.0009682641107561235,
+      "loss": 0.15,
+      "step": 20120
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 0.0009681931132410366,
+      "loss": 0.1707,
+      "step": 20130
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 0.0009681221157259496,
+      "loss": 0.1403,
+      "step": 20140
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0009680511182108627,
+      "loss": 0.1787,
+      "step": 20150
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 0.0009679801206957757,
+      "loss": 0.1708,
+      "step": 20160
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0009679091231806887,
+      "loss": 0.1595,
+      "step": 20170
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 0.0009678381256656017,
+      "loss": 0.1536,
+      "step": 20180
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 0.0009677671281505147,
+      "loss": 0.1783,
+      "step": 20190
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0009676961306354278,
+      "loss": 0.1694,
+      "step": 20200
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 0.0009676251331203409,
+      "loss": 0.1593,
+      "step": 20210
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 0.0009675541356052538,
+      "loss": 0.1378,
+      "step": 20220
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.0009674831380901669,
+      "loss": 0.174,
+      "step": 20230
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 0.0009674121405750799,
+      "loss": 0.1588,
+      "step": 20240
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 0.0009673411430599929,
+      "loss": 0.1726,
+      "step": 20250
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.000967270145544906,
+      "loss": 0.1696,
+      "step": 20260
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 0.0009671991480298191,
+      "loss": 0.1529,
+      "step": 20270
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.000967128150514732,
+      "loss": 0.1869,
+      "step": 20280
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 0.0009670571529996451,
+      "loss": 0.1673,
+      "step": 20290
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 0.000966986155484558,
+      "loss": 0.1682,
+      "step": 20300
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.000966915157969471,
+      "loss": 0.1937,
+      "step": 20310
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 0.0009668441604543842,
+      "loss": 0.1636,
+      "step": 20320
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 0.0009667731629392971,
+      "loss": 0.1535,
+      "step": 20330
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.0009667021654242102,
+      "loss": 0.1794,
+      "step": 20340
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 0.0009666311679091232,
+      "loss": 0.1457,
+      "step": 20350
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 0.0009665601703940362,
+      "loss": 0.168,
+      "step": 20360
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.0009664891728789492,
+      "loss": 0.1501,
+      "step": 20370
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 0.0009664181753638623,
+      "loss": 0.1401,
+      "step": 20380
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 0.0009663471778487753,
+      "loss": 0.1438,
+      "step": 20390
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0009662761803336884,
+      "loss": 0.1199,
+      "step": 20400
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 0.0009662051828186013,
+      "loss": 0.1534,
+      "step": 20410
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.0009661341853035144,
+      "loss": 0.1386,
+      "step": 20420
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 0.0009660631877884274,
+      "loss": 0.1598,
+      "step": 20430
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 0.0009659921902733404,
+      "loss": 0.1612,
+      "step": 20440
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.0009659211927582535,
+      "loss": 0.1319,
+      "step": 20450
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 0.0009658501952431666,
+      "loss": 0.157,
+      "step": 20460
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0009657791977280795,
+      "loss": 0.1411,
+      "step": 20470
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.0009657082002129926,
+      "loss": 0.1287,
+      "step": 20480
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 0.0009656372026979055,
+      "loss": 0.1241,
+      "step": 20490
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 0.0009655662051828186,
+      "loss": 0.168,
+      "step": 20500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.0009654952076677317,
+      "loss": 0.1583,
+      "step": 20510
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 0.0009654242101526447,
+      "loss": 0.1267,
+      "step": 20520
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.0009653532126375577,
+      "loss": 0.1357,
+      "step": 20530
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 0.0009652822151224708,
+      "loss": 0.126,
+      "step": 20540
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 0.0009652112176073837,
+      "loss": 0.1379,
+      "step": 20550
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0009651402200922968,
+      "loss": 0.1572,
+      "step": 20560
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 0.0009650692225772098,
+      "loss": 0.1551,
+      "step": 20570
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 0.0009649982250621229,
+      "loss": 0.1277,
+      "step": 20580
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.0009649272275470359,
+      "loss": 0.1609,
+      "step": 20590
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 0.0009648562300319489,
+      "loss": 0.1459,
+      "step": 20600
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 0.0009647852325168619,
+      "loss": 0.1602,
+      "step": 20610
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.0009647142350017749,
+      "loss": 0.1472,
+      "step": 20620
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 0.000964643237486688,
+      "loss": 0.128,
+      "step": 20630
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.000964572239971601,
+      "loss": 0.1444,
+      "step": 20640
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 0.000964501242456514,
+      "loss": 0.1627,
+      "step": 20650
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 0.000964430244941427,
+      "loss": 0.1418,
+      "step": 20660
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.0009643592474263401,
+      "loss": 0.135,
+      "step": 20670
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 0.000964288249911253,
+      "loss": 0.1337,
+      "step": 20680
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 0.0009642172523961662,
+      "loss": 0.1353,
+      "step": 20690
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.0009641462548810792,
+      "loss": 0.1586,
+      "step": 20700
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 0.0009640752573659922,
+      "loss": 0.1386,
+      "step": 20710
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 0.0009640042598509052,
+      "loss": 0.1369,
+      "step": 20720
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.0009639332623358183,
+      "loss": 0.1357,
+      "step": 20730
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 0.0009638622648207312,
+      "loss": 0.1472,
+      "step": 20740
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 0.0009637912673056444,
+      "loss": 0.1533,
+      "step": 20750
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0009637202697905573,
+      "loss": 0.1584,
+      "step": 20760
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 0.0009636492722754704,
+      "loss": 0.126,
+      "step": 20770
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0009635782747603834,
+      "loss": 0.1642,
+      "step": 20780
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 0.0009635072772452964,
+      "loss": 0.1284,
+      "step": 20790
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 0.0009634362797302094,
+      "loss": 0.1457,
+      "step": 20800
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.0009633652822151226,
+      "loss": 0.1481,
+      "step": 20810
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 0.0009632942847000355,
+      "loss": 0.1639,
+      "step": 20820
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 0.0009632232871849486,
+      "loss": 0.1459,
+      "step": 20830
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0009631522896698616,
+      "loss": 0.1742,
+      "step": 20840
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 0.0009630812921547746,
+      "loss": 0.1388,
+      "step": 20850
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 0.0009630102946396876,
+      "loss": 0.1769,
+      "step": 20860
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.0009629392971246008,
+      "loss": 0.1517,
+      "step": 20870
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 0.0009628682996095137,
+      "loss": 0.1463,
+      "step": 20880
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0009627973020944268,
+      "loss": 0.1135,
+      "step": 20890
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 0.0009627263045793397,
+      "loss": 0.1543,
+      "step": 20900
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 0.0009626553070642528,
+      "loss": 0.1598,
+      "step": 20910
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.0009625843095491658,
+      "loss": 0.1745,
+      "step": 20920
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 0.0009625133120340789,
+      "loss": 0.1595,
+      "step": 20930
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 0.0009624423145189919,
+      "loss": 0.1329,
+      "step": 20940
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0009623713170039048,
+      "loss": 0.1612,
+      "step": 20950
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 0.0009623003194888179,
+      "loss": 0.1528,
+      "step": 20960
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 0.0009622293219737309,
+      "loss": 0.1518,
+      "step": 20970
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.0009621583244586439,
+      "loss": 0.1635,
+      "step": 20980
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 0.000962087326943557,
+      "loss": 0.1446,
+      "step": 20990
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.0009620163294284701,
+      "loss": 0.1471,
+      "step": 21000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 0.000961945331913383,
+      "loss": 0.1476,
+      "step": 21010
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 0.0009618743343982961,
+      "loss": 0.1907,
+      "step": 21020
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.000961803336883209,
+      "loss": 0.1495,
+      "step": 21030
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 0.0009617323393681221,
+      "loss": 0.1817,
+      "step": 21040
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 0.0009616613418530351,
+      "loss": 0.1571,
+      "step": 21050
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0009615903443379483,
+      "loss": 0.1746,
+      "step": 21060
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 0.0009615193468228612,
+      "loss": 0.1451,
+      "step": 21070
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 0.0009614483493077743,
+      "loss": 0.1545,
+      "step": 21080
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0009613773517926872,
+      "loss": 0.148,
+      "step": 21090
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 0.0009613063542776003,
+      "loss": 0.1368,
+      "step": 21100
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 0.0009612353567625133,
+      "loss": 0.1401,
+      "step": 21110
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0009611643592474264,
+      "loss": 0.1375,
+      "step": 21120
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 0.0009610933617323394,
+      "loss": 0.1525,
+      "step": 21130
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.0009610223642172525,
+      "loss": 0.1509,
+      "step": 21140
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 0.0009609513667021654,
+      "loss": 0.1531,
+      "step": 21150
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 0.0009608803691870785,
+      "loss": 0.1096,
+      "step": 21160
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.0009608093716719914,
+      "loss": 0.1678,
+      "step": 21170
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 0.0009607383741569046,
+      "loss": 0.1366,
+      "step": 21180
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 0.0009606673766418176,
+      "loss": 0.1431,
+      "step": 21190
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.0009605963791267306,
+      "loss": 0.1709,
+      "step": 21200
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 0.0009605253816116436,
+      "loss": 0.1812,
+      "step": 21210
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 0.0009604543840965567,
+      "loss": 0.1601,
+      "step": 21220
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0009603833865814696,
+      "loss": 0.1491,
+      "step": 21230
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 0.0009603123890663828,
+      "loss": 0.128,
+      "step": 21240
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0009602413915512958,
+      "loss": 0.1746,
+      "step": 21250
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 0.0009601703940362087,
+      "loss": 0.1417,
+      "step": 21260
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 0.0009600993965211218,
+      "loss": 0.1661,
+      "step": 21270
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0009600283990060347,
+      "loss": 0.1624,
+      "step": 21280
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 0.0009599574014909478,
+      "loss": 0.1768,
+      "step": 21290
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 0.0009598864039758609,
+      "loss": 0.1553,
+      "step": 21300
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.0009598154064607739,
+      "loss": 0.1604,
+      "step": 21310
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 0.0009597444089456869,
+      "loss": 0.1621,
+      "step": 21320
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 0.0009596734114306,
+      "loss": 0.1551,
+      "step": 21330
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.0009596024139155129,
+      "loss": 0.1578,
+      "step": 21340
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 0.000959531416400426,
+      "loss": 0.1668,
+      "step": 21350
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.000959460418885339,
+      "loss": 0.1563,
+      "step": 21360
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 0.0009593894213702521,
+      "loss": 0.1717,
+      "step": 21370
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 0.0009593184238551651,
+      "loss": 0.1706,
+      "step": 21380
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0009592474263400781,
+      "loss": 0.1705,
+      "step": 21390
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 0.0009591764288249911,
+      "loss": 0.1619,
+      "step": 21400
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 0.0009591054313099042,
+      "loss": 0.1597,
+      "step": 21410
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0009590344337948172,
+      "loss": 0.1756,
+      "step": 21420
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 0.0009589634362797303,
+      "loss": 0.1706,
+      "step": 21430
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 0.0009588924387646433,
+      "loss": 0.1657,
+      "step": 21440
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.0009588214412495563,
+      "loss": 0.1659,
+      "step": 21450
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 0.0009587504437344693,
+      "loss": 0.1977,
+      "step": 21460
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 0.0009586794462193823,
+      "loss": 0.1391,
+      "step": 21470
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.0009586084487042954,
+      "loss": 0.1594,
+      "step": 21480
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 0.0009585374511892085,
+      "loss": 0.1668,
+      "step": 21490
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.0009584664536741214,
+      "loss": 0.1544,
+      "step": 21500
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 0.0009583954561590345,
+      "loss": 0.1808,
+      "step": 21510
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 0.0009583244586439475,
+      "loss": 0.1779,
+      "step": 21520
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0009582534611288605,
+      "loss": 0.1392,
+      "step": 21530
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 0.0009581824636137736,
+      "loss": 0.165,
+      "step": 21540
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 0.0009581114660986867,
+      "loss": 0.1705,
+      "step": 21550
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0009580404685835996,
+      "loss": 0.1484,
+      "step": 21560
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 0.0009579694710685126,
+      "loss": 0.1389,
+      "step": 21570
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 0.0009578984735534256,
+      "loss": 0.1566,
+      "step": 21580
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.0009578274760383386,
+      "loss": 0.133,
+      "step": 21590
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 0.0009577564785232517,
+      "loss": 0.1441,
+      "step": 21600
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.0009576854810081647,
+      "loss": 0.1644,
+      "step": 21610
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 0.0009576144834930778,
+      "loss": 0.1805,
+      "step": 21620
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 0.0009575434859779907,
+      "loss": 0.1491,
+      "step": 21630
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0009574724884629038,
+      "loss": 0.1472,
+      "step": 21640
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 0.0009574014909478168,
+      "loss": 0.173,
+      "step": 21650
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 0.0009573304934327298,
+      "loss": 0.1336,
+      "step": 21660
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.0009572594959176429,
+      "loss": 0.1522,
+      "step": 21670
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.000957188498402556,
+      "loss": 0.1586,
+      "step": 21680
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 0.0009571175008874689,
+      "loss": 0.1364,
+      "step": 21690
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.000957046503372382,
+      "loss": 0.1443,
+      "step": 21700
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 0.000956975505857295,
+      "loss": 0.182,
+      "step": 21710
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.000956904508342208,
+      "loss": 0.1481,
+      "step": 21720
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 0.0009568335108271211,
+      "loss": 0.1608,
+      "step": 21730
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 0.0009567625133120342,
+      "loss": 0.1596,
+      "step": 21740
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0009566915157969471,
+      "loss": 0.1546,
+      "step": 21750
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 0.0009566205182818602,
+      "loss": 0.1417,
+      "step": 21760
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 0.0009565495207667731,
+      "loss": 0.1342,
+      "step": 21770
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.0009564785232516862,
+      "loss": 0.1615,
+      "step": 21780
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 0.0009564075257365993,
+      "loss": 0.1452,
+      "step": 21790
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 0.0009563365282215123,
+      "loss": 0.1705,
+      "step": 21800
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0009562655307064253,
+      "loss": 0.1404,
+      "step": 21810
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 0.0009561945331913384,
+      "loss": 0.141,
+      "step": 21820
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0009561235356762513,
+      "loss": 0.1642,
+      "step": 21830
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 0.0009560525381611644,
+      "loss": 0.1654,
+      "step": 21840
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 0.0009559815406460774,
+      "loss": 0.1722,
+      "step": 21850
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0009559105431309905,
+      "loss": 0.1509,
+      "step": 21860
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 0.0009558395456159035,
+      "loss": 0.1557,
+      "step": 21870
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 0.0009557685481008164,
+      "loss": 0.1645,
+      "step": 21880
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0009556975505857295,
+      "loss": 0.1729,
+      "step": 21890
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 0.0009556265530706425,
+      "loss": 0.1471,
+      "step": 21900
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 0.0009555555555555556,
+      "loss": 0.1493,
+      "step": 21910
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0009554845580404686,
+      "loss": 0.1662,
+      "step": 21920
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 0.0009554135605253817,
+      "loss": 0.1034,
+      "step": 21930
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 0.0009553425630102946,
+      "loss": 0.1537,
+      "step": 21940
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0009552715654952077,
+      "loss": 0.1495,
+      "step": 21950
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 0.0009552005679801206,
+      "loss": 0.1488,
+      "step": 21960
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0009551295704650338,
+      "loss": 0.1242,
+      "step": 21970
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 0.0009550585729499468,
+      "loss": 0.1232,
+      "step": 21980
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 0.0009549875754348598,
+      "loss": 0.126,
+      "step": 21990
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0009549165779197728,
+      "loss": 0.1493,
+      "step": 22000
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 0.0009548455804046859,
+      "loss": 0.145,
+      "step": 22010
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 0.0009547745828895988,
+      "loss": 0.1363,
+      "step": 22020
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.000954703585374512,
+      "loss": 0.1487,
+      "step": 22030
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 0.000954632587859425,
+      "loss": 0.15,
+      "step": 22040
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 0.000954561590344338,
+      "loss": 0.1205,
+      "step": 22050
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.000954490592829251,
+      "loss": 0.1431,
+      "step": 22060
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 0.000954419595314164,
+      "loss": 0.1134,
+      "step": 22070
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.000954348597799077,
+      "loss": 0.1431,
+      "step": 22080
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 0.0009542776002839902,
+      "loss": 0.1484,
+      "step": 22090
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 0.0009542066027689031,
+      "loss": 0.1435,
+      "step": 22100
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.0009541356052538162,
+      "loss": 0.1595,
+      "step": 22110
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 0.0009540646077387292,
+      "loss": 0.1407,
+      "step": 22120
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 0.0009539936102236422,
+      "loss": 0.1514,
+      "step": 22130
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0009539226127085552,
+      "loss": 0.1391,
+      "step": 22140
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 0.0009538516151934684,
+      "loss": 0.1266,
+      "step": 22150
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 0.0009537806176783813,
+      "loss": 0.1598,
+      "step": 22160
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0009537096201632944,
+      "loss": 0.1071,
+      "step": 22170
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 0.0009536386226482073,
+      "loss": 0.1529,
+      "step": 22180
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 0.0009535676251331203,
+      "loss": 0.1626,
+      "step": 22190
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0009534966276180334,
+      "loss": 0.0999,
+      "step": 22200
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 0.0009534256301029463,
+      "loss": 0.1244,
+      "step": 22210
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.0009533546325878595,
+      "loss": 0.1518,
+      "step": 22220
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 0.0009532836350727724,
+      "loss": 0.1589,
+      "step": 22230
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 0.0009532126375576855,
+      "loss": 0.128,
+      "step": 22240
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.0009531416400425985,
+      "loss": 0.155,
+      "step": 22250
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 0.0009530706425275115,
+      "loss": 0.1533,
+      "step": 22260
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 0.0009529996450124245,
+      "loss": 0.1312,
+      "step": 22270
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0009529286474973377,
+      "loss": 0.1448,
+      "step": 22280
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 0.0009528576499822506,
+      "loss": 0.1474,
+      "step": 22290
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 0.0009527866524671637,
+      "loss": 0.1018,
+      "step": 22300
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0009527156549520767,
+      "loss": 0.1606,
+      "step": 22310
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 0.0009526446574369897,
+      "loss": 0.1233,
+      "step": 22320
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.0009525736599219027,
+      "loss": 0.1554,
+      "step": 22330
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 0.0009525026624068159,
+      "loss": 0.1245,
+      "step": 22340
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 0.0009524316648917288,
+      "loss": 0.1178,
+      "step": 22350
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.0009523606673766419,
+      "loss": 0.1328,
+      "step": 22360
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 0.0009522896698615548,
+      "loss": 0.1562,
+      "step": 22370
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 0.0009522186723464679,
+      "loss": 0.1414,
+      "step": 22380
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.0009521476748313809,
+      "loss": 0.1537,
+      "step": 22390
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 0.000952076677316294,
+      "loss": 0.1343,
+      "step": 22400
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 0.000952005679801207,
+      "loss": 0.1536,
+      "step": 22410
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.0009519346822861201,
+      "loss": 0.1516,
+      "step": 22420
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 0.000951863684771033,
+      "loss": 0.1512,
+      "step": 22430
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.0009517926872559461,
+      "loss": 0.1475,
+      "step": 22440
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 0.000951721689740859,
+      "loss": 0.1431,
+      "step": 22450
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 0.0009516506922257722,
+      "loss": 0.1415,
+      "step": 22460
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0009515796947106852,
+      "loss": 0.1334,
+      "step": 22470
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 0.0009515086971955982,
+      "loss": 0.1613,
+      "step": 22480
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 0.0009514376996805112,
+      "loss": 0.1442,
+      "step": 22490
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.0009513667021654242,
+      "loss": 0.1368,
+      "step": 22500
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 0.0009512957046503372,
+      "loss": 0.1551,
+      "step": 22510
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 0.0009512247071352503,
+      "loss": 0.1346,
+      "step": 22520
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.0009511537096201633,
+      "loss": 0.1497,
+      "step": 22530
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 0.0009510827121050763,
+      "loss": 0.1537,
+      "step": 22540
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 0.0009510117145899894,
+      "loss": 0.1407,
+      "step": 22550
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0009509407170749023,
+      "loss": 0.1513,
+      "step": 22560
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 0.0009508697195598154,
+      "loss": 0.1479,
+      "step": 22570
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.0009507987220447285,
+      "loss": 0.1532,
+      "step": 22580
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 0.0009507277245296415,
+      "loss": 0.1189,
+      "step": 22590
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 0.0009506567270145545,
+      "loss": 0.1349,
+      "step": 22600
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.0009505857294994676,
+      "loss": 0.1508,
+      "step": 22610
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 0.0009505147319843805,
+      "loss": 0.1711,
+      "step": 22620
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 0.0009504437344692936,
+      "loss": 0.1591,
+      "step": 22630
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.0009503727369542066,
+      "loss": 0.1572,
+      "step": 22640
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 0.0009503017394391197,
+      "loss": 0.1153,
+      "step": 22650
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 0.0009502307419240327,
+      "loss": 0.1261,
+      "step": 22660
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.0009501597444089457,
+      "loss": 0.1392,
+      "step": 22670
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 0.0009500887468938587,
+      "loss": 0.1275,
+      "step": 22680
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.0009500177493787718,
+      "loss": 0.1472,
+      "step": 22690
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 0.0009499467518636848,
+      "loss": 0.1568,
+      "step": 22700
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 0.0009498757543485979,
+      "loss": 0.1386,
+      "step": 22710
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0009498047568335108,
+      "loss": 0.1265,
+      "step": 22720
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 0.0009497337593184239,
+      "loss": 0.1428,
+      "step": 22730
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 0.0009496627618033369,
+      "loss": 0.175,
+      "step": 22740
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.0009495917642882499,
+      "loss": 0.1614,
+      "step": 22750
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 0.000949520766773163,
+      "loss": 0.1728,
+      "step": 22760
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 0.0009494497692580761,
+      "loss": 0.1589,
+      "step": 22770
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.000949378771742989,
+      "loss": 0.1652,
+      "step": 22780
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 0.0009493077742279021,
+      "loss": 0.1545,
+      "step": 22790
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.000949236776712815,
+      "loss": 0.1602,
+      "step": 22800
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 0.000949165779197728,
+      "loss": 0.1615,
+      "step": 22810
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 0.0009490947816826411,
+      "loss": 0.167,
+      "step": 22820
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0009490237841675541,
+      "loss": 0.145,
+      "step": 22830
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 0.0009489527866524672,
+      "loss": 0.15,
+      "step": 22840
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 0.0009488817891373802,
+      "loss": 0.141,
+      "step": 22850
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.0009488107916222932,
+      "loss": 0.1568,
+      "step": 22860
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 0.0009487397941072062,
+      "loss": 0.1405,
+      "step": 22870
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.0009486687965921193,
+      "loss": 0.1496,
+      "step": 22880
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0009485977990770323,
+      "loss": 0.1532,
+      "step": 22890
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 0.0009485268015619454,
+      "loss": 0.1574,
+      "step": 22900
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 0.0009484558040468583,
+      "loss": 0.144,
+      "step": 22910
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0009483848065317714,
+      "loss": 0.1783,
+      "step": 22920
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 0.0009483138090166844,
+      "loss": 0.1661,
+      "step": 22930
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0009482428115015974,
+      "loss": 0.142,
+      "step": 22940
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 0.0009481718139865105,
+      "loss": 0.1407,
+      "step": 22950
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 0.0009481008164714236,
+      "loss": 0.146,
+      "step": 22960
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0009480298189563365,
+      "loss": 0.1423,
+      "step": 22970
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 0.0009479588214412496,
+      "loss": 0.1735,
+      "step": 22980
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 0.0009478878239261626,
+      "loss": 0.1355,
+      "step": 22990
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0009478168264110756,
+      "loss": 0.1344,
+      "step": 23000
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 0.0009477458288959887,
+      "loss": 0.1427,
+      "step": 23010
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 0.0009476748313809018,
+      "loss": 0.1657,
+      "step": 23020
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.0009476038338658147,
+      "loss": 0.1527,
+      "step": 23030
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 0.0009475328363507278,
+      "loss": 0.1503,
+      "step": 23040
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0009474618388356407,
+      "loss": 0.1619,
+      "step": 23050
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 0.0009473908413205538,
+      "loss": 0.1645,
+      "step": 23060
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 0.0009473198438054669,
+      "loss": 0.1143,
+      "step": 23070
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.0009472488462903799,
+      "loss": 0.1702,
+      "step": 23080
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 0.0009471778487752929,
+      "loss": 0.1313,
+      "step": 23090
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 0.000947106851260206,
+      "loss": 0.1716,
+      "step": 23100
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0009470358537451189,
+      "loss": 0.1421,
+      "step": 23110
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 0.0009469648562300319,
+      "loss": 0.1407,
+      "step": 23120
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 0.000946893858714945,
+      "loss": 0.1345,
+      "step": 23130
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.000946822861199858,
+      "loss": 0.162,
+      "step": 23140
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 0.0009467518636847711,
+      "loss": 0.1601,
+      "step": 23150
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.000946680866169684,
+      "loss": 0.1415,
+      "step": 23160
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 0.0009466098686545971,
+      "loss": 0.182,
+      "step": 23170
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 0.00094653887113951,
+      "loss": 0.1519,
+      "step": 23180
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.0009464678736244232,
+      "loss": 0.1856,
+      "step": 23190
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 0.0009463968761093362,
+      "loss": 0.1641,
+      "step": 23200
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 0.0009463258785942493,
+      "loss": 0.1401,
+      "step": 23210
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0009462548810791622,
+      "loss": 0.1536,
+      "step": 23220
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 0.0009461838835640753,
+      "loss": 0.1346,
+      "step": 23230
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 0.0009461128860489882,
+      "loss": 0.1536,
+      "step": 23240
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.0009460418885339014,
+      "loss": 0.1416,
+      "step": 23250
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 0.0009459708910188144,
+      "loss": 0.1659,
+      "step": 23260
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 0.0009458998935037274,
+      "loss": 0.1343,
+      "step": 23270
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0009458288959886404,
+      "loss": 0.1592,
+      "step": 23280
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 0.0009457578984735535,
+      "loss": 0.1342,
+      "step": 23290
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0009456869009584664,
+      "loss": 0.1514,
+      "step": 23300
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 0.0009456159034433796,
+      "loss": 0.1519,
+      "step": 23310
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 0.0009455449059282925,
+      "loss": 0.1524,
+      "step": 23320
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.0009454739084132056,
+      "loss": 0.1608,
+      "step": 23330
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 0.0009454029108981186,
+      "loss": 0.1324,
+      "step": 23340
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 0.0009453319133830316,
+      "loss": 0.1542,
+      "step": 23350
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.0009452609158679446,
+      "loss": 0.1339,
+      "step": 23360
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 0.0009451899183528578,
+      "loss": 0.1496,
+      "step": 23370
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 0.0009451189208377707,
+      "loss": 0.1244,
+      "step": 23380
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0009450479233226838,
+      "loss": 0.1439,
+      "step": 23390
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 0.0009449769258075968,
+      "loss": 0.1165,
+      "step": 23400
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.0009449059282925098,
+      "loss": 0.1267,
+      "step": 23410
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 0.0009448349307774228,
+      "loss": 0.1383,
+      "step": 23420
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 0.0009447639332623357,
+      "loss": 0.1468,
+      "step": 23430
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0009446929357472489,
+      "loss": 0.1508,
+      "step": 23440
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 0.0009446219382321619,
+      "loss": 0.1566,
+      "step": 23450
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 0.0009445509407170749,
+      "loss": 0.1399,
+      "step": 23460
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.0009444799432019879,
+      "loss": 0.1656,
+      "step": 23470
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 0.000944408945686901,
+      "loss": 0.1503,
+      "step": 23480
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 0.0009443379481718139,
+      "loss": 0.1385,
+      "step": 23490
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.0009442669506567271,
+      "loss": 0.1249,
+      "step": 23500
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 0.00094419595314164,
+      "loss": 0.1565,
+      "step": 23510
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 0.0009441249556265531,
+      "loss": 0.1236,
+      "step": 23520
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0009440539581114661,
+      "loss": 0.1381,
+      "step": 23530
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 0.0009439829605963791,
+      "loss": 0.1496,
+      "step": 23540
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0009439119630812921,
+      "loss": 0.1309,
+      "step": 23550
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 0.0009438409655662053,
+      "loss": 0.1029,
+      "step": 23560
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 0.0009437699680511182,
+      "loss": 0.1542,
+      "step": 23570
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0009436989705360313,
+      "loss": 0.1186,
+      "step": 23580
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 0.0009436279730209442,
+      "loss": 0.1395,
+      "step": 23590
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 0.0009435569755058573,
+      "loss": 0.1368,
+      "step": 23600
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.0009434859779907703,
+      "loss": 0.1187,
+      "step": 23610
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 0.0009434149804756834,
+      "loss": 0.1232,
+      "step": 23620
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 0.0009433439829605964,
+      "loss": 0.1449,
+      "step": 23630
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0009432729854455095,
+      "loss": 0.1134,
+      "step": 23640
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 0.0009432019879304224,
+      "loss": 0.1189,
+      "step": 23650
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.0009431309904153355,
+      "loss": 0.1534,
+      "step": 23660
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 0.0009430599929002485,
+      "loss": 0.1376,
+      "step": 23670
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 0.0009429889953851616,
+      "loss": 0.1442,
+      "step": 23680
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.0009429179978700746,
+      "loss": 0.1275,
+      "step": 23690
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 0.0009428470003549877,
+      "loss": 0.1406,
+      "step": 23700
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 0.0009427760028399006,
+      "loss": 0.1367,
+      "step": 23710
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.0009427050053248137,
+      "loss": 0.149,
+      "step": 23720
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 0.0009426340078097266,
+      "loss": 0.1376,
+      "step": 23730
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 0.0009425630102946398,
+      "loss": 0.1431,
+      "step": 23740
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.0009424920127795528,
+      "loss": 0.1326,
+      "step": 23750
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 0.0009424210152644657,
+      "loss": 0.1056,
+      "step": 23760
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.0009423500177493788,
+      "loss": 0.1506,
+      "step": 23770
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 0.0009422790202342917,
+      "loss": 0.153,
+      "step": 23780
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 0.0009422080227192048,
+      "loss": 0.1368,
+      "step": 23790
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.0009421370252041179,
+      "loss": 0.1508,
+      "step": 23800
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 0.000942066027689031,
+      "loss": 0.1398,
+      "step": 23810
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 0.0009419950301739439,
+      "loss": 0.1396,
+      "step": 23820
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.000941924032658857,
+      "loss": 0.1394,
+      "step": 23830
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 0.0009418530351437699,
+      "loss": 0.1578,
+      "step": 23840
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 0.000941782037628683,
+      "loss": 0.1626,
+      "step": 23850
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0009417110401135961,
+      "loss": 0.1428,
+      "step": 23860
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 0.0009416400425985091,
+      "loss": 0.1478,
+      "step": 23870
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 0.0009415690450834221,
+      "loss": 0.1482,
+      "step": 23880
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.0009414980475683352,
+      "loss": 0.1329,
+      "step": 23890
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 0.0009414270500532481,
+      "loss": 0.1287,
+      "step": 23900
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0009413560525381612,
+      "loss": 0.1443,
+      "step": 23910
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 0.0009412850550230742,
+      "loss": 0.1398,
+      "step": 23920
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 0.0009412140575079873,
+      "loss": 0.1211,
+      "step": 23930
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0009411430599929003,
+      "loss": 0.137,
+      "step": 23940
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 0.0009410720624778133,
+      "loss": 0.1539,
+      "step": 23950
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 0.0009410010649627263,
+      "loss": 0.1305,
+      "step": 23960
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0009409300674476394,
+      "loss": 0.1501,
+      "step": 23970
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 0.0009408590699325523,
+      "loss": 0.1517,
+      "step": 23980
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 0.0009407880724174655,
+      "loss": 0.1436,
+      "step": 23990
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.0009407170749023784,
+      "loss": 0.1175,
+      "step": 24000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 0.0009406460773872915,
+      "loss": 0.1333,
+      "step": 24010
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0009405750798722045,
+      "loss": 0.1091,
+      "step": 24020
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 0.0009405040823571175,
+      "loss": 0.143,
+      "step": 24030
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 0.0009404330848420305,
+      "loss": 0.1213,
+      "step": 24040
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.0009403620873269437,
+      "loss": 0.1295,
+      "step": 24050
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 0.0009402910898118566,
+      "loss": 0.1312,
+      "step": 24060
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0009402200922967696,
+      "loss": 0.1402,
+      "step": 24070
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0009401490947816827,
+      "loss": 0.1299,
+      "step": 24080
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 0.0009400780972665956,
+      "loss": 0.1399,
+      "step": 24090
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 0.0009400070997515087,
+      "loss": 0.1339,
+      "step": 24100
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.0009399361022364217,
+      "loss": 0.1145,
+      "step": 24110
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 0.0009398651047213348,
+      "loss": 0.1567,
+      "step": 24120
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0009397941072062478,
+      "loss": 0.1406,
+      "step": 24130
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 0.0009397231096911608,
+      "loss": 0.1522,
+      "step": 24140
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 0.0009396521121760738,
+      "loss": 0.1274,
+      "step": 24150
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.0009395811146609869,
+      "loss": 0.1309,
+      "step": 24160
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 0.0009395101171458999,
+      "loss": 0.1333,
+      "step": 24170
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 0.000939439119630813,
+      "loss": 0.1439,
+      "step": 24180
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.000939368122115726,
+      "loss": 0.1288,
+      "step": 24190
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 0.000939297124600639,
+      "loss": 0.1154,
+      "step": 24200
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 0.000939226127085552,
+      "loss": 0.1208,
+      "step": 24210
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.000939155129570465,
+      "loss": 0.1275,
+      "step": 24220
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 0.0009390841320553781,
+      "loss": 0.136,
+      "step": 24230
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 0.0009390131345402912,
+      "loss": 0.1182,
+      "step": 24240
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0009389421370252041,
+      "loss": 0.136,
+      "step": 24250
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 0.0009388711395101172,
+      "loss": 0.1294,
+      "step": 24260
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0009388001419950302,
+      "loss": 0.1553,
+      "step": 24270
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 0.0009387291444799432,
+      "loss": 0.1494,
+      "step": 24280
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 0.0009386581469648563,
+      "loss": 0.1119,
+      "step": 24290
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0009385871494497694,
+      "loss": 0.1552,
+      "step": 24300
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 0.0009385161519346823,
+      "loss": 0.1289,
+      "step": 24310
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 0.0009384451544195954,
+      "loss": 0.1435,
+      "step": 24320
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.0009383741569045083,
+      "loss": 0.1491,
+      "step": 24330
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 0.0009383031593894214,
+      "loss": 0.1536,
+      "step": 24340
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 0.0009382321618743345,
+      "loss": 0.1418,
+      "step": 24350
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0009381611643592475,
+      "loss": 0.1373,
+      "step": 24360
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 0.0009380901668441605,
+      "loss": 0.1431,
+      "step": 24370
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.0009380191693290734,
+      "loss": 0.1286,
+      "step": 24380
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 0.0009379481718139865,
+      "loss": 0.1437,
+      "step": 24390
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 0.0009378771742988995,
+      "loss": 0.184,
+      "step": 24400
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0009378061767838126,
+      "loss": 0.1034,
+      "step": 24410
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 0.0009377351792687256,
+      "loss": 0.1458,
+      "step": 24420
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 0.0009376641817536387,
+      "loss": 0.1409,
+      "step": 24430
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0009375931842385516,
+      "loss": 0.1224,
+      "step": 24440
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 0.0009375221867234647,
+      "loss": 0.1531,
+      "step": 24450
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 0.0009374511892083777,
+      "loss": 0.1554,
+      "step": 24460
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0009373801916932908,
+      "loss": 0.145,
+      "step": 24470
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 0.0009373091941782038,
+      "loss": 0.1375,
+      "step": 24480
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0009372381966631168,
+      "loss": 0.1538,
+      "step": 24490
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 0.0009371671991480298,
+      "loss": 0.1313,
+      "step": 24500
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 0.0009370962016329429,
+      "loss": 0.1241,
+      "step": 24510
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.0009370252041178558,
+      "loss": 0.1408,
+      "step": 24520
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 0.000936954206602769,
+      "loss": 0.1285,
+      "step": 24530
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 0.000936883209087682,
+      "loss": 0.1444,
+      "step": 24540
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.000936812211572595,
+      "loss": 0.1534,
+      "step": 24550
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 0.000936741214057508,
+      "loss": 0.1407,
+      "step": 24560
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 0.000936670216542421,
+      "loss": 0.1189,
+      "step": 24570
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.000936599219027334,
+      "loss": 0.1567,
+      "step": 24580
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 0.0009365282215122471,
+      "loss": 0.149,
+      "step": 24590
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 0.0009364572239971601,
+      "loss": 0.1743,
+      "step": 24600
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.0009363862264820732,
+      "loss": 0.158,
+      "step": 24610
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 0.0009363152289669862,
+      "loss": 0.1391,
+      "step": 24620
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0009362442314518992,
+      "loss": 0.147,
+      "step": 24630
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 0.0009361732339368122,
+      "loss": 0.1432,
+      "step": 24640
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 0.0009361022364217253,
+      "loss": 0.1382,
+      "step": 24650
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.0009360312389066383,
+      "loss": 0.14,
+      "step": 24660
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 0.0009359602413915513,
+      "loss": 0.1489,
+      "step": 24670
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 0.0009358892438764643,
+      "loss": 0.1443,
+      "step": 24680
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.0009358182463613773,
+      "loss": 0.1329,
+      "step": 24690
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 0.0009357472488462904,
+      "loss": 0.1445,
+      "step": 24700
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 0.0009356762513312033,
+      "loss": 0.159,
+      "step": 24710
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.0009356052538161165,
+      "loss": 0.1628,
+      "step": 24720
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 0.0009355342563010295,
+      "loss": 0.1636,
+      "step": 24730
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.0009354632587859425,
+      "loss": 0.1332,
+      "step": 24740
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 0.0009353922612708555,
+      "loss": 0.1685,
+      "step": 24750
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 0.0009353212637557686,
+      "loss": 0.1439,
+      "step": 24760
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0009352502662406815,
+      "loss": 0.1545,
+      "step": 24770
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 0.0009351792687255947,
+      "loss": 0.173,
+      "step": 24780
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 0.0009351082712105076,
+      "loss": 0.1309,
+      "step": 24790
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.0009350372736954207,
+      "loss": 0.1452,
+      "step": 24800
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 0.0009349662761803337,
+      "loss": 0.1527,
+      "step": 24810
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 0.0009348952786652467,
+      "loss": 0.1403,
+      "step": 24820
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.0009348242811501597,
+      "loss": 0.1439,
+      "step": 24830
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 0.0009347532836350729,
+      "loss": 0.1238,
+      "step": 24840
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.0009346822861199858,
+      "loss": 0.1268,
+      "step": 24850
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 0.0009346112886048989,
+      "loss": 0.1617,
+      "step": 24860
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 0.0009345402910898118,
+      "loss": 0.1181,
+      "step": 24870
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0009344692935747249,
+      "loss": 0.1561,
+      "step": 24880
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 0.0009343982960596379,
+      "loss": 0.1353,
+      "step": 24890
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 0.000934327298544551,
+      "loss": 0.1284,
+      "step": 24900
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.000934256301029464,
+      "loss": 0.1271,
+      "step": 24910
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 0.0009341853035143771,
+      "loss": 0.1071,
+      "step": 24920
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 0.00093411430599929,
+      "loss": 0.1475,
+      "step": 24930
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.0009340433084842031,
+      "loss": 0.1313,
+      "step": 24940
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 0.000933972310969116,
+      "loss": 0.1228,
+      "step": 24950
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.0009339013134540292,
+      "loss": 0.137,
+      "step": 24960
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 0.0009338303159389422,
+      "loss": 0.1521,
+      "step": 24970
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 0.0009337593184238551,
+      "loss": 0.1228,
+      "step": 24980
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0009336883209087682,
+      "loss": 0.1769,
+      "step": 24990
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 0.0009336173233936812,
+      "loss": 0.1282,
+      "step": 25000
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 0.0009335463258785942,
+      "loss": 0.1443,
+      "step": 25010
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.0009334753283635073,
+      "loss": 0.1371,
+      "step": 25020
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 0.0009334043308484204,
+      "loss": 0.1388,
+      "step": 25030
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 0.0009333333333333333,
+      "loss": 0.1664,
+      "step": 25040
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0009332623358182464,
+      "loss": 0.1286,
+      "step": 25050
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 0.0009331913383031593,
+      "loss": 0.1375,
+      "step": 25060
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 0.0009331203407880724,
+      "loss": 0.1182,
+      "step": 25070
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.0009330493432729855,
+      "loss": 0.1052,
+      "step": 25080
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 0.0009329783457578985,
+      "loss": 0.1171,
+      "step": 25090
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0009329073482428115,
+      "loss": 0.1204,
+      "step": 25100
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 0.0009328363507277246,
+      "loss": 0.1288,
+      "step": 25110
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 0.0009327653532126375,
+      "loss": 0.1363,
+      "step": 25120
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.0009326943556975506,
+      "loss": 0.1242,
+      "step": 25130
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 0.0009326233581824637,
+      "loss": 0.0983,
+      "step": 25140
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 0.0009325523606673767,
+      "loss": 0.1232,
+      "step": 25150
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0009324813631522897,
+      "loss": 0.133,
+      "step": 25160
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 0.0009324103656372028,
+      "loss": 0.1143,
+      "step": 25170
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 0.0009323393681221157,
+      "loss": 0.118,
+      "step": 25180
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.0009322683706070288,
+      "loss": 0.1199,
+      "step": 25190
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 0.0009321973730919417,
+      "loss": 0.1186,
+      "step": 25200
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.0009321263755768549,
+      "loss": 0.1374,
+      "step": 25210
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 0.0009320553780617679,
+      "loss": 0.1354,
+      "step": 25220
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 0.0009319843805466809,
+      "loss": 0.1587,
+      "step": 25230
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.0009319133830315939,
+      "loss": 0.1165,
+      "step": 25240
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 0.000931842385516507,
+      "loss": 0.1248,
+      "step": 25250
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 0.0009317713880014199,
+      "loss": 0.1411,
+      "step": 25260
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0009317003904863331,
+      "loss": 0.1265,
+      "step": 25270
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.000931629392971246,
+      "loss": 0.1404,
+      "step": 25280
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 0.000931558395456159,
+      "loss": 0.1166,
+      "step": 25290
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.0009314873979410721,
+      "loss": 0.1257,
+      "step": 25300
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 0.000931416400425985,
+      "loss": 0.1293,
+      "step": 25310
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 0.0009313454029108981,
+      "loss": 0.1311,
+      "step": 25320
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0009312744053958112,
+      "loss": 0.1424,
+      "step": 25330
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 0.0009312034078807242,
+      "loss": 0.123,
+      "step": 25340
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.0009311324103656372,
+      "loss": 0.1401,
+      "step": 25350
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 0.0009310614128505503,
+      "loss": 0.1472,
+      "step": 25360
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 0.0009309904153354632,
+      "loss": 0.1571,
+      "step": 25370
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0009309194178203763,
+      "loss": 0.1117,
+      "step": 25380
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 0.0009308484203052893,
+      "loss": 0.1385,
+      "step": 25390
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 0.0009307774227902024,
+      "loss": 0.1311,
+      "step": 25400
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.0009307064252751154,
+      "loss": 0.122,
+      "step": 25410
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 0.0009306354277600284,
+      "loss": 0.141,
+      "step": 25420
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 0.0009305644302449414,
+      "loss": 0.1267,
+      "step": 25430
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0009304934327298545,
+      "loss": 0.1027,
+      "step": 25440
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 0.0009304224352147675,
+      "loss": 0.1542,
+      "step": 25450
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.0009303514376996806,
+      "loss": 0.1143,
+      "step": 25460
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 0.0009302804401845935,
+      "loss": 0.1454,
+      "step": 25470
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 0.0009302094426695066,
+      "loss": 0.1523,
+      "step": 25480
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.0009301384451544196,
+      "loss": 0.1088,
+      "step": 25490
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 0.0009300674476393326,
+      "loss": 0.1545,
+      "step": 25500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 0.0009299964501242457,
+      "loss": 0.1418,
+      "step": 25510
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0009299254526091588,
+      "loss": 0.1418,
+      "step": 25520
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 0.0009298544550940717,
+      "loss": 0.1424,
+      "step": 25530
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 0.0009297834575789848,
+      "loss": 0.1374,
+      "step": 25540
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0009297124600638978,
+      "loss": 0.148,
+      "step": 25550
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 0.0009296414625488108,
+      "loss": 0.1299,
+      "step": 25560
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.0009295704650337239,
+      "loss": 0.1407,
+      "step": 25570
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 0.000929499467518637,
+      "loss": 0.1308,
+      "step": 25580
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 0.0009294284700035499,
+      "loss": 0.1344,
+      "step": 25590
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.0009293574724884629,
+      "loss": 0.1084,
+      "step": 25600
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 0.0009292864749733759,
+      "loss": 0.1325,
+      "step": 25610
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 0.0009292154774582889,
+      "loss": 0.133,
+      "step": 25620
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.0009291444799432021,
+      "loss": 0.107,
+      "step": 25630
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 0.000929073482428115,
+      "loss": 0.1486,
+      "step": 25640
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 0.0009290024849130281,
+      "loss": 0.1332,
+      "step": 25650
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.000928931487397941,
+      "loss": 0.1404,
+      "step": 25660
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 0.0009288604898828541,
+      "loss": 0.1448,
+      "step": 25670
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 0.0009287894923677671,
+      "loss": 0.1351,
+      "step": 25680
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.0009287184948526802,
+      "loss": 0.1293,
+      "step": 25690
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 0.0009286474973375932,
+      "loss": 0.1368,
+      "step": 25700
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.0009285764998225063,
+      "loss": 0.1199,
+      "step": 25710
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 0.0009285055023074192,
+      "loss": 0.1127,
+      "step": 25720
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 0.0009284345047923323,
+      "loss": 0.146,
+      "step": 25730
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.0009283635072772452,
+      "loss": 0.1567,
+      "step": 25740
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 0.0009282925097621584,
+      "loss": 0.1227,
+      "step": 25750
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 0.0009282215122470714,
+      "loss": 0.1116,
+      "step": 25760
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.0009281505147319844,
+      "loss": 0.1299,
+      "step": 25770
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 0.0009280795172168974,
+      "loss": 0.1208,
+      "step": 25780
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 0.0009280085197018105,
+      "loss": 0.1362,
+      "step": 25790
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0009279375221867234,
+      "loss": 0.1251,
+      "step": 25800
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 0.0009278665246716365,
+      "loss": 0.1411,
+      "step": 25810
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0009277955271565496,
+      "loss": 0.1121,
+      "step": 25820
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 0.0009277245296414626,
+      "loss": 0.1616,
+      "step": 25830
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 0.0009276535321263756,
+      "loss": 0.1454,
+      "step": 25840
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0009275825346112887,
+      "loss": 0.122,
+      "step": 25850
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 0.0009275115370962016,
+      "loss": 0.1452,
+      "step": 25860
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 0.0009274405395811147,
+      "loss": 0.1088,
+      "step": 25870
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.0009273695420660277,
+      "loss": 0.1377,
+      "step": 25880
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 0.0009272985445509408,
+      "loss": 0.1186,
+      "step": 25890
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 0.0009272275470358538,
+      "loss": 0.1247,
+      "step": 25900
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0009271565495207667,
+      "loss": 0.0938,
+      "step": 25910
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 0.0009270855520056798,
+      "loss": 0.1685,
+      "step": 25920
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0009270145544905927,
+      "loss": 0.1359,
+      "step": 25930
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 0.0009269435569755059,
+      "loss": 0.1519,
+      "step": 25940
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 0.0009268725594604189,
+      "loss": 0.1279,
+      "step": 25950
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.000926801561945332,
+      "loss": 0.1069,
+      "step": 25960
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 0.0009267305644302449,
+      "loss": 0.1241,
+      "step": 25970
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 0.000926659566915158,
+      "loss": 0.1594,
+      "step": 25980
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0009265885694000709,
+      "loss": 0.1272,
+      "step": 25990
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 0.0009265175718849841,
+      "loss": 0.1267,
+      "step": 26000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 0.0009264465743698971,
+      "loss": 0.1224,
+      "step": 26010
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.0009263755768548101,
+      "loss": 0.1352,
+      "step": 26020
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 0.0009263045793397231,
+      "loss": 0.1276,
+      "step": 26030
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 0.0009262335818246362,
+      "loss": 0.123,
+      "step": 26040
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.0009261625843095491,
+      "loss": 0.1321,
+      "step": 26050
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 0.0009260915867944623,
+      "loss": 0.1449,
+      "step": 26060
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0009260205892793752,
+      "loss": 0.1492,
+      "step": 26070
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 0.0009259495917642883,
+      "loss": 0.1182,
+      "step": 26080
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 0.0009258785942492013,
+      "loss": 0.1407,
+      "step": 26090
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0009258075967341143,
+      "loss": 0.1472,
+      "step": 26100
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 0.0009257365992190273,
+      "loss": 0.1404,
+      "step": 26110
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 0.0009256656017039405,
+      "loss": 0.1511,
+      "step": 26120
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0009255946041888534,
+      "loss": 0.1255,
+      "step": 26130
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 0.0009255236066737665,
+      "loss": 0.1221,
+      "step": 26140
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 0.0009254526091586794,
+      "loss": 0.1251,
+      "step": 26150
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 0.0009253816116435925,
+      "loss": 0.1232,
+      "step": 26160
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 0.0009253106141285055,
+      "loss": 0.1382,
+      "step": 26170
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0009252396166134186,
+      "loss": 0.135,
+      "step": 26180
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 0.0009251686190983316,
+      "loss": 0.1372,
+      "step": 26190
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 0.0009250976215832447,
+      "loss": 0.1124,
+      "step": 26200
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.0009250266240681576,
+      "loss": 0.1464,
+      "step": 26210
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 0.0009249556265530706,
+      "loss": 0.1388,
+      "step": 26220
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 0.0009248846290379837,
+      "loss": 0.1212,
+      "step": 26230
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0009248136315228967,
+      "loss": 0.1517,
+      "step": 26240
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 0.0009247426340078098,
+      "loss": 0.1009,
+      "step": 26250
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 0.0009246716364927227,
+      "loss": 0.1296,
+      "step": 26260
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.0009246006389776358,
+      "loss": 0.1443,
+      "step": 26270
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 0.0009245296414625488,
+      "loss": 0.1436,
+      "step": 26280
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.0009244586439474618,
+      "loss": 0.1104,
+      "step": 26290
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 0.0009243876464323749,
+      "loss": 0.1109,
+      "step": 26300
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 0.000924316648917288,
+      "loss": 0.1459,
+      "step": 26310
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.0009242456514022009,
+      "loss": 0.131,
+      "step": 26320
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 0.000924174653887114,
+      "loss": 0.1423,
+      "step": 26330
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 0.000924103656372027,
+      "loss": 0.1342,
+      "step": 26340
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.00092403265885694,
+      "loss": 0.1241,
+      "step": 26350
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 0.000923961661341853,
+      "loss": 0.1781,
+      "step": 26360
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 0.0009238906638267661,
+      "loss": 0.116,
+      "step": 26370
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.0009238196663116791,
+      "loss": 0.1799,
+      "step": 26380
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 0.0009237486687965922,
+      "loss": 0.1317,
+      "step": 26390
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 0.0009236776712815051,
+      "loss": 0.0983,
+      "step": 26400
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0009236066737664182,
+      "loss": 0.1497,
+      "step": 26410
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 0.0009235356762513312,
+      "loss": 0.1571,
+      "step": 26420
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.0009234646787362443,
+      "loss": 0.1443,
+      "step": 26430
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 0.0009233936812211573,
+      "loss": 0.1457,
+      "step": 26440
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 0.0009233226837060703,
+      "loss": 0.1255,
+      "step": 26450
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0009232516861909833,
+      "loss": 0.1181,
+      "step": 26460
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 0.0009231806886758964,
+      "loss": 0.1382,
+      "step": 26470
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.0009231096911608093,
+      "loss": 0.1318,
+      "step": 26480
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 0.0009230386936457225,
+      "loss": 0.1427,
+      "step": 26490
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 0.0009229676961306355,
+      "loss": 0.1359,
+      "step": 26500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 0.0009228966986155485,
+      "loss": 0.1099,
+      "step": 26510
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0009228257011004615,
+      "loss": 0.1713,
+      "step": 26520
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 0.0009227547035853744,
+      "loss": 0.1392,
+      "step": 26530
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.0009226837060702875,
+      "loss": 0.123,
+      "step": 26540
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 0.0009226127085552006,
+      "loss": 0.1638,
+      "step": 26550
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 0.0009225417110401136,
+      "loss": 0.152,
+      "step": 26560
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.0009224707135250266,
+      "loss": 0.1197,
+      "step": 26570
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 0.0009223997160099397,
+      "loss": 0.1322,
+      "step": 26580
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 0.0009223287184948526,
+      "loss": 0.1272,
+      "step": 26590
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0009222577209797657,
+      "loss": 0.1141,
+      "step": 26600
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 0.0009221867234646788,
+      "loss": 0.122,
+      "step": 26610
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 0.0009221157259495918,
+      "loss": 0.1188,
+      "step": 26620
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0009220447284345048,
+      "loss": 0.1244,
+      "step": 26630
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 0.0009219737309194178,
+      "loss": 0.1568,
+      "step": 26640
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 0.0009219027334043308,
+      "loss": 0.1023,
+      "step": 26650
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.0009218317358892439,
+      "loss": 0.1231,
+      "step": 26660
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 0.0009217607383741569,
+      "loss": 0.1095,
+      "step": 26670
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.00092168974085907,
+      "loss": 0.123,
+      "step": 26680
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 0.000921618743343983,
+      "loss": 0.1071,
+      "step": 26690
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 0.000921547745828896,
+      "loss": 0.1295,
+      "step": 26700
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.000921476748313809,
+      "loss": 0.1227,
+      "step": 26710
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 0.000921405750798722,
+      "loss": 0.1109,
+      "step": 26720
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 0.0009213347532836351,
+      "loss": 0.1039,
+      "step": 26730
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.0009212637557685482,
+      "loss": 0.1291,
+      "step": 26740
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 0.0009211927582534611,
+      "loss": 0.11,
+      "step": 26750
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 0.0009211217607383742,
+      "loss": 0.1387,
+      "step": 26760
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 0.0009210507632232872,
+      "loss": 0.1014,
+      "step": 26770
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 0.0009209797657082002,
+      "loss": 0.1332,
+      "step": 26780
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0009209087681931133,
+      "loss": 0.1355,
+      "step": 26790
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 0.0009208377706780264,
+      "loss": 0.1203,
+      "step": 26800
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 0.0009207667731629393,
+      "loss": 0.1282,
+      "step": 26810
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 0.0009206957756478524,
+      "loss": 0.1236,
+      "step": 26820
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 0.0009206247781327653,
+      "loss": 0.1301,
+      "step": 26830
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 0.0009205537806176783,
+      "loss": 0.1248,
+      "step": 26840
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0009204827831025915,
+      "loss": 0.1393,
+      "step": 26850
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 0.0009204117855875044,
+      "loss": 0.1158,
+      "step": 26860
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 0.0009203407880724175,
+      "loss": 0.1415,
+      "step": 26870
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.0009202697905573305,
+      "loss": 0.1166,
+      "step": 26880
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 0.0009201987930422435,
+      "loss": 0.1273,
+      "step": 26890
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.0009201277955271565,
+      "loss": 0.1322,
+      "step": 26900
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 0.0009200567980120697,
+      "loss": 0.1042,
+      "step": 26910
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 0.0009199858004969826,
+      "loss": 0.1692,
+      "step": 26920
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0009199148029818957,
+      "loss": 0.0913,
+      "step": 26930
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 0.0009198438054668086,
+      "loss": 0.1293,
+      "step": 26940
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 0.0009197728079517217,
+      "loss": 0.1275,
+      "step": 26950
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0009197018104366347,
+      "loss": 0.1086,
+      "step": 26960
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 0.0009196308129215477,
+      "loss": 0.1151,
+      "step": 26970
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 0.0009195598154064608,
+      "loss": 0.1449,
+      "step": 26980
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.0009194888178913739,
+      "loss": 0.1332,
+      "step": 26990
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 0.0009194178203762868,
+      "loss": 0.1148,
+      "step": 27000
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 0.0009193468228611999,
+      "loss": 0.1344,
+      "step": 27010
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.0009192758253461128,
+      "loss": 0.1061,
+      "step": 27020
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 0.0009192048278310259,
+      "loss": 0.1154,
+      "step": 27030
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.000919133830315939,
+      "loss": 0.0995,
+      "step": 27040
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 0.000919062832800852,
+      "loss": 0.1339,
+      "step": 27050
+    },
+    {
+      "epoch": 17.29,
+      "learning_rate": 0.000918991835285765,
+      "loss": 0.1048,
+      "step": 27060
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.0009189208377706781,
+      "loss": 0.1519,
+      "step": 27070
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 0.000918849840255591,
+      "loss": 0.078,
+      "step": 27080
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 0.0009187788427405041,
+      "loss": 0.1348,
+      "step": 27090
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.0009187078452254172,
+      "loss": 0.1258,
+      "step": 27100
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 0.0009186368477103302,
+      "loss": 0.1221,
+      "step": 27110
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 0.0009185658501952432,
+      "loss": 0.1198,
+      "step": 27120
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.0009184948526801563,
+      "loss": 0.0938,
+      "step": 27130
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 0.0009184238551650692,
+      "loss": 0.1246,
+      "step": 27140
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0009183528576499822,
+      "loss": 0.132,
+      "step": 27150
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 0.0009182818601348953,
+      "loss": 0.1265,
+      "step": 27160
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 0.0009182108626198083,
+      "loss": 0.1348,
+      "step": 27170
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.0009181398651047214,
+      "loss": 0.1407,
+      "step": 27180
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 0.0009180688675896343,
+      "loss": 0.1409,
+      "step": 27190
+    },
+    {
+      "epoch": 17.38,
+      "learning_rate": 0.0009179978700745474,
+      "loss": 0.0996,
+      "step": 27200
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0009179268725594603,
+      "loss": 0.1391,
+      "step": 27210
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 0.0009178558750443735,
+      "loss": 0.1252,
+      "step": 27220
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 0.0009177848775292865,
+      "loss": 0.1404,
+      "step": 27230
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.0009177138800141995,
+      "loss": 0.1238,
+      "step": 27240
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 0.0009176428824991125,
+      "loss": 0.126,
+      "step": 27250
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0009175718849840256,
+      "loss": 0.1264,
+      "step": 27260
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 0.0009175008874689385,
+      "loss": 0.1355,
+      "step": 27270
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 0.0009174298899538517,
+      "loss": 0.1192,
+      "step": 27280
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0009173588924387647,
+      "loss": 0.1355,
+      "step": 27290
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 0.0009172878949236777,
+      "loss": 0.1259,
+      "step": 27300
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 0.0009172168974085907,
+      "loss": 0.1207,
+      "step": 27310
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0009171458998935038,
+      "loss": 0.111,
+      "step": 27320
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 0.0009170749023784167,
+      "loss": 0.1248,
+      "step": 27330
+    },
+    {
+      "epoch": 17.47,
+      "learning_rate": 0.0009170039048633299,
+      "loss": 0.1155,
+      "step": 27340
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.0009169329073482428,
+      "loss": 0.1202,
+      "step": 27350
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 0.0009168619098331559,
+      "loss": 0.1468,
+      "step": 27360
+    },
+    {
+      "epoch": 17.49,
+      "learning_rate": 0.0009167909123180689,
+      "loss": 0.1278,
+      "step": 27370
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0009167199148029819,
+      "loss": 0.1361,
+      "step": 27380
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 0.0009166489172878949,
+      "loss": 0.1249,
+      "step": 27390
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.0009165779197728081,
+      "loss": 0.1392,
+      "step": 27400
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 0.000916506922257721,
+      "loss": 0.1001,
+      "step": 27410
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 0.0009164359247426341,
+      "loss": 0.1354,
+      "step": 27420
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.000916364927227547,
+      "loss": 0.1063,
+      "step": 27430
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 0.0009162939297124601,
+      "loss": 0.1433,
+      "step": 27440
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 0.0009162229321973731,
+      "loss": 0.1294,
+      "step": 27450
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.0009161519346822862,
+      "loss": 0.1183,
+      "step": 27460
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 0.0009160809371671992,
+      "loss": 0.1353,
+      "step": 27470
+    },
+    {
+      "epoch": 17.56,
+      "learning_rate": 0.0009160099396521122,
+      "loss": 0.1012,
+      "step": 27480
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0009159389421370252,
+      "loss": 0.1388,
+      "step": 27490
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 0.0009158679446219382,
+      "loss": 0.1329,
+      "step": 27500
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.0009157969471068513,
+      "loss": 0.1465,
+      "step": 27510
+    },
+    {
+      "epoch": 17.58,
+      "learning_rate": 0.0009157259495917643,
+      "loss": 0.1405,
+      "step": 27520
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 0.0009156549520766774,
+      "loss": 0.1375,
+      "step": 27530
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.0009155839545615903,
+      "loss": 0.1431,
+      "step": 27540
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 0.0009155129570465034,
+      "loss": 0.1102,
+      "step": 27550
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 0.0009154419595314164,
+      "loss": 0.1477,
+      "step": 27560
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0009153709620163294,
+      "loss": 0.1203,
+      "step": 27570
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 0.0009152999645012424,
+      "loss": 0.142,
+      "step": 27580
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 0.0009152289669861556,
+      "loss": 0.1107,
+      "step": 27590
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0009151579694710685,
+      "loss": 0.0862,
+      "step": 27600
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 0.0009150869719559816,
+      "loss": 0.1198,
+      "step": 27610
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.0009150159744408945,
+      "loss": 0.1224,
+      "step": 27620
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 0.0009149449769258076,
+      "loss": 0.1223,
+      "step": 27630
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 0.0009148739794107206,
+      "loss": 0.1656,
+      "step": 27640
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.0009148029818956337,
+      "loss": 0.1391,
+      "step": 27650
+    },
+    {
+      "epoch": 17.67,
+      "learning_rate": 0.0009147319843805467,
+      "loss": 0.1391,
+      "step": 27660
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 0.0009146609868654598,
+      "loss": 0.1292,
+      "step": 27670
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0009145899893503727,
+      "loss": 0.1195,
+      "step": 27680
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0009145189918352858,
+      "loss": 0.1563,
+      "step": 27690
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 0.0009144479943201987,
+      "loss": 0.1435,
+      "step": 27700
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0009143769968051119,
+      "loss": 0.1365,
+      "step": 27710
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 0.0009143059992900249,
+      "loss": 0.1296,
+      "step": 27720
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 0.000914235001774938,
+      "loss": 0.1118,
+      "step": 27730
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.0009141640042598509,
+      "loss": 0.115,
+      "step": 27740
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 0.000914093006744764,
+      "loss": 0.1173,
+      "step": 27750
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.0009140220092296769,
+      "loss": 0.1318,
+      "step": 27760
+    },
+    {
+      "epoch": 17.74,
+      "learning_rate": 0.0009139510117145901,
+      "loss": 0.1547,
+      "step": 27770
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 0.0009138800141995031,
+      "loss": 0.1173,
+      "step": 27780
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.000913809016684416,
+      "loss": 0.1173,
+      "step": 27790
+    },
+    {
+      "epoch": 17.76,
+      "learning_rate": 0.0009137380191693291,
+      "loss": 0.1154,
+      "step": 27800
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 0.000913667021654242,
+      "loss": 0.1305,
+      "step": 27810
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.0009135960241391551,
+      "loss": 0.1241,
+      "step": 27820
+    },
+    {
+      "epoch": 17.78,
+      "learning_rate": 0.0009135250266240682,
+      "loss": 0.1159,
+      "step": 27830
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 0.0009134540291089812,
+      "loss": 0.1501,
+      "step": 27840
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0009133830315938942,
+      "loss": 0.1342,
+      "step": 27850
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 0.0009133120340788073,
+      "loss": 0.1278,
+      "step": 27860
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0009132410365637202,
+      "loss": 0.134,
+      "step": 27870
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 0.0009131700390486333,
+      "loss": 0.1363,
+      "step": 27880
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 0.0009130990415335464,
+      "loss": 0.1314,
+      "step": 27890
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.0009130280440184594,
+      "loss": 0.1324,
+      "step": 27900
+    },
+    {
+      "epoch": 17.83,
+      "learning_rate": 0.0009129570465033724,
+      "loss": 0.1197,
+      "step": 27910
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 0.0009128860489882854,
+      "loss": 0.1307,
+      "step": 27920
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0009128150514731984,
+      "loss": 0.1197,
+      "step": 27930
+    },
+    {
+      "epoch": 17.85,
+      "learning_rate": 0.0009127440539581115,
+      "loss": 0.1454,
+      "step": 27940
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 0.0009126730564430245,
+      "loss": 0.1155,
+      "step": 27950
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 0.0009126020589279376,
+      "loss": 0.1238,
+      "step": 27960
+    },
+    {
+      "epoch": 17.87,
+      "learning_rate": 0.0009125310614128506,
+      "loss": 0.102,
+      "step": 27970
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.0009124600638977636,
+      "loss": 0.1501,
+      "step": 27980
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 0.0009123890663826766,
+      "loss": 0.1435,
+      "step": 27990
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 0.0009123180688675897,
+      "loss": 0.1381,
+      "step": 28000
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0009122470713525027,
+      "loss": 0.1478,
+      "step": 28010
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 0.0009121760738374158,
+      "loss": 0.1608,
+      "step": 28020
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 0.0009121050763223287,
+      "loss": 0.16,
+      "step": 28030
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.0009120340788072418,
+      "loss": 0.1383,
+      "step": 28040
+    },
+    {
+      "epoch": 17.92,
+      "learning_rate": 0.0009119630812921548,
+      "loss": 0.1365,
+      "step": 28050
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 0.0009118920837770678,
+      "loss": 0.1274,
+      "step": 28060
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.0009118210862619809,
+      "loss": 0.1176,
+      "step": 28070
+    },
+    {
+      "epoch": 17.94,
+      "learning_rate": 0.000911750088746894,
+      "loss": 0.1474,
+      "step": 28080
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.0009116790912318069,
+      "loss": 0.1065,
+      "step": 28090
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 0.0009116080937167199,
+      "loss": 0.1241,
+      "step": 28100
+    },
+    {
+      "epoch": 17.96,
+      "learning_rate": 0.000911537096201633,
+      "loss": 0.123,
+      "step": 28110
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0009114660986865459,
+      "loss": 0.1231,
+      "step": 28120
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 0.0009113951011714591,
+      "loss": 0.1273,
+      "step": 28130
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 0.000911324103656372,
+      "loss": 0.136,
+      "step": 28140
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.0009112531061412851,
+      "loss": 0.1333,
+      "step": 28150
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 0.0009111821086261981,
+      "loss": 0.1464,
+      "step": 28160
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 0.0009111111111111111,
+      "loss": 0.1176,
+      "step": 28170
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0009110401135960241,
+      "loss": 0.1254,
+      "step": 28180
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 0.0009109691160809372,
+      "loss": 0.111,
+      "step": 28190
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 0.0009108981185658502,
+      "loss": 0.1208,
+      "step": 28200
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.0009108271210507633,
+      "loss": 0.1233,
+      "step": 28210
+    },
+    {
+      "epoch": 18.03,
+      "learning_rate": 0.0009107561235356762,
+      "loss": 0.097,
+      "step": 28220
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0009106851260205893,
+      "loss": 0.1101,
+      "step": 28230
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 0.0009106141285055023,
+      "loss": 0.0998,
+      "step": 28240
+    },
+    {
+      "epoch": 18.05,
+      "learning_rate": 0.0009105431309904153,
+      "loss": 0.116,
+      "step": 28250
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.0009104721334753284,
+      "loss": 0.1068,
+      "step": 28260
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 0.0009104011359602415,
+      "loss": 0.1056,
+      "step": 28270
+    },
+    {
+      "epoch": 18.07,
+      "learning_rate": 0.0009103301384451544,
+      "loss": 0.1343,
+      "step": 28280
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0009102591409300675,
+      "loss": 0.1656,
+      "step": 28290
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 0.0009101881434149804,
+      "loss": 0.1124,
+      "step": 28300
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 0.0009101171458998935,
+      "loss": 0.1164,
+      "step": 28310
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.0009100461483848066,
+      "loss": 0.0916,
+      "step": 28320
+    },
+    {
+      "epoch": 18.1,
+      "learning_rate": 0.0009099751508697196,
+      "loss": 0.1054,
+      "step": 28330
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0009099041533546326,
+      "loss": 0.1102,
+      "step": 28340
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 0.0009098331558395457,
+      "loss": 0.1293,
+      "step": 28350
+    },
+    {
+      "epoch": 18.12,
+      "learning_rate": 0.0009097621583244586,
+      "loss": 0.1283,
+      "step": 28360
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.0009096911608093717,
+      "loss": 0.1168,
+      "step": 28370
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 0.0009096201632942848,
+      "loss": 0.0809,
+      "step": 28380
+    },
+    {
+      "epoch": 18.14,
+      "learning_rate": 0.0009095491657791978,
+      "loss": 0.1197,
+      "step": 28390
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0009094781682641108,
+      "loss": 0.1149,
+      "step": 28400
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 0.0009094071707490237,
+      "loss": 0.1162,
+      "step": 28410
+    },
+    {
+      "epoch": 18.16,
+      "learning_rate": 0.0009093361732339368,
+      "loss": 0.1258,
+      "step": 28420
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.0009092651757188498,
+      "loss": 0.1266,
+      "step": 28430
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 0.0009091941782037629,
+      "loss": 0.1502,
+      "step": 28440
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 0.0009091231806886759,
+      "loss": 0.1195,
+      "step": 28450
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.000909052183173589,
+      "loss": 0.1035,
+      "step": 28460
+    },
+    {
+      "epoch": 18.19,
+      "learning_rate": 0.0009089811856585019,
+      "loss": 0.1482,
+      "step": 28470
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.000908910188143415,
+      "loss": 0.1187,
+      "step": 28480
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 0.000908839190628328,
+      "loss": 0.1204,
+      "step": 28490
+    },
+    {
+      "epoch": 18.21,
+      "learning_rate": 0.0009087681931132411,
+      "loss": 0.1059,
+      "step": 28500
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0009086971955981541,
+      "loss": 0.1164,
+      "step": 28510
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 0.0009086261980830671,
+      "loss": 0.1168,
+      "step": 28520
+    },
+    {
+      "epoch": 18.23,
+      "learning_rate": 0.0009085552005679801,
+      "loss": 0.1324,
+      "step": 28530
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.0009084842030528932,
+      "loss": 0.124,
+      "step": 28540
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 0.0009084132055378061,
+      "loss": 0.1028,
+      "step": 28550
+    },
+    {
+      "epoch": 18.25,
+      "learning_rate": 0.0009083422080227193,
+      "loss": 0.1057,
+      "step": 28560
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0009082712105076323,
+      "loss": 0.1132,
+      "step": 28570
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 0.0009082002129925453,
+      "loss": 0.1164,
+      "step": 28580
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.0009081292154774583,
+      "loss": 0.1129,
+      "step": 28590
+    },
+    {
+      "epoch": 18.27,
+      "learning_rate": 0.0009080582179623713,
+      "loss": 0.1185,
+      "step": 28600
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 0.0009079872204472843,
+      "loss": 0.1268,
+      "step": 28610
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 0.0009079162229321975,
+      "loss": 0.1184,
+      "step": 28620
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 0.0009078452254171104,
+      "loss": 0.0942,
+      "step": 28630
+    },
+    {
+      "epoch": 18.3,
+      "learning_rate": 0.0009077742279020235,
+      "loss": 0.109,
+      "step": 28640
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.0009077032303869365,
+      "loss": 0.0995,
+      "step": 28650
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 0.0009076322328718495,
+      "loss": 0.1432,
+      "step": 28660
+    },
+    {
+      "epoch": 18.32,
+      "learning_rate": 0.0009075612353567625,
+      "loss": 0.1517,
+      "step": 28670
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0009074902378416757,
+      "loss": 0.1044,
+      "step": 28680
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 0.0009074192403265886,
+      "loss": 0.0846,
+      "step": 28690
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.0009073482428115017,
+      "loss": 0.1275,
+      "step": 28700
+    },
+    {
+      "epoch": 18.34,
+      "learning_rate": 0.0009072772452964146,
+      "loss": 0.0935,
+      "step": 28710
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 0.0009072062477813276,
+      "loss": 0.0898,
+      "step": 28720
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.0009071352502662407,
+      "loss": 0.127,
+      "step": 28730
+    },
+    {
+      "epoch": 18.36,
+      "learning_rate": 0.0009070642527511536,
+      "loss": 0.1172,
+      "step": 28740
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 0.0009069932552360668,
+      "loss": 0.1069,
+      "step": 28750
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0009069222577209798,
+      "loss": 0.1378,
+      "step": 28760
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 0.0009068512602058928,
+      "loss": 0.1268,
+      "step": 28770
+    },
+    {
+      "epoch": 18.39,
+      "learning_rate": 0.0009067802626908058,
+      "loss": 0.0926,
+      "step": 28780
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0009067092651757188,
+      "loss": 0.1208,
+      "step": 28790
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 0.0009066382676606318,
+      "loss": 0.1234,
+      "step": 28800
+    },
+    {
+      "epoch": 18.41,
+      "learning_rate": 0.000906567270145545,
+      "loss": 0.1548,
+      "step": 28810
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.0009064962726304579,
+      "loss": 0.1121,
+      "step": 28820
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 0.000906425275115371,
+      "loss": 0.1457,
+      "step": 28830
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.000906354277600284,
+      "loss": 0.1252,
+      "step": 28840
+    },
+    {
+      "epoch": 18.43,
+      "learning_rate": 0.000906283280085197,
+      "loss": 0.1345,
+      "step": 28850
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 0.00090621228257011,
+      "loss": 0.1101,
+      "step": 28860
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.0009061412850550232,
+      "loss": 0.1146,
+      "step": 28870
+    },
+    {
+      "epoch": 18.45,
+      "learning_rate": 0.0009060702875399361,
+      "loss": 0.1232,
+      "step": 28880
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.0009059992900248492,
+      "loss": 0.1239,
+      "step": 28890
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.0009059282925097621,
+      "loss": 0.1202,
+      "step": 28900
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 0.0009058572949946752,
+      "loss": 0.1124,
+      "step": 28910
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 0.0009057862974795882,
+      "loss": 0.1148,
+      "step": 28920
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0009057152999645013,
+      "loss": 0.1111,
+      "step": 28930
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 0.0009056443024494143,
+      "loss": 0.1243,
+      "step": 28940
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 0.0009055733049343274,
+      "loss": 0.1227,
+      "step": 28950
+    },
+    {
+      "epoch": 18.5,
+      "learning_rate": 0.0009055023074192403,
+      "loss": 0.0993,
+      "step": 28960
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 0.0009054313099041534,
+      "loss": 0.1286,
+      "step": 28970
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0009053603123890663,
+      "loss": 0.1054,
+      "step": 28980
+    },
+    {
+      "epoch": 18.52,
+      "learning_rate": 0.0009052893148739795,
+      "loss": 0.1196,
+      "step": 28990
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 0.0009052183173588925,
+      "loss": 0.1255,
+      "step": 29000
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.0009051473198438055,
+      "loss": 0.1325,
+      "step": 29010
+    },
+    {
+      "epoch": 18.54,
+      "learning_rate": 0.0009050763223287185,
+      "loss": 0.1161,
+      "step": 29020
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 0.0009050053248136315,
+      "loss": 0.1144,
+      "step": 29030
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.0009049343272985445,
+      "loss": 0.1293,
+      "step": 29040
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 0.0009048633297834576,
+      "loss": 0.1284,
+      "step": 29050
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0009047923322683707,
+      "loss": 0.1086,
+      "step": 29060
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 0.0009047213347532836,
+      "loss": 0.1142,
+      "step": 29070
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 0.0009046503372381967,
+      "loss": 0.1223,
+      "step": 29080
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0009045793397231096,
+      "loss": 0.137,
+      "step": 29090
+    },
+    {
+      "epoch": 18.59,
+      "learning_rate": 0.0009045083422080227,
+      "loss": 0.1287,
+      "step": 29100
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 0.0009044373446929358,
+      "loss": 0.1208,
+      "step": 29110
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0009043663471778488,
+      "loss": 0.1086,
+      "step": 29120
+    },
+    {
+      "epoch": 18.61,
+      "learning_rate": 0.0009042953496627618,
+      "loss": 0.125,
+      "step": 29130
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 0.0009042243521476749,
+      "loss": 0.1193,
+      "step": 29140
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.0009041533546325878,
+      "loss": 0.1054,
+      "step": 29150
+    },
+    {
+      "epoch": 18.63,
+      "learning_rate": 0.0009040823571175009,
+      "loss": 0.1231,
+      "step": 29160
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 0.000904011359602414,
+      "loss": 0.1086,
+      "step": 29170
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.000903940362087327,
+      "loss": 0.1391,
+      "step": 29180
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 0.00090386936457224,
+      "loss": 0.1039,
+      "step": 29190
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.000903798367057153,
+      "loss": 0.14,
+      "step": 29200
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 0.000903727369542066,
+      "loss": 0.1216,
+      "step": 29210
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 0.0009036563720269791,
+      "loss": 0.1337,
+      "step": 29220
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0009035853745118921,
+      "loss": 0.1436,
+      "step": 29230
+    },
+    {
+      "epoch": 18.68,
+      "learning_rate": 0.0009035143769968052,
+      "loss": 0.1275,
+      "step": 29240
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 0.0009034433794817182,
+      "loss": 0.1191,
+      "step": 29250
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0009033723819666312,
+      "loss": 0.1228,
+      "step": 29260
+    },
+    {
+      "epoch": 18.7,
+      "learning_rate": 0.0009033013844515442,
+      "loss": 0.1387,
+      "step": 29270
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 0.0009032303869364573,
+      "loss": 0.146,
+      "step": 29280
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.0009031593894213703,
+      "loss": 0.1387,
+      "step": 29290
+    },
+    {
+      "epoch": 18.72,
+      "learning_rate": 0.0009030883919062834,
+      "loss": 0.0893,
+      "step": 29300
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0009030173943911963,
+      "loss": 0.1221,
+      "step": 29310
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 0.0009029463968761094,
+      "loss": 0.1158,
+      "step": 29320
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 0.0009028753993610224,
+      "loss": 0.1393,
+      "step": 29330
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.0009028044018459353,
+      "loss": 0.1195,
+      "step": 29340
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 0.0009027334043308484,
+      "loss": 0.1365,
+      "step": 29350
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 0.0009026624068157615,
+      "loss": 0.1153,
+      "step": 29360
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0009025914093006745,
+      "loss": 0.1396,
+      "step": 29370
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 0.0009025204117855875,
+      "loss": 0.1326,
+      "step": 29380
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 0.0009024494142705005,
+      "loss": 0.1302,
+      "step": 29390
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0009023784167554135,
+      "loss": 0.142,
+      "step": 29400
+    },
+    {
+      "epoch": 18.79,
+      "learning_rate": 0.0009023074192403266,
+      "loss": 0.1146,
+      "step": 29410
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0009022364217252396,
+      "loss": 0.1048,
+      "step": 29420
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 0.0009021654242101527,
+      "loss": 0.1186,
+      "step": 29430
+    },
+    {
+      "epoch": 18.81,
+      "learning_rate": 0.0009020944266950657,
+      "loss": 0.1196,
+      "step": 29440
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.0009020234291799787,
+      "loss": 0.1468,
+      "step": 29450
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 0.0009019524316648917,
+      "loss": 0.1145,
+      "step": 29460
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 0.0009018814341498048,
+      "loss": 0.112,
+      "step": 29470
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0009018104366347178,
+      "loss": 0.1304,
+      "step": 29480
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 0.0009017394391196309,
+      "loss": 0.1395,
+      "step": 29490
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 0.0009016684416045438,
+      "loss": 0.1276,
+      "step": 29500
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0009015974440894569,
+      "loss": 0.097,
+      "step": 29510
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 0.0009015264465743699,
+      "loss": 0.1287,
+      "step": 29520
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 0.0009014554490592829,
+      "loss": 0.1234,
+      "step": 29530
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.000901384451544196,
+      "loss": 0.1509,
+      "step": 29540
+    },
+    {
+      "epoch": 18.88,
+      "learning_rate": 0.0009013134540291091,
+      "loss": 0.1317,
+      "step": 29550
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.000901242456514022,
+      "loss": 0.1198,
+      "step": 29560
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 0.0009011714589989351,
+      "loss": 0.1212,
+      "step": 29570
+    },
+    {
+      "epoch": 18.9,
+      "learning_rate": 0.000901100461483848,
+      "loss": 0.1124,
+      "step": 29580
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0009010294639687611,
+      "loss": 0.1289,
+      "step": 29590
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 0.0009009584664536742,
+      "loss": 0.0943,
+      "step": 29600
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 0.0009008874689385872,
+      "loss": 0.1308,
+      "step": 29610
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0009008164714235002,
+      "loss": 0.1405,
+      "step": 29620
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 0.0009007454739084133,
+      "loss": 0.1067,
+      "step": 29630
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 0.0009006744763933262,
+      "loss": 0.1219,
+      "step": 29640
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0009006034788782392,
+      "loss": 0.1205,
+      "step": 29650
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 0.0009005324813631524,
+      "loss": 0.1326,
+      "step": 29660
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.0009004614838480653,
+      "loss": 0.1112,
+      "step": 29670
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 0.0009003904863329784,
+      "loss": 0.1184,
+      "step": 29680
+    },
+    {
+      "epoch": 18.97,
+      "learning_rate": 0.0009003194888178913,
+      "loss": 0.1426,
+      "step": 29690
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0009002484913028044,
+      "loss": 0.1358,
+      "step": 29700
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 0.0009001774937877174,
+      "loss": 0.1562,
+      "step": 29710
+    },
+    {
+      "epoch": 18.99,
+      "learning_rate": 0.0009001064962726305,
+      "loss": 0.1396,
+      "step": 29720
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.0009000354987575435,
+      "loss": 0.1173,
+      "step": 29730
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 0.0008999645012424566,
+      "loss": 0.1017,
+      "step": 29740
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 0.0008998935037273695,
+      "loss": 0.1158,
+      "step": 29750
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0008998225062122826,
+      "loss": 0.109,
+      "step": 29760
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 0.0008997515086971955,
+      "loss": 0.1313,
+      "step": 29770
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 0.0008996805111821087,
+      "loss": 0.1366,
+      "step": 29780
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.0008996095136670217,
+      "loss": 0.1002,
+      "step": 29790
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 0.0008995385161519347,
+      "loss": 0.0708,
+      "step": 29800
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.0008994675186368477,
+      "loss": 0.1137,
+      "step": 29810
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 0.0008993965211217608,
+      "loss": 0.0956,
+      "step": 29820
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 0.0008993255236066737,
+      "loss": 0.1274,
+      "step": 29830
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.0008992545260915869,
+      "loss": 0.1048,
+      "step": 29840
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 0.0008991835285764999,
+      "loss": 0.0961,
+      "step": 29850
+    },
+    {
+      "epoch": 19.08,
+      "learning_rate": 0.0008991125310614129,
+      "loss": 0.125,
+      "step": 29860
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.0008990415335463259,
+      "loss": 0.1302,
+      "step": 29870
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 0.000898970536031239,
+      "loss": 0.1347,
+      "step": 29880
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 0.0008988995385161519,
+      "loss": 0.1157,
+      "step": 29890
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.0008988285410010651,
+      "loss": 0.1186,
+      "step": 29900
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 0.000898757543485978,
+      "loss": 0.1363,
+      "step": 29910
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0008986865459708911,
+      "loss": 0.1319,
+      "step": 29920
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 0.0008986155484558041,
+      "loss": 0.1121,
+      "step": 29930
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 0.0008985445509407171,
+      "loss": 0.0922,
+      "step": 29940
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.0008984735534256301,
+      "loss": 0.1178,
+      "step": 29950
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 0.000898402555910543,
+      "loss": 0.1193,
+      "step": 29960
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 0.0008983315583954562,
+      "loss": 0.111,
+      "step": 29970
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.0008982605608803692,
+      "loss": 0.0945,
+      "step": 29980
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 0.0008981895633652822,
+      "loss": 0.119,
+      "step": 29990
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 0.0008981185658501952,
+      "loss": 0.1203,
+      "step": 30000
+    }
+  ],
+  "max_steps": 156500,
+  "num_train_epochs": 100,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}