diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7267 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999586007037881,
+  "global_step": 12077,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6559032952475576e-08,
+      "loss": 6.3857,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.311806590495115e-08,
+      "loss": 6.43,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.967709885742673e-08,
+      "loss": 6.4599,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.62361318099023e-08,
+      "loss": 6.4068,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.279516476237789e-08,
+      "loss": 6.5193,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.935419771485346e-08,
+      "loss": 6.3464,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1591323066732903e-07,
+      "loss": 6.234,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.324722636198046e-07,
+      "loss": 6.41,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.490312965722802e-07,
+      "loss": 6.3997,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6559032952475578e-07,
+      "loss": 6.4958,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8214936247723137e-07,
+      "loss": 6.3384,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9870839542970692e-07,
+      "loss": 6.3671,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1526742838218252e-07,
+      "loss": 6.3716,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3182646133465806e-07,
+      "loss": 6.2636,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4838549428713366e-07,
+      "loss": 6.3201,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.649445272396092e-07,
+      "loss": 6.3198,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.815035601920848e-07,
+      "loss": 6.2535,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.980625931445604e-07,
+      "loss": 6.2163,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1462162609703595e-07,
+      "loss": 6.2817,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3118065904951155e-07,
+      "loss": 6.2388,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.477396920019871e-07,
+      "loss": 6.3169,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6429872495446275e-07,
+      "loss": 6.2179,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.808577579069383e-07,
+      "loss": 6.2893,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9741679085941384e-07,
+      "loss": 6.2153,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.139758238118894e-07,
+      "loss": 6.3111,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3053485676436504e-07,
+      "loss": 6.1546,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.470938897168406e-07,
+      "loss": 6.0874,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6365292266931613e-07,
+      "loss": 6.2002,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.802119556217917e-07,
+      "loss": 6.1429,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967709885742673e-07,
+      "loss": 6.187,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.133300215267429e-07,
+      "loss": 6.0405,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.298890544792184e-07,
+      "loss": 6.0928,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.46448087431694e-07,
+      "loss": 5.9895,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.630071203841696e-07,
+      "loss": 6.0476,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.795661533366452e-07,
+      "loss": 6.0089,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.961251862891208e-07,
+      "loss": 6.0452,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.126842192415964e-07,
+      "loss": 5.9284,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.292432521940719e-07,
+      "loss": 5.9368,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.458022851465474e-07,
+      "loss": 5.9543,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.623613180990231e-07,
+      "loss": 5.923,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.789203510514986e-07,
+      "loss": 5.7986,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.954793840039742e-07,
+      "loss": 5.8673,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.120384169564498e-07,
+      "loss": 5.8395,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.285974499089255e-07,
+      "loss": 5.76,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.451564828614009e-07,
+      "loss": 5.8973,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.617155158138766e-07,
+      "loss": 5.8379,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.78274548766352e-07,
+      "loss": 5.7986,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.948335817188277e-07,
+      "loss": 5.7387,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.113926146713033e-07,
+      "loss": 5.6814,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.279516476237788e-07,
+      "loss": 5.7203,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.445106805762544e-07,
+      "loss": 5.5967,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.610697135287301e-07,
+      "loss": 5.6956,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.776287464812055e-07,
+      "loss": 5.6753,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.941877794336812e-07,
+      "loss": 5.612,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.107468123861567e-07,
+      "loss": 5.5828,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.273058453386323e-07,
+      "loss": 5.6709,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.438648782911079e-07,
+      "loss": 5.4728,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.604239112435833e-07,
+      "loss": 5.4503,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.76982944196059e-07,
+      "loss": 5.4991,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935419771485347e-07,
+      "loss": 5.5169,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 5.5044,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0266600430534857e-06,
+      "loss": 5.373,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0432190760059612e-06,
+      "loss": 5.338,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0597781089584368e-06,
+      "loss": 5.3108,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0763371419109125e-06,
+      "loss": 5.3302,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.092896174863388e-06,
+      "loss": 5.249,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1094552078158636e-06,
+      "loss": 5.2276,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1260142407683392e-06,
+      "loss": 5.2067,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1425732737208149e-06,
+      "loss": 5.1474,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1591323066732903e-06,
+      "loss": 5.3222,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.175691339625766e-06,
+      "loss": 5.068,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1922503725782416e-06,
+      "loss": 5.0145,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.208809405530717e-06,
+      "loss": 5.0543,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2253684384831927e-06,
+      "loss": 5.0019,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2419274714356684e-06,
+      "loss": 4.9581,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2584865043881438e-06,
+      "loss": 4.9894,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2750455373406195e-06,
+      "loss": 5.005,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.291604570293095e-06,
+      "loss": 5.0153,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3081636032455706e-06,
+      "loss": 4.8505,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3247226361980462e-06,
+      "loss": 5.0054,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3412816691505219e-06,
+      "loss": 4.9017,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3578407021029973e-06,
+      "loss": 4.869,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3743997350554727e-06,
+      "loss": 4.7603,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.3909587680079484e-06,
+      "loss": 4.7279,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.407517800960424e-06,
+      "loss": 4.9057,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4240768339128997e-06,
+      "loss": 4.7596,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4406358668653753e-06,
+      "loss": 4.752,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.457194899817851e-06,
+      "loss": 4.6213,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4737539327703262e-06,
+      "loss": 4.6757,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4903129657228019e-06,
+      "loss": 4.5897,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5068719986752775e-06,
+      "loss": 4.5763,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5234310316277532e-06,
+      "loss": 4.5543,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5399900645802288e-06,
+      "loss": 4.5276,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.556549097532704e-06,
+      "loss": 4.4686,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5731081304851797e-06,
+      "loss": 4.5276,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.5896671634376554e-06,
+      "loss": 4.4484,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.606226196390131e-06,
+      "loss": 4.38,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6227852293426067e-06,
+      "loss": 4.3073,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6393442622950819e-06,
+      "loss": 4.4981,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6559032952475575e-06,
+      "loss": 4.3106,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6724623282000332e-06,
+      "loss": 4.3446,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6890213611525088e-06,
+      "loss": 4.3745,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7055803941049845e-06,
+      "loss": 4.2771,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7221394270574601e-06,
+      "loss": 4.3331,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7386984600099354e-06,
+      "loss": 4.182,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.755257492962411e-06,
+      "loss": 4.3352,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7718165259148867e-06,
+      "loss": 4.2325,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.7883755588673623e-06,
+      "loss": 4.2712,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.804934591819838e-06,
+      "loss": 4.2565,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8214936247723134e-06,
+      "loss": 4.2287,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.838052657724789e-06,
+      "loss": 4.0568,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8546116906772645e-06,
+      "loss": 4.1905,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8711707236297402e-06,
+      "loss": 4.1169,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8877297565822158e-06,
+      "loss": 4.0983,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9042887895346913e-06,
+      "loss": 4.0022,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9208478224871667e-06,
+      "loss": 4.0904,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9374068554396426e-06,
+      "loss": 3.9949,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.953965888392118e-06,
+      "loss": 4.0095,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970524921344594e-06,
+      "loss": 3.9118,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9870839542970693e-06,
+      "loss": 3.9597,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0036429872495447e-06,
+      "loss": 4.0454,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 3.8968,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.036761053154496e-06,
+      "loss": 3.9477,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0533200861069715e-06,
+      "loss": 4.0273,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0698791190594474e-06,
+      "loss": 3.7427,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.0864381520119224e-06,
+      "loss": 3.846,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1029971849643982e-06,
+      "loss": 3.7779,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1195562179168737e-06,
+      "loss": 3.7887,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1361152508693495e-06,
+      "loss": 3.7965,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.152674283821825e-06,
+      "loss": 3.7958,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1692333167743004e-06,
+      "loss": 3.6792,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.185792349726776e-06,
+      "loss": 3.6996,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.2023513826792517e-06,
+      "loss": 3.5333,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.218910415631727e-06,
+      "loss": 3.563,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.235469448584203e-06,
+      "loss": 3.6404,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.2520284815366785e-06,
+      "loss": 3.5061,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.268587514489154e-06,
+      "loss": 3.5279,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.2851465474416298e-06,
+      "loss": 3.5147,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.301705580394105e-06,
+      "loss": 3.5111,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3182646133465806e-06,
+      "loss": 3.4339,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3348236462990565e-06,
+      "loss": 3.3923,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.351382679251532e-06,
+      "loss": 3.372,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3679417122040074e-06,
+      "loss": 3.4274,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.3845007451564832e-06,
+      "loss": 3.4303,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.4010597781089587e-06,
+      "loss": 3.4981,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.417618811061434e-06,
+      "loss": 3.2494,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.4341778440139096e-06,
+      "loss": 3.2626,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.4507368769663854e-06,
+      "loss": 3.2556,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.467295909918861e-06,
+      "loss": 3.2183,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.4838549428713367e-06,
+      "loss": 3.2121,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.500413975823812e-06,
+      "loss": 3.2533,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5169730087762876e-06,
+      "loss": 3.2347,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5335320417287635e-06,
+      "loss": 3.1927,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.550091074681239e-06,
+      "loss": 3.0843,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.566650107633714e-06,
+      "loss": 3.1224,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.58320914058619e-06,
+      "loss": 3.0779,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.5997681735386652e-06,
+      "loss": 3.0327,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.616327206491141e-06,
+      "loss": 2.9708,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6328862394436165e-06,
+      "loss": 3.1392,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6494452723960924e-06,
+      "loss": 3.0131,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.666004305348568e-06,
+      "loss": 3.0038,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6825633383010437e-06,
+      "loss": 2.9489,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.699122371253519e-06,
+      "loss": 2.9183,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7156814042059946e-06,
+      "loss": 2.8645,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7322404371584705e-06,
+      "loss": 2.8819,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7487994701109455e-06,
+      "loss": 2.8819,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7653585030634213e-06,
+      "loss": 2.7823,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7819175360158968e-06,
+      "loss": 2.7256,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.798476568968372e-06,
+      "loss": 2.7996,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.815035601920848e-06,
+      "loss": 2.713,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8315946348733235e-06,
+      "loss": 2.706,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8481536678257994e-06,
+      "loss": 2.7418,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.864712700778275e-06,
+      "loss": 2.7116,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8812717337307507e-06,
+      "loss": 2.6976,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.897830766683226e-06,
+      "loss": 2.6727,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.914389799635702e-06,
+      "loss": 2.6178,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.930948832588177e-06,
+      "loss": 2.5004,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9475078655406524e-06,
+      "loss": 2.6062,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9640668984931283e-06,
+      "loss": 2.5896,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9806259314456037e-06,
+      "loss": 2.6111,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.997184964398079e-06,
+      "loss": 2.5346,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.013743997350555e-06,
+      "loss": 2.576,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 2.5316,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.0468620632555064e-06,
+      "loss": 2.4878,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.063421096207982e-06,
+      "loss": 2.5387,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.0799801291604577e-06,
+      "loss": 2.464,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.0965391621129327e-06,
+      "loss": 2.4736,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.113098195065408e-06,
+      "loss": 2.3853,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.129657228017884e-06,
+      "loss": 2.4447,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1462162609703594e-06,
+      "loss": 2.3735,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1627752939228353e-06,
+      "loss": 2.3644,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1793343268753107e-06,
+      "loss": 2.4373,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.1958933598277866e-06,
+      "loss": 2.4221,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.212452392780262e-06,
+      "loss": 2.3043,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.2290114257327375e-06,
+      "loss": 2.2928,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.2455704586852133e-06,
+      "loss": 2.3373,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.2621294916376888e-06,
+      "loss": 2.2466,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.2786885245901638e-06,
+      "loss": 2.2875,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.2952475575426396e-06,
+      "loss": 2.1438,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.311806590495115e-06,
+      "loss": 2.1487,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.328365623447591e-06,
+      "loss": 2.161,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3449246564000664e-06,
+      "loss": 2.1174,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3614836893525422e-06,
+      "loss": 2.169,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3780427223050177e-06,
+      "loss": 2.1465,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.3946017552574936e-06,
+      "loss": 2.0953,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.411160788209969e-06,
+      "loss": 2.1087,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.4277198211624444e-06,
+      "loss": 2.0311,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.4442788541149203e-06,
+      "loss": 2.0938,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.4608378870673953e-06,
+      "loss": 2.053,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.4773969200198708e-06,
+      "loss": 2.0856,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.4939559529723466e-06,
+      "loss": 2.0346,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.510514985924822e-06,
+      "loss": 1.9762,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.527074018877298e-06,
+      "loss": 2.0013,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5436330518297734e-06,
+      "loss": 1.8917,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5601920847822492e-06,
+      "loss": 1.9639,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5767511177347247e-06,
+      "loss": 2.012,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.5933101506872005e-06,
+      "loss": 1.9868,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.609869183639676e-06,
+      "loss": 1.9896,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.626428216592151e-06,
+      "loss": 1.8686,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.642987249544627e-06,
+      "loss": 1.9306,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.6595462824971023e-06,
+      "loss": 1.8979,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.676105315449578e-06,
+      "loss": 1.8627,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.6926643484020536e-06,
+      "loss": 1.8561,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.709223381354529e-06,
+      "loss": 1.8493,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.725782414307005e-06,
+      "loss": 1.8237,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7423414472594803e-06,
+      "loss": 1.8373,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.758900480211956e-06,
+      "loss": 1.8354,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7754595131644316e-06,
+      "loss": 1.8166,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.7920185461169075e-06,
+      "loss": 1.7415,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.8085775790693825e-06,
+      "loss": 1.8276,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.825136612021858e-06,
+      "loss": 1.6788,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.841695644974333e-06,
+      "loss": 1.7545,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.85825467792681e-06,
+      "loss": 1.7134,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.874813710879285e-06,
+      "loss": 1.7235,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.8913727438317606e-06,
+      "loss": 1.6864,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.907931776784236e-06,
+      "loss": 1.6798,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.9244908097367114e-06,
+      "loss": 1.665,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.941049842689188e-06,
+      "loss": 1.6521,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.957608875641663e-06,
+      "loss": 1.6432,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.974167908594139e-06,
+      "loss": 1.6523,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.990726941546614e-06,
+      "loss": 1.6178,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0072859744990895e-06,
+      "loss": 1.6118,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.023845007451565e-06,
+      "loss": 1.6448,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 1.6524,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.056963073356517e-06,
+      "loss": 1.5844,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.073522106308992e-06,
+      "loss": 1.5728,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.0900811392614675e-06,
+      "loss": 1.534,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.106640172213943e-06,
+      "loss": 1.5531,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.123199205166418e-06,
+      "loss": 1.534,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.139758238118895e-06,
+      "loss": 1.5183,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.156317271071369e-06,
+      "loss": 1.5182,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.172876304023845e-06,
+      "loss": 1.5267,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.189435336976321e-06,
+      "loss": 1.4795,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2059943699287965e-06,
+      "loss": 1.5563,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.222553402881272e-06,
+      "loss": 1.4825,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.239112435833747e-06,
+      "loss": 1.5297,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.255671468786224e-06,
+      "loss": 1.4852,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.272230501738699e-06,
+      "loss": 1.4829,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.2887895346911745e-06,
+      "loss": 1.4442,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.30534856764365e-06,
+      "loss": 1.4681,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.321907600596125e-06,
+      "loss": 1.4145,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.338466633548601e-06,
+      "loss": 1.4249,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.355025666501076e-06,
+      "loss": 1.4261,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.371584699453552e-06,
+      "loss": 1.4147,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.388143732406028e-06,
+      "loss": 1.4496,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4047027653585034e-06,
+      "loss": 1.4531,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.421261798310979e-06,
+      "loss": 1.4158,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.437820831263454e-06,
+      "loss": 1.3977,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.454379864215931e-06,
+      "loss": 1.3926,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.470938897168406e-06,
+      "loss": 1.3849,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4874979301208815e-06,
+      "loss": 1.4039,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.504056963073357e-06,
+      "loss": 1.3232,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.520615996025832e-06,
+      "loss": 1.3711,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.537175028978308e-06,
+      "loss": 1.3558,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.553734061930783e-06,
+      "loss": 1.3352,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.5702930948832595e-06,
+      "loss": 1.3546,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.586852127835735e-06,
+      "loss": 1.3369,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.60341116078821e-06,
+      "loss": 1.3016,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619970193740686e-06,
+      "loss": 1.3003,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.636529226693161e-06,
+      "loss": 1.3518,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6530882596456376e-06,
+      "loss": 1.2875,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.669647292598113e-06,
+      "loss": 1.3135,
+      "step": 2820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.686206325550588e-06,
+      "loss": 1.302,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.702765358503064e-06,
+      "loss": 1.2536,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.719324391455539e-06,
+      "loss": 1.3275,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.735883424408015e-06,
+      "loss": 1.2927,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.75244245736049e-06,
+      "loss": 1.3173,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7690014903129665e-06,
+      "loss": 1.2696,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.785560523265442e-06,
+      "loss": 1.2797,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.802119556217917e-06,
+      "loss": 1.2277,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.818678589170393e-06,
+      "loss": 1.2779,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.835237622122868e-06,
+      "loss": 1.2377,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.8517966550753445e-06,
+      "loss": 1.2771,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.868355688027819e-06,
+      "loss": 1.2402,
+      "step": 2940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.884914720980295e-06,
+      "loss": 1.2476,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.901473753932771e-06,
+      "loss": 1.2217,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.918032786885246e-06,
+      "loss": 1.2297,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.934591819837722e-06,
+      "loss": 1.204,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.951150852790197e-06,
+      "loss": 1.2238,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9677098857426735e-06,
+      "loss": 1.2229,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.984268918695149e-06,
+      "loss": 1.2204,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.000827951647624e-06,
+      "loss": 1.2156,
+      "step": 3020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.0173869846001e-06,
+      "loss": 1.1825,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.033946017552575e-06,
+      "loss": 1.2509,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.0505050505050515e-06,
+      "loss": 1.2098,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.067064083457527e-06,
+      "loss": 1.1757,
+      "step": 3060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.083623116410002e-06,
+      "loss": 1.1787,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.100182149362478e-06,
+      "loss": 1.1779,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.116741182314954e-06,
+      "loss": 1.1744,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.133300215267428e-06,
+      "loss": 1.1566,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.149859248219904e-06,
+      "loss": 1.1769,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.16641828117238e-06,
+      "loss": 1.1614,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.182977314124855e-06,
+      "loss": 1.1894,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.1995363470773305e-06,
+      "loss": 1.1298,
+      "step": 3140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.216095380029807e-06,
+      "loss": 1.1647,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.232654412982282e-06,
+      "loss": 1.1222,
+      "step": 3160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.249213445934758e-06,
+      "loss": 1.1146,
+      "step": 3170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.265772478887233e-06,
+      "loss": 1.1434,
+      "step": 3180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.2823315118397085e-06,
+      "loss": 1.1263,
+      "step": 3190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.298890544792185e-06,
+      "loss": 1.1113,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.31544957774466e-06,
+      "loss": 1.0988,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.332008610697136e-06,
+      "loss": 1.1267,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.348567643649611e-06,
+      "loss": 1.118,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.365126676602087e-06,
+      "loss": 1.1468,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.381685709554563e-06,
+      "loss": 1.1379,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.398244742507038e-06,
+      "loss": 1.0794,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.414803775459514e-06,
+      "loss": 1.1028,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.431362808411989e-06,
+      "loss": 1.0537,
+      "step": 3280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4479218413644655e-06,
+      "loss": 1.0812,
+      "step": 3290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.464480874316941e-06,
+      "loss": 1.1172,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.4810399072694155e-06,
+      "loss": 1.0838,
+      "step": 3310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.497598940221891e-06,
+      "loss": 1.0837,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.514157973174366e-06,
+      "loss": 1.1244,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.530717006126843e-06,
+      "loss": 1.1133,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.547276039079318e-06,
+      "loss": 1.0968,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.5638350720317935e-06,
+      "loss": 1.0954,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.580394104984269e-06,
+      "loss": 1.0744,
+      "step": 3370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.596953137936744e-06,
+      "loss": 1.082,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.613512170889221e-06,
+      "loss": 1.0591,
+      "step": 3390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.630071203841696e-06,
+      "loss": 1.0604,
+      "step": 3400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.646630236794172e-06,
+      "loss": 1.0427,
+      "step": 3410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.663189269746647e-06,
+      "loss": 1.0584,
+      "step": 3420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.679748302699123e-06,
+      "loss": 1.0528,
+      "step": 3430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.696307335651599e-06,
+      "loss": 1.0578,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.712866368604074e-06,
+      "loss": 1.0571,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.72942540155655e-06,
+      "loss": 1.055,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.745984434509025e-06,
+      "loss": 1.0195,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.762543467461501e-06,
+      "loss": 1.0358,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.779102500413977e-06,
+      "loss": 1.0599,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.795661533366452e-06,
+      "loss": 1.0098,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.812220566318928e-06,
+      "loss": 1.0381,
+      "step": 3510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.828779599271404e-06,
+      "loss": 1.0187,
+      "step": 3520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.845338632223878e-06,
+      "loss": 1.0731,
+      "step": 3530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.861897665176354e-06,
+      "loss": 1.0089,
+      "step": 3540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.8784566981288294e-06,
+      "loss": 1.0614,
+      "step": 3550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.895015731081305e-06,
+      "loss": 1.0106,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91157476403378e-06,
+      "loss": 1.0282,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.928133796986257e-06,
+      "loss": 1.0272,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.944692829938732e-06,
+      "loss": 1.0104,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9612518628912075e-06,
+      "loss": 1.0085,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.977810895843683e-06,
+      "loss": 1.021,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.994369928796158e-06,
+      "loss": 1.001,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.010928961748635e-06,
+      "loss": 1.0097,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.02748799470111e-06,
+      "loss": 0.9971,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.0440470276535855e-06,
+      "loss": 1.0118,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.0519,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.077165093558537e-06,
+      "loss": 1.0166,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 6.093724126511013e-06,
+      "loss": 1.0027,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.110283159463488e-06,
+      "loss": 1.0049,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.126842192415964e-06,
+      "loss": 1.0118,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.143401225368439e-06,
+      "loss": 1.0102,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.159960258320915e-06,
+      "loss": 0.9987,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.176519291273391e-06,
+      "loss": 1.0121,
+      "step": 3730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.193078324225865e-06,
+      "loss": 1.0163,
+      "step": 3740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.209637357178341e-06,
+      "loss": 0.9831,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.226196390130816e-06,
+      "loss": 1.0143,
+      "step": 3760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.242755423083292e-06,
+      "loss": 0.9793,
+      "step": 3770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.259314456035768e-06,
+      "loss": 0.996,
+      "step": 3780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.275873488988243e-06,
+      "loss": 0.9769,
+      "step": 3790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 6.292432521940719e-06,
+      "loss": 0.9822,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.308991554893194e-06,
+      "loss": 1.0093,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.3255505878456706e-06,
+      "loss": 0.9681,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.342109620798146e-06,
+      "loss": 0.9801,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.3586686537506214e-06,
+      "loss": 0.976,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.375227686703097e-06,
+      "loss": 0.9723,
+      "step": 3850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.391786719655573e-06,
+      "loss": 0.9894,
+      "step": 3860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.408345752608049e-06,
+      "loss": 0.9974,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.424904785560524e-06,
+      "loss": 0.964,
+      "step": 3880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.4414638185129995e-06,
+      "loss": 0.9913,
+      "step": 3890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.458022851465475e-06,
+      "loss": 0.9732,
+      "step": 3900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.474581884417951e-06,
+      "loss": 0.9555,
+      "step": 3910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 6.491140917370427e-06,
+      "loss": 0.9302,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.507699950322902e-06,
+      "loss": 0.9583,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.5242589832753775e-06,
+      "loss": 0.9462,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.540818016227852e-06,
+      "loss": 0.9529,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.5573770491803276e-06,
+      "loss": 0.9703,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.573936082132804e-06,
+      "loss": 0.9635,
+      "step": 3970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.590495115085279e-06,
+      "loss": 0.9523,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.607054148037755e-06,
+      "loss": 0.9875,
+      "step": 3990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.62361318099023e-06,
+      "loss": 0.9385,
+      "step": 4000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.6401722139427065e-06,
+      "loss": 0.9688,
+      "step": 4010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.656731246895182e-06,
+      "loss": 0.969,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.673290279847657e-06,
+      "loss": 0.9651,
+      "step": 4030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 6.689849312800133e-06,
+      "loss": 0.926,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.706408345752608e-06,
+      "loss": 0.949,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.7229673787050845e-06,
+      "loss": 0.9236,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.73952641165756e-06,
+      "loss": 0.972,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.756085444610035e-06,
+      "loss": 0.9395,
+      "step": 4080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.772644477562511e-06,
+      "loss": 0.941,
+      "step": 4090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.789203510514987e-06,
+      "loss": 0.9344,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.8057625434674625e-06,
+      "loss": 0.9492,
+      "step": 4110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.822321576419938e-06,
+      "loss": 0.9413,
+      "step": 4120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.8388806093724134e-06,
+      "loss": 0.9335,
+      "step": 4130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.855439642324889e-06,
+      "loss": 0.9448,
+      "step": 4140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.871998675277365e-06,
+      "loss": 0.9299,
+      "step": 4150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 6.888557708229841e-06,
+      "loss": 0.9448,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.905116741182315e-06,
+      "loss": 0.9429,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.921675774134791e-06,
+      "loss": 0.9064,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.938234807087266e-06,
+      "loss": 0.9484,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.9547938400397415e-06,
+      "loss": 0.9343,
+      "step": 4200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.971352872992218e-06,
+      "loss": 0.9438,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 6.987911905944693e-06,
+      "loss": 0.9309,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.004470938897169e-06,
+      "loss": 0.9316,
+      "step": 4230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.021029971849644e-06,
+      "loss": 0.9348,
+      "step": 4240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.03758900480212e-06,
+      "loss": 0.9274,
+      "step": 4250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.054148037754596e-06,
+      "loss": 0.9285,
+      "step": 4260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.070707070707071e-06,
+      "loss": 0.93,
+      "step": 4270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.087266103659547e-06,
+      "loss": 0.9462,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.103825136612022e-06,
+      "loss": 0.9255,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1203841695644984e-06,
+      "loss": 0.9192,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.136943202516974e-06,
+      "loss": 0.9366,
+      "step": 4310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.153502235469449e-06,
+      "loss": 0.9061,
+      "step": 4320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.170061268421925e-06,
+      "loss": 0.9285,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.186620301374401e-06,
+      "loss": 0.8876,
+      "step": 4340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2031793343268765e-06,
+      "loss": 0.9178,
+      "step": 4350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.219738367279352e-06,
+      "loss": 0.9101,
+      "step": 4360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.236297400231827e-06,
+      "loss": 0.8923,
+      "step": 4370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.252856433184302e-06,
+      "loss": 0.8974,
+      "step": 4380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.269415466136777e-06,
+      "loss": 0.9264,
+      "step": 4390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.285974499089254e-06,
+      "loss": 0.8971,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302533532041729e-06,
+      "loss": 0.9137,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3190925649942046e-06,
+      "loss": 0.8909,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33565159794668e-06,
+      "loss": 0.9044,
+      "step": 4430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.352210630899156e-06,
+      "loss": 0.9125,
+      "step": 4440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.368769663851632e-06,
+      "loss": 0.8978,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.385328696804107e-06,
+      "loss": 0.9047,
+      "step": 4460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.401887729756583e-06,
+      "loss": 0.8895,
+      "step": 4470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.418446762709058e-06,
+      "loss": 0.9048,
+      "step": 4480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.435005795661534e-06,
+      "loss": 0.892,
+      "step": 4490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.45156482861401e-06,
+      "loss": 0.9188,
+      "step": 4500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.468123861566485e-06,
+      "loss": 0.9063,
+      "step": 4510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.484682894518961e-06,
+      "loss": 0.9058,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.501241927471437e-06,
+      "loss": 0.8871,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.517800960423912e-06,
+      "loss": 0.8848,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.534359993376388e-06,
+      "loss": 0.897,
+      "step": 4550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.550919026328863e-06,
+      "loss": 0.874,
+      "step": 4560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.567478059281339e-06,
+      "loss": 0.8843,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.584037092233815e-06,
+      "loss": 0.8942,
+      "step": 4580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.60059612518629e-06,
+      "loss": 0.8711,
+      "step": 4590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.617155158138765e-06,
+      "loss": 0.8907,
+      "step": 4600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.63371419109124e-06,
+      "loss": 0.8909,
+      "step": 4610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.650273224043716e-06,
+      "loss": 0.8909,
+      "step": 4620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.666832256996191e-06,
+      "loss": 0.8926,
+      "step": 4630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.683391289948667e-06,
+      "loss": 0.8662,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.699950322901142e-06,
+      "loss": 0.8906,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.71650935585362e-06,
+      "loss": 0.8656,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.733068388806095e-06,
+      "loss": 0.8792,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.74962742175857e-06,
+      "loss": 0.8893,
+      "step": 4680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.766186454711046e-06,
+      "loss": 0.8799,
+      "step": 4690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.782745487663521e-06,
+      "loss": 0.9227,
+      "step": 4700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.799304520615997e-06,
+      "loss": 0.8472,
+      "step": 4710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.815863553568472e-06,
+      "loss": 0.8502,
+      "step": 4720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.832422586520947e-06,
+      "loss": 0.8596,
+      "step": 4730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.848981619473423e-06,
+      "loss": 0.8757,
+      "step": 4740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.8655406524259e-06,
+      "loss": 0.896,
+      "step": 4750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.882099685378375e-06,
+      "loss": 0.8661,
+      "step": 4760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.898658718330851e-06,
+      "loss": 0.8426,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.915217751283326e-06,
+      "loss": 0.8908,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.931776784235802e-06,
+      "loss": 0.8549,
+      "step": 4790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.948335817188277e-06,
+      "loss": 0.8743,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.964894850140753e-06,
+      "loss": 0.8626,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.981453883093228e-06,
+      "loss": 0.8673,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.998012916045704e-06,
+      "loss": 0.8593,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.014571948998179e-06,
+      "loss": 0.8624,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.031130981950654e-06,
+      "loss": 0.8652,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.04769001490313e-06,
+      "loss": 0.8663,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.064249047855605e-06,
+      "loss": 0.8848,
+      "step": 4870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.08080808080808e-06,
+      "loss": 0.8322,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.097367113760556e-06,
+      "loss": 0.8553,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.113926146713033e-06,
+      "loss": 0.8816,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.130485179665509e-06,
+      "loss": 0.8685,
+      "step": 4910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.147044212617984e-06,
+      "loss": 0.8464,
+      "step": 4920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.16360324557046e-06,
+      "loss": 0.8501,
+      "step": 4930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.180162278522935e-06,
+      "loss": 0.8498,
+      "step": 4940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.19672131147541e-06,
+      "loss": 0.8549,
+      "step": 4950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.213280344427886e-06,
+      "loss": 0.8466,
+      "step": 4960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.229839377380361e-06,
+      "loss": 0.844,
+      "step": 4970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.246398410332837e-06,
+      "loss": 0.8767,
+      "step": 4980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.262957443285314e-06,
+      "loss": 0.8473,
+      "step": 4990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.27951647623779e-06,
+      "loss": 0.8575,
+      "step": 5000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.296075509190265e-06,
+      "loss": 0.8178,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.312634542142739e-06,
+      "loss": 0.8311,
+      "step": 5020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.329193575095214e-06,
+      "loss": 0.8519,
+      "step": 5030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.34575260804769e-06,
+      "loss": 0.8827,
+      "step": 5040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.362311641000167e-06,
+      "loss": 0.8403,
+      "step": 5050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.378870673952642e-06,
+      "loss": 0.8458,
+      "step": 5060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.395429706905117e-06,
+      "loss": 0.8386,
+      "step": 5070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.411988739857593e-06,
+      "loss": 0.8652,
+      "step": 5080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.428547772810068e-06,
+      "loss": 0.8101,
+      "step": 5090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.445106805762544e-06,
+      "loss": 0.827,
+      "step": 5100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.46166583871502e-06,
+      "loss": 0.8309,
+      "step": 5110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.478224871667495e-06,
+      "loss": 0.8642,
+      "step": 5120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.49478390461997e-06,
+      "loss": 0.8426,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.511342937572447e-06,
+      "loss": 0.8326,
+      "step": 5140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.527901970524923e-06,
+      "loss": 0.8413,
+      "step": 5150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.544461003477398e-06,
+      "loss": 0.8172,
+      "step": 5160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.561020036429874e-06,
+      "loss": 0.8408,
+      "step": 5170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.577579069382349e-06,
+      "loss": 0.8358,
+      "step": 5180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.594138102334824e-06,
+      "loss": 0.8556,
+      "step": 5190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.6106971352873e-06,
+      "loss": 0.8369,
+      "step": 5200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.627256168239775e-06,
+      "loss": 0.8438,
+      "step": 5210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.64381520119225e-06,
+      "loss": 0.8498,
+      "step": 5220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.660374234144726e-06,
+      "loss": 0.8276,
+      "step": 5230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.676933267097202e-06,
+      "loss": 0.8264,
+      "step": 5240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.693492300049677e-06,
+      "loss": 0.8554,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.710051333002153e-06,
+      "loss": 0.8386,
+      "step": 5260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.726610365954628e-06,
+      "loss": 0.8509,
+      "step": 5270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.743169398907103e-06,
+      "loss": 0.8333,
+      "step": 5280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.75972843185958e-06,
+      "loss": 0.8222,
+      "step": 5290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.776287464812056e-06,
+      "loss": 0.8021,
+      "step": 5300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.792846497764531e-06,
+      "loss": 0.8355,
+      "step": 5310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.809405530717007e-06,
+      "loss": 0.8254,
+      "step": 5320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.825964563669482e-06,
+      "loss": 0.8366,
+      "step": 5330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.842523596621958e-06,
+      "loss": 0.7972,
+      "step": 5340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.859082629574433e-06,
+      "loss": 0.8485,
+      "step": 5350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.875641662526909e-06,
+      "loss": 0.8302,
+      "step": 5360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.892200695479386e-06,
+      "loss": 0.8096,
+      "step": 5370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.908759728431861e-06,
+      "loss": 0.8105,
+      "step": 5380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.925318761384337e-06,
+      "loss": 0.8076,
+      "step": 5390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.941877794336812e-06,
+      "loss": 0.8301,
+      "step": 5400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.958436827289288e-06,
+      "loss": 0.8283,
+      "step": 5410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.974995860241763e-06,
+      "loss": 0.8065,
+      "step": 5420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.991554893194238e-06,
+      "loss": 0.8066,
+      "step": 5430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.008113926146714e-06,
+      "loss": 0.8032,
+      "step": 5440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.02467295909919e-06,
+      "loss": 0.8203,
+      "step": 5450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.041231992051665e-06,
+      "loss": 0.8144,
+      "step": 5460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.05779102500414e-06,
+      "loss": 0.8232,
+      "step": 5470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.074350057956616e-06,
+      "loss": 0.8131,
+      "step": 5480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.8015,
+      "step": 5490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.107468123861566e-06,
+      "loss": 0.8121,
+      "step": 5500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.124027156814042e-06,
+      "loss": 0.8296,
+      "step": 5510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.140586189766519e-06,
+      "loss": 0.8069,
+      "step": 5520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.157145222718994e-06,
+      "loss": 0.8356,
+      "step": 5530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.17370425567147e-06,
+      "loss": 0.8147,
+      "step": 5540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.190263288623945e-06,
+      "loss": 0.833,
+      "step": 5550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.20682232157642e-06,
+      "loss": 0.7915,
+      "step": 5560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.223381354528896e-06,
+      "loss": 0.8287,
+      "step": 5570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.239940387481372e-06,
+      "loss": 0.8181,
+      "step": 5580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.256499420433847e-06,
+      "loss": 0.8051,
+      "step": 5590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.273058453386323e-06,
+      "loss": 0.8149,
+      "step": 5600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.2896174863388e-06,
+      "loss": 0.7904,
+      "step": 5610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.306176519291275e-06,
+      "loss": 0.8078,
+      "step": 5620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.32273555224375e-06,
+      "loss": 0.8066,
+      "step": 5630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.339294585196226e-06,
+      "loss": 0.7877,
+      "step": 5640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.355853618148701e-06,
+      "loss": 0.797,
+      "step": 5650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.372412651101175e-06,
+      "loss": 0.7991,
+      "step": 5660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.388971684053652e-06,
+      "loss": 0.8052,
+      "step": 5670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.405530717006128e-06,
+      "loss": 0.7839,
+      "step": 5680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.422089749958603e-06,
+      "loss": 0.7638,
+      "step": 5690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.438648782911079e-06,
+      "loss": 0.8076,
+      "step": 5700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.455207815863554e-06,
+      "loss": 0.7958,
+      "step": 5710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.47176684881603e-06,
+      "loss": 0.8012,
+      "step": 5720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.488325881768505e-06,
+      "loss": 0.7708,
+      "step": 5730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.50488491472098e-06,
+      "loss": 0.7921,
+      "step": 5740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.521443947673456e-06,
+      "loss": 0.8043,
+      "step": 5750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.538002980625933e-06,
+      "loss": 0.7744,
+      "step": 5760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.554562013578408e-06,
+      "loss": 0.7982,
+      "step": 5770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.571121046530884e-06,
+      "loss": 0.8058,
+      "step": 5780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.58768007948336e-06,
+      "loss": 0.793,
+      "step": 5790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.604239112435835e-06,
+      "loss": 0.8107,
+      "step": 5800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.62079814538831e-06,
+      "loss": 0.7907,
+      "step": 5810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.637357178340786e-06,
+      "loss": 0.818,
+      "step": 5820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.653916211293261e-06,
+      "loss": 0.7918,
+      "step": 5830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.670475244245737e-06,
+      "loss": 0.7815,
+      "step": 5840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.687034277198214e-06,
+      "loss": 0.8019,
+      "step": 5850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.703593310150689e-06,
+      "loss": 0.8088,
+      "step": 5860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.720152343103163e-06,
+      "loss": 0.7652,
+      "step": 5870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.736711376055638e-06,
+      "loss": 0.7754,
+      "step": 5880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.753270409008114e-06,
+      "loss": 0.7706,
+      "step": 5890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.76982944196059e-06,
+      "loss": 0.7679,
+      "step": 5900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.786388474913066e-06,
+      "loss": 0.7796,
+      "step": 5910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.802947507865542e-06,
+      "loss": 0.7877,
+      "step": 5920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.819506540818017e-06,
+      "loss": 0.7773,
+      "step": 5930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.836065573770493e-06,
+      "loss": 0.809,
+      "step": 5940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.852624606722968e-06,
+      "loss": 0.7951,
+      "step": 5950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.869183639675443e-06,
+      "loss": 0.7736,
+      "step": 5960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.885742672627919e-06,
+      "loss": 0.7919,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.902301705580394e-06,
+      "loss": 0.7981,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.91886073853287e-06,
+      "loss": 0.7748,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.935419771485347e-06,
+      "loss": 0.779,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.951978804437822e-06,
+      "loss": 0.7923,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.968537837390298e-06,
+      "loss": 0.791,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.985096870342773e-06,
+      "loss": 0.7799,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.998343822457768e-06,
+      "loss": 0.7615,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.981782047035444e-06,
+      "loss": 0.765,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.965220271613117e-06,
+      "loss": 0.7687,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.948658496190793e-06,
+      "loss": 0.7758,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.932096720768467e-06,
+      "loss": 0.7813,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.915534945346142e-06,
+      "loss": 0.7745,
+      "step": 6090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898973169923817e-06,
+      "loss": 0.7687,
+      "step": 6100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.882411394501491e-06,
+      "loss": 0.7762,
+      "step": 6110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.865849619079166e-06,
+      "loss": 0.7736,
+      "step": 6120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.84928784365684e-06,
+      "loss": 0.7646,
+      "step": 6130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.832726068234515e-06,
+      "loss": 0.7549,
+      "step": 6140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.816164292812191e-06,
+      "loss": 0.7736,
+      "step": 6150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.799602517389865e-06,
+      "loss": 0.7321,
+      "step": 6160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.78304074196754e-06,
+      "loss": 0.7794,
+      "step": 6170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.766478966545214e-06,
+      "loss": 0.7847,
+      "step": 6180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.749917191122889e-06,
+      "loss": 0.7732,
+      "step": 6190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.733355415700564e-06,
+      "loss": 0.778,
+      "step": 6200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.716793640278238e-06,
+      "loss": 0.7685,
+      "step": 6210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700231864855914e-06,
+      "loss": 0.7647,
+      "step": 6220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.683670089433587e-06,
+      "loss": 0.7607,
+      "step": 6230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.667108314011263e-06,
+      "loss": 0.7546,
+      "step": 6240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.650546538588938e-06,
+      "loss": 0.7532,
+      "step": 6250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.633984763166613e-06,
+      "loss": 0.766,
+      "step": 6260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.617422987744287e-06,
+      "loss": 0.75,
+      "step": 6270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.60086121232196e-06,
+      "loss": 0.7782,
+      "step": 6280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.584299436899636e-06,
+      "loss": 0.7481,
+      "step": 6290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.567737661477312e-06,
+      "loss": 0.7572,
+      "step": 6300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.551175886054987e-06,
+      "loss": 0.7561,
+      "step": 6310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.53461411063266e-06,
+      "loss": 0.759,
+      "step": 6320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.518052335210334e-06,
+      "loss": 0.7637,
+      "step": 6330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.50149055978801e-06,
+      "loss": 0.7511,
+      "step": 6340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.484928784365685e-06,
+      "loss": 0.758,
+      "step": 6350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46836700894336e-06,
+      "loss": 0.747,
+      "step": 6360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451805233521034e-06,
+      "loss": 0.7767,
+      "step": 6370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.435243458098708e-06,
+      "loss": 0.7354,
+      "step": 6380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.418681682676383e-06,
+      "loss": 0.7691,
+      "step": 6390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402119907254059e-06,
+      "loss": 0.74,
+      "step": 6400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.385558131831734e-06,
+      "loss": 0.7684,
+      "step": 6410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368996356409408e-06,
+      "loss": 0.7491,
+      "step": 6420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.352434580987083e-06,
+      "loss": 0.7664,
+      "step": 6430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.335872805564757e-06,
+      "loss": 0.764,
+      "step": 6440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.319311030142432e-06,
+      "loss": 0.74,
+      "step": 6450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.302749254720108e-06,
+      "loss": 0.7579,
+      "step": 6460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286187479297781e-06,
+      "loss": 0.7676,
+      "step": 6470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.269625703875457e-06,
+      "loss": 0.7652,
+      "step": 6480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25306392845313e-06,
+      "loss": 0.7515,
+      "step": 6490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236502153030806e-06,
+      "loss": 0.7549,
+      "step": 6500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219940377608481e-06,
+      "loss": 0.7603,
+      "step": 6510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203378602186155e-06,
+      "loss": 0.7805,
+      "step": 6520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18681682676383e-06,
+      "loss": 0.7571,
+      "step": 6530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170255051341504e-06,
+      "loss": 0.7364,
+      "step": 6540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15369327591918e-06,
+      "loss": 0.7657,
+      "step": 6550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137131500496855e-06,
+      "loss": 0.7527,
+      "step": 6560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120569725074528e-06,
+      "loss": 0.7746,
+      "step": 6570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104007949652204e-06,
+      "loss": 0.734,
+      "step": 6580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.087446174229878e-06,
+      "loss": 0.7373,
+      "step": 6590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.070884398807553e-06,
+      "loss": 0.7354,
+      "step": 6600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.054322623385228e-06,
+      "loss": 0.7409,
+      "step": 6610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.037760847962902e-06,
+      "loss": 0.733,
+      "step": 6620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.021199072540577e-06,
+      "loss": 0.7218,
+      "step": 6630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004637297118251e-06,
+      "loss": 0.7816,
+      "step": 6640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988075521695927e-06,
+      "loss": 0.7423,
+      "step": 6650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971513746273602e-06,
+      "loss": 0.7444,
+      "step": 6660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954951970851276e-06,
+      "loss": 0.7266,
+      "step": 6670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.938390195428951e-06,
+      "loss": 0.7514,
+      "step": 6680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921828420006625e-06,
+      "loss": 0.767,
+      "step": 6690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9052666445843e-06,
+      "loss": 0.7357,
+      "step": 6700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.888704869161975e-06,
+      "loss": 0.7237,
+      "step": 6710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.87214309373965e-06,
+      "loss": 0.7514,
+      "step": 6720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.855581318317325e-06,
+      "loss": 0.7392,
+      "step": 6730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.839019542894998e-06,
+      "loss": 0.7369,
+      "step": 6740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.822457767472674e-06,
+      "loss": 0.7441,
+      "step": 6750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.805895992050349e-06,
+      "loss": 0.7343,
+      "step": 6760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.789334216628024e-06,
+      "loss": 0.7368,
+      "step": 6770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.772772441205698e-06,
+      "loss": 0.7386,
+      "step": 6780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.756210665783372e-06,
+      "loss": 0.7177,
+      "step": 6790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.739648890361047e-06,
+      "loss": 0.7355,
+      "step": 6800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.723087114938723e-06,
+      "loss": 0.7536,
+      "step": 6810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706525339516398e-06,
+      "loss": 0.737,
+      "step": 6820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.689963564094072e-06,
+      "loss": 0.7201,
+      "step": 6830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.673401788671745e-06,
+      "loss": 0.7337,
+      "step": 6840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.65684001324942e-06,
+      "loss": 0.7494,
+      "step": 6850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.640278237827096e-06,
+      "loss": 0.7176,
+      "step": 6860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.623716462404772e-06,
+      "loss": 0.7382,
+      "step": 6870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.607154686982445e-06,
+      "loss": 0.7303,
+      "step": 6880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.590592911560119e-06,
+      "loss": 0.7441,
+      "step": 6890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.574031136137794e-06,
+      "loss": 0.7629,
+      "step": 6900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.55746936071547e-06,
+      "loss": 0.7316,
+      "step": 6910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.540907585293145e-06,
+      "loss": 0.7366,
+      "step": 6920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.524345809870819e-06,
+      "loss": 0.7291,
+      "step": 6930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.507784034448494e-06,
+      "loss": 0.7286,
+      "step": 6940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.491222259026168e-06,
+      "loss": 0.7488,
+      "step": 6950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.474660483603843e-06,
+      "loss": 0.7304,
+      "step": 6960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.458098708181519e-06,
+      "loss": 0.7369,
+      "step": 6970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.441536932759192e-06,
+      "loss": 0.7378,
+      "step": 6980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.424975157336868e-06,
+      "loss": 0.7161,
+      "step": 6990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.408413381914541e-06,
+      "loss": 0.7381,
+      "step": 7000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.391851606492217e-06,
+      "loss": 0.737,
+      "step": 7010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.375289831069892e-06,
+      "loss": 0.7309,
+      "step": 7020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.358728055647566e-06,
+      "loss": 0.7539,
+      "step": 7030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.342166280225241e-06,
+      "loss": 0.7238,
+      "step": 7040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.325604504802915e-06,
+      "loss": 0.772,
+      "step": 7050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.30904272938059e-06,
+      "loss": 0.7197,
+      "step": 7060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.292480953958266e-06,
+      "loss": 0.6917,
+      "step": 7070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.27591917853594e-06,
+      "loss": 0.745,
+      "step": 7080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.259357403113615e-06,
+      "loss": 0.7299,
+      "step": 7090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.242795627691289e-06,
+      "loss": 0.7047,
+      "step": 7100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.226233852268964e-06,
+      "loss": 0.7161,
+      "step": 7110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.20967207684664e-06,
+      "loss": 0.7424,
+      "step": 7120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.193110301424313e-06,
+      "loss": 0.7169,
+      "step": 7130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.176548526001988e-06,
+      "loss": 0.74,
+      "step": 7140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.159986750579662e-06,
+      "loss": 0.7332,
+      "step": 7150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.143424975157338e-06,
+      "loss": 0.721,
+      "step": 7160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.126863199735013e-06,
+      "loss": 0.7346,
+      "step": 7170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.110301424312687e-06,
+      "loss": 0.7189,
+      "step": 7180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.093739648890362e-06,
+      "loss": 0.729,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.077177873468036e-06,
+      "loss": 0.7321,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.060616098045711e-06,
+      "loss": 0.7362,
+      "step": 7210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.044054322623386e-06,
+      "loss": 0.7429,
+      "step": 7220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.02749254720106e-06,
+      "loss": 0.7278,
+      "step": 7230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.010930771778736e-06,
+      "loss": 0.7124,
+      "step": 7240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.99436899635641e-06,
+      "loss": 0.7154,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.977807220934085e-06,
+      "loss": 0.7358,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.96124544551176e-06,
+      "loss": 0.6947,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.944683670089435e-06,
+      "loss": 0.7242,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.928121894667109e-06,
+      "loss": 0.7357,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.911560119244783e-06,
+      "loss": 0.7199,
+      "step": 7300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.894998343822458e-06,
+      "loss": 0.7427,
+      "step": 7310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.878436568400134e-06,
+      "loss": 0.7354,
+      "step": 7320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.861874792977809e-06,
+      "loss": 0.7171,
+      "step": 7330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.845313017555483e-06,
+      "loss": 0.7214,
+      "step": 7340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.828751242133156e-06,
+      "loss": 0.7303,
+      "step": 7350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.812189466710832e-06,
+      "loss": 0.7201,
+      "step": 7360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.795627691288507e-06,
+      "loss": 0.7059,
+      "step": 7370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.779065915866183e-06,
+      "loss": 0.7127,
+      "step": 7380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.762504140443856e-06,
+      "loss": 0.7232,
+      "step": 7390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.74594236502153e-06,
+      "loss": 0.7253,
+      "step": 7400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.729380589599205e-06,
+      "loss": 0.7191,
+      "step": 7410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.71281881417688e-06,
+      "loss": 0.7256,
+      "step": 7420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.696257038754556e-06,
+      "loss": 0.701,
+      "step": 7430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.67969526333223e-06,
+      "loss": 0.7153,
+      "step": 7440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.663133487909905e-06,
+      "loss": 0.6816,
+      "step": 7450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.646571712487579e-06,
+      "loss": 0.7188,
+      "step": 7460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.630009937065254e-06,
+      "loss": 0.7052,
+      "step": 7470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.613448161642929e-06,
+      "loss": 0.6967,
+      "step": 7480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.596886386220603e-06,
+      "loss": 0.7202,
+      "step": 7490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.580324610798279e-06,
+      "loss": 0.7185,
+      "step": 7500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.563762835375953e-06,
+      "loss": 0.717,
+      "step": 7510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.547201059953627e-06,
+      "loss": 0.7185,
+      "step": 7520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.530639284531302e-06,
+      "loss": 0.722,
+      "step": 7530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.514077509108977e-06,
+      "loss": 0.7197,
+      "step": 7540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.497515733686652e-06,
+      "loss": 0.7212,
+      "step": 7550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.480953958264327e-06,
+      "loss": 0.7164,
+      "step": 7560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.464392182842001e-06,
+      "loss": 0.7232,
+      "step": 7570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.447830407419676e-06,
+      "loss": 0.693,
+      "step": 7580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.4312686319973505e-06,
+      "loss": 0.7195,
+      "step": 7590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.414706856575026e-06,
+      "loss": 0.6996,
+      "step": 7600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3981450811527e-06,
+      "loss": 0.7277,
+      "step": 7610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.381583305730375e-06,
+      "loss": 0.6954,
+      "step": 7620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3650215303080495e-06,
+      "loss": 0.6941,
+      "step": 7630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.348459754885724e-06,
+      "loss": 0.7269,
+      "step": 7640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.3318979794633995e-06,
+      "loss": 0.7036,
+      "step": 7650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.315336204041073e-06,
+      "loss": 0.718,
+      "step": 7660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.2987744286187486e-06,
+      "loss": 0.7109,
+      "step": 7670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.282212653196423e-06,
+      "loss": 0.7113,
+      "step": 7680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.265650877774098e-06,
+      "loss": 0.7204,
+      "step": 7690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.249089102351773e-06,
+      "loss": 0.7243,
+      "step": 7700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.232527326929447e-06,
+      "loss": 0.6937,
+      "step": 7710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.215965551507122e-06,
+      "loss": 0.712,
+      "step": 7720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.199403776084797e-06,
+      "loss": 0.7091,
+      "step": 7730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.182842000662471e-06,
+      "loss": 0.7127,
+      "step": 7740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.166280225240147e-06,
+      "loss": 0.6984,
+      "step": 7750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.14971844981782e-06,
+      "loss": 0.7022,
+      "step": 7760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.133156674395496e-06,
+      "loss": 0.7113,
+      "step": 7770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.11659489897317e-06,
+      "loss": 0.7228,
+      "step": 7780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.100033123550846e-06,
+      "loss": 0.7072,
+      "step": 7790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.08347134812852e-06,
+      "loss": 0.711,
+      "step": 7800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.066909572706194e-06,
+      "loss": 0.7287,
+      "step": 7810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.050347797283869e-06,
+      "loss": 0.7124,
+      "step": 7820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.033786021861544e-06,
+      "loss": 0.7212,
+      "step": 7830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.017224246439219e-06,
+      "loss": 0.707,
+      "step": 7840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.000662471016894e-06,
+      "loss": 0.7172,
+      "step": 7850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.984100695594567e-06,
+      "loss": 0.7385,
+      "step": 7860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.967538920172243e-06,
+      "loss": 0.7272,
+      "step": 7870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.950977144749917e-06,
+      "loss": 0.6972,
+      "step": 7880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.934415369327593e-06,
+      "loss": 0.7251,
+      "step": 7890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.917853593905267e-06,
+      "loss": 0.7199,
+      "step": 7900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.901291818482941e-06,
+      "loss": 0.7161,
+      "step": 7910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.884730043060616e-06,
+      "loss": 0.6993,
+      "step": 7920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.868168267638291e-06,
+      "loss": 0.6977,
+      "step": 7930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.851606492215966e-06,
+      "loss": 0.6985,
+      "step": 7940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.835044716793641e-06,
+      "loss": 0.7077,
+      "step": 7950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.818482941371316e-06,
+      "loss": 0.6873,
+      "step": 7960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.80192116594899e-06,
+      "loss": 0.7166,
+      "step": 7970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7853593905266645e-06,
+      "loss": 0.7008,
+      "step": 7980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.76879761510434e-06,
+      "loss": 0.6944,
+      "step": 7990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.752235839682014e-06,
+      "loss": 0.7062,
+      "step": 8000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.73567406425969e-06,
+      "loss": 0.7045,
+      "step": 8010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7191122888373635e-06,
+      "loss": 0.7091,
+      "step": 8020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.702550513415038e-06,
+      "loss": 0.6981,
+      "step": 8030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6859887379927134e-06,
+      "loss": 0.7218,
+      "step": 8040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.669426962570388e-06,
+      "loss": 0.6965,
+      "step": 8050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.652865187148063e-06,
+      "loss": 0.7097,
+      "step": 8060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.636303411725737e-06,
+      "loss": 0.6902,
+      "step": 8070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6197416363034125e-06,
+      "loss": 0.7025,
+      "step": 8080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.603179860881087e-06,
+      "loss": 0.702,
+      "step": 8090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.5866180854587615e-06,
+      "loss": 0.7285,
+      "step": 8100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.570056310036437e-06,
+      "loss": 0.719,
+      "step": 8110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.553494534614111e-06,
+      "loss": 0.7101,
+      "step": 8120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.536932759191786e-06,
+      "loss": 0.6924,
+      "step": 8130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.5203709837694606e-06,
+      "loss": 0.6861,
+      "step": 8140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.503809208347135e-06,
+      "loss": 0.6941,
+      "step": 8150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.4872474329248105e-06,
+      "loss": 0.6738,
+      "step": 8160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.470685657502484e-06,
+      "loss": 0.672,
+      "step": 8170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.45412388208016e-06,
+      "loss": 0.7043,
+      "step": 8180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.437562106657834e-06,
+      "loss": 0.7002,
+      "step": 8190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.421000331235509e-06,
+      "loss": 0.716,
+      "step": 8200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.404438555813184e-06,
+      "loss": 0.6976,
+      "step": 8210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.387876780390858e-06,
+      "loss": 0.7162,
+      "step": 8220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.371315004968533e-06,
+      "loss": 0.6882,
+      "step": 8230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.354753229546208e-06,
+      "loss": 0.6995,
+      "step": 8240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.338191454123883e-06,
+      "loss": 0.6934,
+      "step": 8250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.321629678701558e-06,
+      "loss": 0.6956,
+      "step": 8260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.305067903279231e-06,
+      "loss": 0.7101,
+      "step": 8270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.288506127856907e-06,
+      "loss": 0.7014,
+      "step": 8280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.271944352434581e-06,
+      "loss": 0.6877,
+      "step": 8290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.255382577012257e-06,
+      "loss": 0.6999,
+      "step": 8300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.238820801589931e-06,
+      "loss": 0.7118,
+      "step": 8310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.222259026167605e-06,
+      "loss": 0.7084,
+      "step": 8320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.20569725074528e-06,
+      "loss": 0.6969,
+      "step": 8330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.189135475322955e-06,
+      "loss": 0.7048,
+      "step": 8340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.17257369990063e-06,
+      "loss": 0.6875,
+      "step": 8350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.156011924478305e-06,
+      "loss": 0.6991,
+      "step": 8360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1394501490559784e-06,
+      "loss": 0.692,
+      "step": 8370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.122888373633654e-06,
+      "loss": 0.6997,
+      "step": 8380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.106326598211328e-06,
+      "loss": 0.6819,
+      "step": 8390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.089764822789004e-06,
+      "loss": 0.6732,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.073203047366678e-06,
+      "loss": 0.6852,
+      "step": 8410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.056641271944354e-06,
+      "loss": 0.7219,
+      "step": 8420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.040079496522027e-06,
+      "loss": 0.711,
+      "step": 8430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.023517721099702e-06,
+      "loss": 0.6958,
+      "step": 8440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.006955945677377e-06,
+      "loss": 0.6748,
+      "step": 8450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.990394170255052e-06,
+      "loss": 0.6976,
+      "step": 8460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.973832394832727e-06,
+      "loss": 0.7087,
+      "step": 8470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.957270619410401e-06,
+      "loss": 0.7002,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9407088439880755e-06,
+      "loss": 0.7171,
+      "step": 8490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.924147068565751e-06,
+      "loss": 0.6987,
+      "step": 8500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9075852931434254e-06,
+      "loss": 0.7037,
+      "step": 8510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.891023517721101e-06,
+      "loss": 0.6969,
+      "step": 8520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.8744617422987745e-06,
+      "loss": 0.6996,
+      "step": 8530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.857899966876449e-06,
+      "loss": 0.6903,
+      "step": 8540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.8413381914541245e-06,
+      "loss": 0.6639,
+      "step": 8550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.824776416031799e-06,
+      "loss": 0.6978,
+      "step": 8560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.808214640609474e-06,
+      "loss": 0.7142,
+      "step": 8570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.791652865187148e-06,
+      "loss": 0.6976,
+      "step": 8580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.7750910897648235e-06,
+      "loss": 0.715,
+      "step": 8590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.758529314342498e-06,
+      "loss": 0.7161,
+      "step": 8600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.7419675389201726e-06,
+      "loss": 0.6939,
+      "step": 8610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.725405763497848e-06,
+      "loss": 0.6906,
+      "step": 8620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.708843988075522e-06,
+      "loss": 0.7068,
+      "step": 8630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.692282212653197e-06,
+      "loss": 0.6941,
+      "step": 8640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.675720437230872e-06,
+      "loss": 0.6829,
+      "step": 8650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.659158661808546e-06,
+      "loss": 0.6916,
+      "step": 8660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.6425968863862215e-06,
+      "loss": 0.6966,
+      "step": 8670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.626035110963895e-06,
+      "loss": 0.6911,
+      "step": 8680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.609473335541571e-06,
+      "loss": 0.6893,
+      "step": 8690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.592911560119245e-06,
+      "loss": 0.7303,
+      "step": 8700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.57634978469692e-06,
+      "loss": 0.6844,
+      "step": 8710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.559788009274595e-06,
+      "loss": 0.7024,
+      "step": 8720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.543226233852269e-06,
+      "loss": 0.6972,
+      "step": 8730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.526664458429944e-06,
+      "loss": 0.7005,
+      "step": 8740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.510102683007619e-06,
+      "loss": 0.7068,
+      "step": 8750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.493540907585294e-06,
+      "loss": 0.6922,
+      "step": 8760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.476979132162969e-06,
+      "loss": 0.693,
+      "step": 8770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.460417356740642e-06,
+      "loss": 0.6939,
+      "step": 8780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.443855581318318e-06,
+      "loss": 0.6825,
+      "step": 8790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.427293805895992e-06,
+      "loss": 0.695,
+      "step": 8800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.410732030473668e-06,
+      "loss": 0.6852,
+      "step": 8810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.394170255051342e-06,
+      "loss": 0.7059,
+      "step": 8820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.377608479629016e-06,
+      "loss": 0.6859,
+      "step": 8830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.361046704206691e-06,
+      "loss": 0.6971,
+      "step": 8840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.344484928784366e-06,
+      "loss": 0.7077,
+      "step": 8850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.327923153362041e-06,
+      "loss": 0.6679,
+      "step": 8860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.311361377939716e-06,
+      "loss": 0.673,
+      "step": 8870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2947996025173895e-06,
+      "loss": 0.7001,
+      "step": 8880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.278237827095065e-06,
+      "loss": 0.6848,
+      "step": 8890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.261676051672739e-06,
+      "loss": 0.6831,
+      "step": 8900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.245114276250415e-06,
+      "loss": 0.6886,
+      "step": 8910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.228552500828089e-06,
+      "loss": 0.7001,
+      "step": 8920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.211990725405765e-06,
+      "loss": 0.7103,
+      "step": 8930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1954289499834384e-06,
+      "loss": 0.6866,
+      "step": 8940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.178867174561113e-06,
+      "loss": 0.6891,
+      "step": 8950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.162305399138788e-06,
+      "loss": 0.6854,
+      "step": 8960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.145743623716463e-06,
+      "loss": 0.6756,
+      "step": 8970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.129181848294138e-06,
+      "loss": 0.6809,
+      "step": 8980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.112620072871812e-06,
+      "loss": 0.691,
+      "step": 8990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0960582974494865e-06,
+      "loss": 0.6843,
+      "step": 9000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.079496522027162e-06,
+      "loss": 0.7142,
+      "step": 9010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0629347466048365e-06,
+      "loss": 0.6793,
+      "step": 9020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.046372971182512e-06,
+      "loss": 0.6803,
+      "step": 9030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0298111957601856e-06,
+      "loss": 0.6687,
+      "step": 9040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.01324942033786e-06,
+      "loss": 0.7009,
+      "step": 9050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9966876449155355e-06,
+      "loss": 0.7167,
+      "step": 9060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.98012586949321e-06,
+      "loss": 0.6812,
+      "step": 9070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9635640940708846e-06,
+      "loss": 0.7035,
+      "step": 9080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.947002318648559e-06,
+      "loss": 0.6729,
+      "step": 9090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9304405432262345e-06,
+      "loss": 0.6778,
+      "step": 9100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.913878767803909e-06,
+      "loss": 0.7033,
+      "step": 9110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.897316992381584e-06,
+      "loss": 0.6915,
+      "step": 9120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.880755216959259e-06,
+      "loss": 0.6843,
+      "step": 9130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.864193441536933e-06,
+      "loss": 0.6891,
+      "step": 9140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.847631666114608e-06,
+      "loss": 0.6869,
+      "step": 9150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.831069890692283e-06,
+      "loss": 0.6755,
+      "step": 9160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.814508115269957e-06,
+      "loss": 0.7038,
+      "step": 9170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7979463398476325e-06,
+      "loss": 0.677,
+      "step": 9180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.781384564425306e-06,
+      "loss": 0.6954,
+      "step": 9190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.764822789002982e-06,
+      "loss": 0.6997,
+      "step": 9200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.748261013580656e-06,
+      "loss": 0.6874,
+      "step": 9210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.731699238158331e-06,
+      "loss": 0.7152,
+      "step": 9220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.715137462736006e-06,
+      "loss": 0.6918,
+      "step": 9230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.698575687313681e-06,
+      "loss": 0.6889,
+      "step": 9240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.682013911891355e-06,
+      "loss": 0.6707,
+      "step": 9250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.66545213646903e-06,
+      "loss": 0.6801,
+      "step": 9260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.648890361046704e-06,
+      "loss": 0.6742,
+      "step": 9270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.63232858562438e-06,
+      "loss": 0.663,
+      "step": 9280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.615766810202054e-06,
+      "loss": 0.699,
+      "step": 9290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.599205034779729e-06,
+      "loss": 0.6819,
+      "step": 9300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.582643259357403e-06,
+      "loss": 0.6865,
+      "step": 9310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.566081483935078e-06,
+      "loss": 0.6819,
+      "step": 9320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.549519708512753e-06,
+      "loss": 0.6966,
+      "step": 9330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.532957933090428e-06,
+      "loss": 0.7008,
+      "step": 9340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.516396157668102e-06,
+      "loss": 0.682,
+      "step": 9350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.499834382245777e-06,
+      "loss": 0.6832,
+      "step": 9360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.483272606823451e-06,
+      "loss": 0.6936,
+      "step": 9370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.466710831401127e-06,
+      "loss": 0.6745,
+      "step": 9380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.450149055978801e-06,
+      "loss": 0.6812,
+      "step": 9390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.433587280556476e-06,
+      "loss": 0.6681,
+      "step": 9400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.417025505134151e-06,
+      "loss": 0.6757,
+      "step": 9410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.400463729711825e-06,
+      "loss": 0.6762,
+      "step": 9420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3839019542895e-06,
+      "loss": 0.667,
+      "step": 9430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.367340178867175e-06,
+      "loss": 0.6863,
+      "step": 9440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3507784034448495e-06,
+      "loss": 0.7007,
+      "step": 9450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.334216628022525e-06,
+      "loss": 0.6846,
+      "step": 9460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.317654852600199e-06,
+      "loss": 0.6705,
+      "step": 9470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.301093077177874e-06,
+      "loss": 0.6711,
+      "step": 9480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2845313017555485e-06,
+      "loss": 0.6903,
+      "step": 9490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267969526333223e-06,
+      "loss": 0.6952,
+      "step": 9500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.251407750910898e-06,
+      "loss": 0.6843,
+      "step": 9510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.234845975488573e-06,
+      "loss": 0.6805,
+      "step": 9520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2182842000662475e-06,
+      "loss": 0.6827,
+      "step": 9530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.201722424643922e-06,
+      "loss": 0.6874,
+      "step": 9540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185160649221597e-06,
+      "loss": 0.6884,
+      "step": 9550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.168598873799272e-06,
+      "loss": 0.6888,
+      "step": 9560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1520370983769465e-06,
+      "loss": 0.689,
+      "step": 9570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.135475322954621e-06,
+      "loss": 0.6901,
+      "step": 9580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.118913547532296e-06,
+      "loss": 0.6735,
+      "step": 9590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.10235177210997e-06,
+      "loss": 0.6679,
+      "step": 9600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0857899966876455e-06,
+      "loss": 0.6882,
+      "step": 9610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.06922822126532e-06,
+      "loss": 0.6837,
+      "step": 9620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.052666445842995e-06,
+      "loss": 0.6831,
+      "step": 9630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.03610467042067e-06,
+      "loss": 0.6666,
+      "step": 9640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.019542894998344e-06,
+      "loss": 0.6784,
+      "step": 9650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002981119576019e-06,
+      "loss": 0.6769,
+      "step": 9660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.986419344153694e-06,
+      "loss": 0.6958,
+      "step": 9670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969857568731368e-06,
+      "loss": 0.6726,
+      "step": 9680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9532957933090436e-06,
+      "loss": 0.7066,
+      "step": 9690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936734017886718e-06,
+      "loss": 0.6978,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.920172242464393e-06,
+      "loss": 0.6548,
+      "step": 9710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.903610467042067e-06,
+      "loss": 0.6825,
+      "step": 9720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.887048691619742e-06,
+      "loss": 0.7016,
+      "step": 9730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.870486916197417e-06,
+      "loss": 0.6798,
+      "step": 9740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.853925140775092e-06,
+      "loss": 0.6777,
+      "step": 9750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.837363365352766e-06,
+      "loss": 0.6824,
+      "step": 9760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.820801589930441e-06,
+      "loss": 0.6776,
+      "step": 9770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8042398145081157e-06,
+      "loss": 0.6804,
+      "step": 9780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7876780390857903e-06,
+      "loss": 0.6841,
+      "step": 9790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7711162636634652e-06,
+      "loss": 0.6955,
+      "step": 9800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7545544882411398e-06,
+      "loss": 0.6941,
+      "step": 9810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7379927128188143e-06,
+      "loss": 0.6867,
+      "step": 9820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7214309373964893e-06,
+      "loss": 0.6879,
+      "step": 9830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.704869161974164e-06,
+      "loss": 0.6672,
+      "step": 9840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.688307386551839e-06,
+      "loss": 0.6823,
+      "step": 9850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6717456111295133e-06,
+      "loss": 0.6929,
+      "step": 9860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6551838357071883e-06,
+      "loss": 0.6916,
+      "step": 9870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6386220602848624e-06,
+      "loss": 0.6784,
+      "step": 9880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6220602848625374e-06,
+      "loss": 0.7008,
+      "step": 9890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6054985094402124e-06,
+      "loss": 0.6741,
+      "step": 9900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.588936734017887e-06,
+      "loss": 0.6697,
+      "step": 9910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.572374958595562e-06,
+      "loss": 0.6841,
+      "step": 9920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.555813183173236e-06,
+      "loss": 0.6887,
+      "step": 9930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.539251407750911e-06,
+      "loss": 0.6728,
+      "step": 9940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.522689632328586e-06,
+      "loss": 0.7005,
+      "step": 9950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5061278569062605e-06,
+      "loss": 0.6911,
+      "step": 9960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4895660814839354e-06,
+      "loss": 0.6689,
+      "step": 9970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4730043060616104e-06,
+      "loss": 0.6729,
+      "step": 9980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4564425306392845e-06,
+      "loss": 0.6849,
+      "step": 9990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4398807552169595e-06,
+      "loss": 0.7005,
+      "step": 10000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.423318979794634e-06,
+      "loss": 0.6662,
+      "step": 10010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.406757204372309e-06,
+      "loss": 0.6629,
+      "step": 10020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.390195428949984e-06,
+      "loss": 0.6909,
+      "step": 10030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3736336535276585e-06,
+      "loss": 0.6753,
+      "step": 10040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.357071878105333e-06,
+      "loss": 0.69,
+      "step": 10050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3405101026830076e-06,
+      "loss": 0.7027,
+      "step": 10060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3239483272606826e-06,
+      "loss": 0.6927,
+      "step": 10070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3073865518383575e-06,
+      "loss": 0.6925,
+      "step": 10080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.290824776416032e-06,
+      "loss": 0.6948,
+      "step": 10090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2742630009937066e-06,
+      "loss": 0.678,
+      "step": 10100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.257701225571381e-06,
+      "loss": 0.6871,
+      "step": 10110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.241139450149056e-06,
+      "loss": 0.6901,
+      "step": 10120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.224577674726731e-06,
+      "loss": 0.711,
+      "step": 10130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2080158993044056e-06,
+      "loss": 0.6787,
+      "step": 10140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1914541238820806e-06,
+      "loss": 0.6866,
+      "step": 10150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1748923484597547e-06,
+      "loss": 0.6625,
+      "step": 10160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1583305730374297e-06,
+      "loss": 0.6877,
+      "step": 10170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1417687976151047e-06,
+      "loss": 0.6541,
+      "step": 10180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125207022192779e-06,
+      "loss": 0.7141,
+      "step": 10190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.108645246770454e-06,
+      "loss": 0.6778,
+      "step": 10200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.092083471348129e-06,
+      "loss": 0.6597,
+      "step": 10210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0755216959258033e-06,
+      "loss": 0.6767,
+      "step": 10220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0589599205034782e-06,
+      "loss": 0.669,
+      "step": 10230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0423981450811528e-06,
+      "loss": 0.6776,
+      "step": 10240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0258363696588277e-06,
+      "loss": 0.6938,
+      "step": 10250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0092745942365027e-06,
+      "loss": 0.6821,
+      "step": 10260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.992712818814177e-06,
+      "loss": 0.7042,
+      "step": 10270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.976151043391852e-06,
+      "loss": 0.6736,
+      "step": 10280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9595892679695263e-06,
+      "loss": 0.6702,
+      "step": 10290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9430274925472013e-06,
+      "loss": 0.6893,
+      "step": 10300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9264657171248763e-06,
+      "loss": 0.7001,
+      "step": 10310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.909903941702551e-06,
+      "loss": 0.6849,
+      "step": 10320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8933421662802254e-06,
+      "loss": 0.6926,
+      "step": 10330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8767803908579e-06,
+      "loss": 0.6747,
+      "step": 10340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.860218615435575e-06,
+      "loss": 0.6738,
+      "step": 10350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.84365684001325e-06,
+      "loss": 0.6696,
+      "step": 10360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8270950645909244e-06,
+      "loss": 0.6668,
+      "step": 10370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8105332891685993e-06,
+      "loss": 0.6557,
+      "step": 10380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7939715137462735e-06,
+      "loss": 0.6953,
+      "step": 10390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7774097383239484e-06,
+      "loss": 0.6735,
+      "step": 10400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7608479629016234e-06,
+      "loss": 0.6647,
+      "step": 10410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.744286187479298e-06,
+      "loss": 0.6768,
+      "step": 10420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.727724412056973e-06,
+      "loss": 0.6879,
+      "step": 10430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.711162636634647e-06,
+      "loss": 0.6894,
+      "step": 10440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.694600861212322e-06,
+      "loss": 0.6771,
+      "step": 10450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.678039085789997e-06,
+      "loss": 0.6847,
+      "step": 10460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6614773103676715e-06,
+      "loss": 0.674,
+      "step": 10470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6449155349453465e-06,
+      "loss": 0.6718,
+      "step": 10480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6283537595230214e-06,
+      "loss": 0.6741,
+      "step": 10490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6117919841006956e-06,
+      "loss": 0.6862,
+      "step": 10500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5952302086783705e-06,
+      "loss": 0.6771,
+      "step": 10510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.578668433256045e-06,
+      "loss": 0.6869,
+      "step": 10520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.56210665783372e-06,
+      "loss": 0.6686,
+      "step": 10530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.545544882411395e-06,
+      "loss": 0.7111,
+      "step": 10540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5289831069890695e-06,
+      "loss": 0.6667,
+      "step": 10550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.512421331566744e-06,
+      "loss": 0.7005,
+      "step": 10560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4958595561444186e-06,
+      "loss": 0.6951,
+      "step": 10570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4792977807220936e-06,
+      "loss": 0.6828,
+      "step": 10580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4627360052997686e-06,
+      "loss": 0.6806,
+      "step": 10590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.446174229877443e-06,
+      "loss": 0.6937,
+      "step": 10600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4296124544551176e-06,
+      "loss": 0.657,
+      "step": 10610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4130506790327926e-06,
+      "loss": 0.6596,
+      "step": 10620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.396488903610467e-06,
+      "loss": 0.6831,
+      "step": 10630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.379927128188142e-06,
+      "loss": 0.6823,
+      "step": 10640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3633653527658167e-06,
+      "loss": 0.7053,
+      "step": 10650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3468035773434912e-06,
+      "loss": 0.6847,
+      "step": 10660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.330241801921166e-06,
+      "loss": 0.6786,
+      "step": 10670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.313680026498841e-06,
+      "loss": 0.6821,
+      "step": 10680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2971182510765157e-06,
+      "loss": 0.6789,
+      "step": 10690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2805564756541902e-06,
+      "loss": 0.6876,
+      "step": 10700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2639947002318648e-06,
+      "loss": 0.6953,
+      "step": 10710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2474329248095397e-06,
+      "loss": 0.6708,
+      "step": 10720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2308711493872147e-06,
+      "loss": 0.6672,
+      "step": 10730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2143093739648893e-06,
+      "loss": 0.6915,
+      "step": 10740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.197747598542564e-06,
+      "loss": 0.6778,
+      "step": 10750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1811858231202388e-06,
+      "loss": 0.6638,
+      "step": 10760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1646240476979133e-06,
+      "loss": 0.6837,
+      "step": 10770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1480622722755883e-06,
+      "loss": 0.6754,
+      "step": 10780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.131500496853263e-06,
+      "loss": 0.6907,
+      "step": 10790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1149387214309374e-06,
+      "loss": 0.6872,
+      "step": 10800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0983769460086123e-06,
+      "loss": 0.654,
+      "step": 10810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0818151705862873e-06,
+      "loss": 0.6903,
+      "step": 10820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.065253395163962e-06,
+      "loss": 0.689,
+      "step": 10830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0486916197416364e-06,
+      "loss": 0.6788,
+      "step": 10840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0321298443193113e-06,
+      "loss": 0.6777,
+      "step": 10850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.015568068896986e-06,
+      "loss": 0.6582,
+      "step": 10860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.999006293474661e-06,
+      "loss": 0.6589,
+      "step": 10870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9824445180523354e-06,
+      "loss": 0.6819,
+      "step": 10880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.96588274263001e-06,
+      "loss": 0.6713,
+      "step": 10890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.949320967207685e-06,
+      "loss": 0.6515,
+      "step": 10900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9327591917853595e-06,
+      "loss": 0.6986,
+      "step": 10910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9161974163630344e-06,
+      "loss": 0.682,
+      "step": 10920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8996356409407092e-06,
+      "loss": 0.6916,
+      "step": 10930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8830738655183837e-06,
+      "loss": 0.6816,
+      "step": 10940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8665120900960585e-06,
+      "loss": 0.6927,
+      "step": 10950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8499503146737332e-06,
+      "loss": 0.683,
+      "step": 10960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8333885392514078e-06,
+      "loss": 0.6598,
+      "step": 10970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8168267638290825e-06,
+      "loss": 0.6946,
+      "step": 10980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8002649884067575e-06,
+      "loss": 0.683,
+      "step": 10990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.783703212984432e-06,
+      "loss": 0.6824,
+      "step": 11000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7671414375621068e-06,
+      "loss": 0.6555,
+      "step": 11010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7505796621397815e-06,
+      "loss": 0.6743,
+      "step": 11020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.734017886717456e-06,
+      "loss": 0.6909,
+      "step": 11030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.717456111295131e-06,
+      "loss": 0.6726,
+      "step": 11040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7008943358728058e-06,
+      "loss": 0.6888,
+      "step": 11050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6843325604504804e-06,
+      "loss": 0.6622,
+      "step": 11060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6677707850281551e-06,
+      "loss": 0.6765,
+      "step": 11070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6512090096058297e-06,
+      "loss": 0.6746,
+      "step": 11080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6346472341835046e-06,
+      "loss": 0.6621,
+      "step": 11090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6180854587611794e-06,
+      "loss": 0.6723,
+      "step": 11100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601523683338854e-06,
+      "loss": 0.6568,
+      "step": 11110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5849619079165287e-06,
+      "loss": 0.6768,
+      "step": 11120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5684001324942036e-06,
+      "loss": 0.6681,
+      "step": 11130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5518383570718782e-06,
+      "loss": 0.6835,
+      "step": 11140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.535276581649553e-06,
+      "loss": 0.6768,
+      "step": 11150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5187148062272277e-06,
+      "loss": 0.659,
+      "step": 11160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5021530308049022e-06,
+      "loss": 0.6635,
+      "step": 11170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4855912553825772e-06,
+      "loss": 0.6824,
+      "step": 11180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.469029479960252e-06,
+      "loss": 0.6661,
+      "step": 11190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4524677045379265e-06,
+      "loss": 0.669,
+      "step": 11200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4359059291156013e-06,
+      "loss": 0.6538,
+      "step": 11210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4193441536932762e-06,
+      "loss": 0.668,
+      "step": 11220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4027823782709508e-06,
+      "loss": 0.6698,
+      "step": 11230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3862206028486255e-06,
+      "loss": 0.6953,
+      "step": 11240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3696588274263e-06,
+      "loss": 0.6649,
+      "step": 11250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3530970520039748e-06,
+      "loss": 0.6917,
+      "step": 11260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3365352765816498e-06,
+      "loss": 0.6786,
+      "step": 11270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3199735011593243e-06,
+      "loss": 0.6706,
+      "step": 11280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303411725736999e-06,
+      "loss": 0.6443,
+      "step": 11290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2868499503146738e-06,
+      "loss": 0.6557,
+      "step": 11300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2702881748923484e-06,
+      "loss": 0.6793,
+      "step": 11310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2537263994700234e-06,
+      "loss": 0.6842,
+      "step": 11320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.237164624047698e-06,
+      "loss": 0.6824,
+      "step": 11330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2206028486253729e-06,
+      "loss": 0.655,
+      "step": 11340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2040410732030474e-06,
+      "loss": 0.6759,
+      "step": 11350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1874792977807222e-06,
+      "loss": 0.6719,
+      "step": 11360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.170917522358397e-06,
+      "loss": 0.7026,
+      "step": 11370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1543557469360717e-06,
+      "loss": 0.6874,
+      "step": 11380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1377939715137464e-06,
+      "loss": 0.6738,
+      "step": 11390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.121232196091421e-06,
+      "loss": 0.6653,
+      "step": 11400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.104670420669096e-06,
+      "loss": 0.7066,
+      "step": 11410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0881086452467705e-06,
+      "loss": 0.6657,
+      "step": 11420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0715468698244452e-06,
+      "loss": 0.6847,
+      "step": 11430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.05498509440212e-06,
+      "loss": 0.6879,
+      "step": 11440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0384233189797947e-06,
+      "loss": 0.6816,
+      "step": 11450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0218615435574695e-06,
+      "loss": 0.6863,
+      "step": 11460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0052997681351443e-06,
+      "loss": 0.6828,
+      "step": 11470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.88737992712819e-07,
+      "loss": 0.6975,
+      "step": 11480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.721762172904936e-07,
+      "loss": 0.6667,
+      "step": 11490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.556144418681683e-07,
+      "loss": 0.6542,
+      "step": 11500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.390526664458431e-07,
+      "loss": 0.6654,
+      "step": 11510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.224908910235178e-07,
+      "loss": 0.664,
+      "step": 11520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.059291156011925e-07,
+      "loss": 0.6745,
+      "step": 11530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.893673401788673e-07,
+      "loss": 0.6779,
+      "step": 11540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.72805564756542e-07,
+      "loss": 0.6713,
+      "step": 11550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.562437893342166e-07,
+      "loss": 0.6756,
+      "step": 11560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.396820139118914e-07,
+      "loss": 0.6827,
+      "step": 11570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.231202384895661e-07,
+      "loss": 0.6704,
+      "step": 11580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.065584630672409e-07,
+      "loss": 0.6769,
+      "step": 11590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.899966876449155e-07,
+      "loss": 0.6684,
+      "step": 11600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.734349122225904e-07,
+      "loss": 0.6743,
+      "step": 11610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.56873136800265e-07,
+      "loss": 0.684,
+      "step": 11620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.403113613779397e-07,
+      "loss": 0.6763,
+      "step": 11630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.237495859556146e-07,
+      "loss": 0.6851,
+      "step": 11640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.071878105332892e-07,
+      "loss": 0.6731,
+      "step": 11650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.90626035110964e-07,
+      "loss": 0.6533,
+      "step": 11660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.740642596886386e-07,
+      "loss": 0.6933,
+      "step": 11670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.575024842663135e-07,
+      "loss": 0.6818,
+      "step": 11680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.409407088439881e-07,
+      "loss": 0.6569,
+      "step": 11690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.243789334216629e-07,
+      "loss": 0.671,
+      "step": 11700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.078171579993375e-07,
+      "loss": 0.6904,
+      "step": 11710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.912553825770123e-07,
+      "loss": 0.661,
+      "step": 11720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.74693607154687e-07,
+      "loss": 0.6925,
+      "step": 11730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.581318317323618e-07,
+      "loss": 0.6645,
+      "step": 11740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.415700563100365e-07,
+      "loss": 0.6615,
+      "step": 11750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.250082808877112e-07,
+      "loss": 0.6738,
+      "step": 11760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.08446505465386e-07,
+      "loss": 0.6582,
+      "step": 11770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.918847300430606e-07,
+      "loss": 0.6757,
+      "step": 11780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.753229546207354e-07,
+      "loss": 0.6614,
+      "step": 11790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.587611791984101e-07,
+      "loss": 0.6549,
+      "step": 11800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.421994037760848e-07,
+      "loss": 0.6602,
+      "step": 11810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2563762835375957e-07,
+      "loss": 0.6911,
+      "step": 11820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0907585293143427e-07,
+      "loss": 0.6826,
+      "step": 11830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.92514077509109e-07,
+      "loss": 0.6793,
+      "step": 11840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.759523020867838e-07,
+      "loss": 0.6798,
+      "step": 11850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5939052666445843e-07,
+      "loss": 0.6768,
+      "step": 11860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.428287512421332e-07,
+      "loss": 0.6803,
+      "step": 11870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.262669758198079e-07,
+      "loss": 0.6607,
+      "step": 11880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0970520039748264e-07,
+      "loss": 0.669,
+      "step": 11890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9314342497515734e-07,
+      "loss": 0.6741,
+      "step": 11900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.765816495528321e-07,
+      "loss": 0.6704,
+      "step": 11910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.600198741305068e-07,
+      "loss": 0.6772,
+      "step": 11920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4345809870818156e-07,
+      "loss": 0.675,
+      "step": 11930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2689632328585626e-07,
+      "loss": 0.6804,
+      "step": 11940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.10334547863531e-07,
+      "loss": 0.6593,
+      "step": 11950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9377277244120572e-07,
+      "loss": 0.6659,
+      "step": 11960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7721099701888042e-07,
+      "loss": 0.6811,
+      "step": 11970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6064922159655515e-07,
+      "loss": 0.6964,
+      "step": 11980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4408744617422987e-07,
+      "loss": 0.6941,
+      "step": 11990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2752567075190463e-07,
+      "loss": 0.6733,
+      "step": 12000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1096389532957933e-07,
+      "loss": 0.6892,
+      "step": 12010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.440211990725406e-08,
+      "loss": 0.6775,
+      "step": 12020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.784034448492879e-08,
+      "loss": 0.6826,
+      "step": 12030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.127856906260352e-08,
+      "loss": 0.7036,
+      "step": 12040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.471679364027824e-08,
+      "loss": 0.6749,
+      "step": 12050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8155018217952968e-08,
+      "loss": 0.6851,
+      "step": 12060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1593242795627693e-08,
+      "loss": 0.6726,
+      "step": 12070
+    },
+    {
+      "epoch": 1.0,
+      "step": 12077,
+      "total_flos": 2.2086208395348541e+18,
+      "train_loss": 1.444948594361097,
+      "train_runtime": 31915.8638,
+      "train_samples_per_second": 12.109,
+      "train_steps_per_second": 0.378
+    }
+  ],
+  "max_steps": 12077,
+  "num_train_epochs": 1,
+  "total_flos": 2.2086208395348541e+18,
+  "trial_name": null,
+  "trial_params": null
+}