diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9619 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.13653449842771992,
+  "eval_steps": 500,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 2.7297,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 2.7093,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 2.7496,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 2.7079,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 2.776,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 2.6999,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 2.6693,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 2.6798,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 2.6673,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 2.6582,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 2.6594,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 2.6427,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 2.6587,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7e-05,
+      "loss": 2.6303,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.6187,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-05,
+      "loss": 2.601,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.5e-05,
+      "loss": 2.6368,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-05,
+      "loss": 2.5823,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.5e-05,
+      "loss": 2.5823,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 2.6158,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999819709864e-05,
+      "loss": 2.5563,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999999278839467e-05,
+      "loss": 2.6209,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999998377388849e-05,
+      "loss": 2.5776,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999997115358075e-05,
+      "loss": 2.5985,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999995492747234e-05,
+      "loss": 2.5543,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999993509556447e-05,
+      "loss": 2.5397,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999991165785854e-05,
+      "loss": 2.5433,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999988461435623e-05,
+      "loss": 2.5178,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999985396505953e-05,
+      "loss": 2.538,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999981970997063e-05,
+      "loss": 2.5748,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999978184909198e-05,
+      "loss": 2.5068,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999974038242635e-05,
+      "loss": 2.533,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999969530997668e-05,
+      "loss": 2.503,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999964663174628e-05,
+      "loss": 2.4623,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999959434773862e-05,
+      "loss": 2.5012,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999953845795749e-05,
+      "loss": 2.5356,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999947896240689e-05,
+      "loss": 2.5203,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999941586109114e-05,
+      "loss": 2.4911,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999934915401479e-05,
+      "loss": 2.5594,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999927884118265e-05,
+      "loss": 2.5002,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999920492259978e-05,
+      "loss": 2.462,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999912739827151e-05,
+      "loss": 2.4508,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999904626820342e-05,
+      "loss": 2.4541,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99989615324014e-05,
+      "loss": 2.4389,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999887319087154e-05,
+      "loss": 2.4549,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999878124362018e-05,
+      "loss": 2.4773,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999868569065401e-05,
+      "loss": 2.4742,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999858653197986e-05,
+      "loss": 2.5173,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999848376760493e-05,
+      "loss": 2.4365,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999837739753662e-05,
+      "loss": 2.4962,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999826742178258e-05,
+      "loss": 2.4823,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999815384035075e-05,
+      "loss": 2.4612,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999803665324932e-05,
+      "loss": 2.4552,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999791586048677e-05,
+      "loss": 2.4297,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999779146207177e-05,
+      "loss": 2.5136,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999766345801331e-05,
+      "loss": 2.4892,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999753184832062e-05,
+      "loss": 2.4471,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999739663300319e-05,
+      "loss": 2.4357,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999725781207076e-05,
+      "loss": 2.4141,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999711538553337e-05,
+      "loss": 2.4309,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999696935340127e-05,
+      "loss": 2.4301,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9996819715685e-05,
+      "loss": 2.4173,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999666647239533e-05,
+      "loss": 2.4822,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999650962354334e-05,
+      "loss": 2.4642,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999634916914033e-05,
+      "loss": 2.435,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999618510919788e-05,
+      "loss": 2.4362,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999601744372779e-05,
+      "loss": 2.4506,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999584617274218e-05,
+      "loss": 2.4675,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99956712962534e-05,
+      "loss": 2.4715,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999549281427406e-05,
+      "loss": 2.4242,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999531072681702e-05,
+      "loss": 2.4689,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999512503389541e-05,
+      "loss": 2.371,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999493573552264e-05,
+      "loss": 2.4047,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999474283171235e-05,
+      "loss": 2.4193,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999454632247847e-05,
+      "loss": 2.4168,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999434620783514e-05,
+      "loss": 2.3914,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99941424877968e-05,
+      "loss": 2.4214,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999393516237815e-05,
+      "loss": 2.3991,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999372423159415e-05,
+      "loss": 2.3518,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999350969545998e-05,
+      "loss": 2.4358,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999329155399116e-05,
+      "loss": 2.4059,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999306980720338e-05,
+      "loss": 2.3931,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999284445511264e-05,
+      "loss": 2.3776,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99926154977352e-05,
+      "loss": 2.4113,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999238293508757e-05,
+      "loss": 2.3481,
+      "step": 425
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999214676718651e-05,
+      "loss": 2.4262,
+      "step": 430
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999190699404908e-05,
+      "loss": 2.3812,
+      "step": 435
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999166361569255e-05,
+      "loss": 2.396,
+      "step": 440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999141663213446e-05,
+      "loss": 2.3798,
+      "step": 445
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999116604339264e-05,
+      "loss": 2.3647,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999091184948517e-05,
+      "loss": 2.4017,
+      "step": 455
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999065405043035e-05,
+      "loss": 2.3834,
+      "step": 460
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99903926462468e-05,
+      "loss": 2.4582,
+      "step": 465
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999012763695337e-05,
+      "loss": 2.3242,
+      "step": 470
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998985902256915e-05,
+      "loss": 2.3769,
+      "step": 475
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998958680311353e-05,
+      "loss": 2.3924,
+      "step": 480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998931097860613e-05,
+      "loss": 2.3822,
+      "step": 485
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998903154906685e-05,
+      "loss": 2.3611,
+      "step": 490
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998874851451582e-05,
+      "loss": 2.4209,
+      "step": 495
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998846187497351e-05,
+      "loss": 2.3927,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998817163046053e-05,
+      "loss": 2.3387,
+      "step": 505
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998787778099783e-05,
+      "loss": 2.3779,
+      "step": 510
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99875803266066e-05,
+      "loss": 2.4246,
+      "step": 515
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99872792673083e-05,
+      "loss": 2.406,
+      "step": 520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998697460312463e-05,
+      "loss": 2.2727,
+      "step": 525
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998666633407758e-05,
+      "loss": 2.3335,
+      "step": 530
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998635446018936e-05,
+      "loss": 2.4138,
+      "step": 535
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998603898148247e-05,
+      "loss": 2.3348,
+      "step": 540
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998571989797967e-05,
+      "loss": 2.3799,
+      "step": 545
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998539720970394e-05,
+      "loss": 2.3255,
+      "step": 550
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998507091667859e-05,
+      "loss": 2.3932,
+      "step": 555
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998474101892712e-05,
+      "loss": 2.3416,
+      "step": 560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998440751647335e-05,
+      "loss": 2.38,
+      "step": 565
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998407040934132e-05,
+      "loss": 2.3981,
+      "step": 570
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998372969755532e-05,
+      "loss": 2.3835,
+      "step": 575
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998338538113993e-05,
+      "loss": 2.326,
+      "step": 580
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998303746012002e-05,
+      "loss": 2.363,
+      "step": 585
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998268593452063e-05,
+      "loss": 2.3389,
+      "step": 590
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998233080436711e-05,
+      "loss": 2.346,
+      "step": 595
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998197206968511e-05,
+      "loss": 2.3378,
+      "step": 600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99816097305005e-05,
+      "loss": 2.3944,
+      "step": 605
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998124378683935e-05,
+      "loss": 2.3273,
+      "step": 610
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998087423872811e-05,
+      "loss": 2.3643,
+      "step": 615
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998050108619341e-05,
+      "loss": 2.3675,
+      "step": 620
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998012432926216e-05,
+      "loss": 2.2795,
+      "step": 625
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997974396796155e-05,
+      "loss": 2.3167,
+      "step": 630
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997936000231898e-05,
+      "loss": 2.3327,
+      "step": 635
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997897243236214e-05,
+      "loss": 2.3441,
+      "step": 640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997858125811901e-05,
+      "loss": 2.31,
+      "step": 645
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997818647961777e-05,
+      "loss": 2.289,
+      "step": 650
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99777880968869e-05,
+      "loss": 2.3149,
+      "step": 655
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997738610995515e-05,
+      "loss": 2.295,
+      "step": 660
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997698051885148e-05,
+      "loss": 2.2835,
+      "step": 665
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997657132360514e-05,
+      "loss": 2.3303,
+      "step": 670
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997615852424567e-05,
+      "loss": 2.3221,
+      "step": 675
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99757421208028e-05,
+      "loss": 2.3166,
+      "step": 680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99753221133066e-05,
+      "loss": 2.3123,
+      "step": 685
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997489850178732e-05,
+      "loss": 2.3864,
+      "step": 690
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997447128627555e-05,
+      "loss": 2.3351,
+      "step": 695
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997404046680206e-05,
+      "loss": 2.2974,
+      "step": 700
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997360604339793e-05,
+      "loss": 2.3133,
+      "step": 705
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997316801609451e-05,
+      "loss": 2.3086,
+      "step": 710
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997272638492337e-05,
+      "loss": 2.3344,
+      "step": 715
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997228114991637e-05,
+      "loss": 2.2582,
+      "step": 720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997183231110561e-05,
+      "loss": 2.298,
+      "step": 725
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997137986852345e-05,
+      "loss": 2.2755,
+      "step": 730
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997092382220253e-05,
+      "loss": 2.3259,
+      "step": 735
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997046417217573e-05,
+      "loss": 2.3079,
+      "step": 740
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997000091847621e-05,
+      "loss": 2.3017,
+      "step": 745
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996953406113739e-05,
+      "loss": 2.2718,
+      "step": 750
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996906360019289e-05,
+      "loss": 2.291,
+      "step": 755
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99685895356767e-05,
+      "loss": 2.3429,
+      "step": 760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996811186762296e-05,
+      "loss": 2.2799,
+      "step": 765
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996763059606611e-05,
+      "loss": 2.3232,
+      "step": 770
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99671457210409e-05,
+      "loss": 2.3127,
+      "step": 775
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99666572425823e-05,
+      "loss": 2.2612,
+      "step": 780
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99661651607255e-05,
+      "loss": 2.3229,
+      "step": 785
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996566947550598e-05,
+      "loss": 2.266,
+      "step": 790
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996517018695952e-05,
+      "loss": 2.2997,
+      "step": 795
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996466729512213e-05,
+      "loss": 2.2767,
+      "step": 800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996416080003006e-05,
+      "loss": 2.3467,
+      "step": 805
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996365070171982e-05,
+      "loss": 2.2605,
+      "step": 810
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996313700022824e-05,
+      "loss": 2.3337,
+      "step": 815
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996261969559233e-05,
+      "loss": 2.3242,
+      "step": 820
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996209878784942e-05,
+      "loss": 2.2541,
+      "step": 825
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996157427703703e-05,
+      "loss": 2.2836,
+      "step": 830
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996104616319303e-05,
+      "loss": 2.2745,
+      "step": 835
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996051444635551e-05,
+      "loss": 2.319,
+      "step": 840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99599791265628e-05,
+      "loss": 2.2845,
+      "step": 845
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99594402038535e-05,
+      "loss": 2.3005,
+      "step": 850
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995889767826647e-05,
+      "loss": 2.3264,
+      "step": 855
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995835154984084e-05,
+      "loss": 2.3229,
+      "step": 860
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995780181861601e-05,
+      "loss": 2.2728,
+      "step": 865
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995724848463161e-05,
+      "loss": 2.2675,
+      "step": 870
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.995669154792755e-05,
+      "loss": 2.3078,
+      "step": 875
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9956131008544e-05,
+      "loss": 2.3089,
+      "step": 880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995556686652136e-05,
+      "loss": 2.3142,
+      "step": 885
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995499912190034e-05,
+      "loss": 2.2928,
+      "step": 890
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995442777472187e-05,
+      "loss": 2.292,
+      "step": 895
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995385282502716e-05,
+      "loss": 2.2979,
+      "step": 900
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995327427285765e-05,
+      "loss": 2.2995,
+      "step": 905
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99526921182551e-05,
+      "loss": 2.2658,
+      "step": 910
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995210636126148e-05,
+      "loss": 2.3132,
+      "step": 915
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995151700191901e-05,
+      "loss": 2.2923,
+      "step": 920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995092404027022e-05,
+      "loss": 2.2885,
+      "step": 925
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995032747635787e-05,
+      "loss": 2.2838,
+      "step": 930
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994972731022496e-05,
+      "loss": 2.2734,
+      "step": 935
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994912354191479e-05,
+      "loss": 2.2622,
+      "step": 940
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99485161714709e-05,
+      "loss": 2.3007,
+      "step": 945
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994790519893706e-05,
+      "loss": 2.2944,
+      "step": 950
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99472906243574e-05,
+      "loss": 2.2571,
+      "step": 955
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994667244777618e-05,
+      "loss": 2.3393,
+      "step": 960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994605066923799e-05,
+      "loss": 2.2864,
+      "step": 965
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99454252887877e-05,
+      "loss": 2.2792,
+      "step": 970
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994479630647037e-05,
+      "loss": 2.319,
+      "step": 975
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994416372233139e-05,
+      "loss": 2.226,
+      "step": 980
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994352753641635e-05,
+      "loss": 2.246,
+      "step": 985
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994288774877118e-05,
+      "loss": 2.2956,
+      "step": 990
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994224435944197e-05,
+      "loss": 2.2692,
+      "step": 995
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994159736847515e-05,
+      "loss": 2.2431,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994094677591734e-05,
+      "loss": 2.2092,
+      "step": 1005
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99402925818155e-05,
+      "loss": 2.2931,
+      "step": 1010
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99396347862168e-05,
+      "loss": 2.2596,
+      "step": 1015
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993897338916864e-05,
+      "loss": 2.2861,
+      "step": 1020
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993830839071875e-05,
+      "loss": 2.2458,
+      "step": 1025
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99376397909151e-05,
+      "loss": 2.2483,
+      "step": 1030
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993696758980589e-05,
+      "loss": 2.2892,
+      "step": 1035
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99362917874396e-05,
+      "loss": 2.265,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993561238386494e-05,
+      "loss": 2.296,
+      "step": 1045
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993492937913094e-05,
+      "loss": 2.2434,
+      "step": 1050
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993424277328685e-05,
+      "loss": 2.2678,
+      "step": 1055
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993355256638216e-05,
+      "loss": 2.2604,
+      "step": 1060
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993285875846667e-05,
+      "loss": 2.228,
+      "step": 1065
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993216134959042e-05,
+      "loss": 2.2633,
+      "step": 1070
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993146033980367e-05,
+      "loss": 2.2369,
+      "step": 1075
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993075572915702e-05,
+      "loss": 2.2116,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993004751770125e-05,
+      "loss": 2.2631,
+      "step": 1085
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992933570548745e-05,
+      "loss": 2.2146,
+      "step": 1090
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992862029256694e-05,
+      "loss": 2.2355,
+      "step": 1095
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992790127899132e-05,
+      "loss": 2.2717,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992717866481244e-05,
+      "loss": 2.2648,
+      "step": 1105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992645245008244e-05,
+      "loss": 2.268,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992572263485365e-05,
+      "loss": 2.2406,
+      "step": 1115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99249892191787e-05,
+      "loss": 2.3071,
+      "step": 1120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992425220311052e-05,
+      "loss": 2.2263,
+      "step": 1125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992351158670224e-05,
+      "loss": 2.292,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992276737000726e-05,
+      "loss": 2.2511,
+      "step": 1135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992201955307924e-05,
+      "loss": 2.248,
+      "step": 1140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992126813597216e-05,
+      "loss": 2.231,
+      "step": 1145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992051311874017e-05,
+      "loss": 2.2864,
+      "step": 1150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991975450143772e-05,
+      "loss": 2.2932,
+      "step": 1155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991899228411954e-05,
+      "loss": 2.26,
+      "step": 1160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991822646684058e-05,
+      "loss": 2.2538,
+      "step": 1165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991745704965606e-05,
+      "loss": 2.2455,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991668403262149e-05,
+      "loss": 2.2473,
+      "step": 1175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991590741579259e-05,
+      "loss": 2.2233,
+      "step": 1180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99151271992254e-05,
+      "loss": 2.257,
+      "step": 1185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991434338297617e-05,
+      "loss": 2.2439,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991355596710141e-05,
+      "loss": 2.2127,
+      "step": 1195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991276495165794e-05,
+      "loss": 2.2573,
+      "step": 1200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991197033670277e-05,
+      "loss": 2.2529,
+      "step": 1205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991117212229321e-05,
+      "loss": 2.2293,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.991037030848685e-05,
+      "loss": 2.259,
+      "step": 1215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990956489534148e-05,
+      "loss": 2.2543,
+      "step": 1220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990875588291521e-05,
+      "loss": 2.2011,
+      "step": 1225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990794327126638e-05,
+      "loss": 2.2488,
+      "step": 1230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990712706045358e-05,
+      "loss": 2.2673,
+      "step": 1235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990630725053567e-05,
+      "loss": 2.1819,
+      "step": 1240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990548384157177e-05,
+      "loss": 2.1845,
+      "step": 1245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990465683362129e-05,
+      "loss": 2.2586,
+      "step": 1250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990382622674385e-05,
+      "loss": 2.279,
+      "step": 1255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990299202099933e-05,
+      "loss": 2.2524,
+      "step": 1260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990215421644793e-05,
+      "loss": 2.2246,
+      "step": 1265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990131281315005e-05,
+      "loss": 2.2482,
+      "step": 1270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.990046781116635e-05,
+      "loss": 2.2675,
+      "step": 1275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989961921055779e-05,
+      "loss": 2.2318,
+      "step": 1280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989876701138557e-05,
+      "loss": 2.2426,
+      "step": 1285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989791121371114e-05,
+      "loss": 2.2536,
+      "step": 1290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989705181759622e-05,
+      "loss": 2.1798,
+      "step": 1295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989618882310279e-05,
+      "loss": 2.2312,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989532223029306e-05,
+      "loss": 2.2495,
+      "step": 1305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989445203922956e-05,
+      "loss": 2.2027,
+      "step": 1310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989357824997502e-05,
+      "loss": 2.2185,
+      "step": 1315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989270086259246e-05,
+      "loss": 2.2289,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989181987714518e-05,
+      "loss": 2.2361,
+      "step": 1325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989093529369667e-05,
+      "loss": 2.2164,
+      "step": 1330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.989004711231075e-05,
+      "loss": 2.2083,
+      "step": 1335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988915533305147e-05,
+      "loss": 2.2596,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988825995598312e-05,
+      "loss": 2.2736,
+      "step": 1345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98873609811703e-05,
+      "loss": 2.2601,
+      "step": 1350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988645840867782e-05,
+      "loss": 2.246,
+      "step": 1355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98855522385708e-05,
+      "loss": 2.2459,
+      "step": 1360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988464247091456e-05,
+      "loss": 2.2403,
+      "step": 1365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98837291057747e-05,
+      "loss": 2.2501,
+      "step": 1370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988281214321713e-05,
+      "loss": 2.1828,
+      "step": 1375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988189158330794e-05,
+      "loss": 2.188,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988096742611353e-05,
+      "loss": 2.2586,
+      "step": 1385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.988003967170055e-05,
+      "loss": 2.1453,
+      "step": 1390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98791083201359e-05,
+      "loss": 2.2125,
+      "step": 1395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987817337148674e-05,
+      "loss": 2.2291,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987723482582051e-05,
+      "loss": 2.1992,
+      "step": 1405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987629268320489e-05,
+      "loss": 2.1922,
+      "step": 1410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987534694370781e-05,
+      "loss": 2.2721,
+      "step": 1415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98743976073975e-05,
+      "loss": 2.2059,
+      "step": 1420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987344467434239e-05,
+      "loss": 2.2222,
+      "step": 1425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987248814461123e-05,
+      "loss": 2.2523,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.987152801827299e-05,
+      "loss": 2.2351,
+      "step": 1435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98705642953969e-05,
+      "loss": 2.2103,
+      "step": 1440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986959697605247e-05,
+      "loss": 2.2197,
+      "step": 1445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986862606030947e-05,
+      "loss": 2.2446,
+      "step": 1450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.98676515482379e-05,
+      "loss": 2.2533,
+      "step": 1455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.986667343990803e-05,
+      "loss": 2.2112,
+      "step": 1460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986569173539044e-05,
+      "loss": 2.2153,
+      "step": 1465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986470643475587e-05,
+      "loss": 2.2329,
+      "step": 1470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986371753807543e-05,
+      "loss": 2.2147,
+      "step": 1475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98627250454204e-05,
+      "loss": 2.1377,
+      "step": 1480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986172895686236e-05,
+      "loss": 2.2269,
+      "step": 1485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986072927247316e-05,
+      "loss": 2.2709,
+      "step": 1490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985972599232489e-05,
+      "loss": 2.2383,
+      "step": 1495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985871911648987e-05,
+      "loss": 2.2562,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985770864504076e-05,
+      "loss": 2.237,
+      "step": 1505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98566945780504e-05,
+      "loss": 2.2035,
+      "step": 1510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985567691559192e-05,
+      "loss": 2.1849,
+      "step": 1515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985465565773872e-05,
+      "loss": 2.1778,
+      "step": 1520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985363080456446e-05,
+      "loss": 2.1869,
+      "step": 1525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985260235614302e-05,
+      "loss": 2.2434,
+      "step": 1530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98515703125486e-05,
+      "loss": 2.1914,
+      "step": 1535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98505346738556e-05,
+      "loss": 2.2126,
+      "step": 1540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984949544013873e-05,
+      "loss": 2.1904,
+      "step": 1545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984845261147292e-05,
+      "loss": 2.1837,
+      "step": 1550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984740618793335e-05,
+      "loss": 2.1911,
+      "step": 1555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984635616959553e-05,
+      "loss": 2.2533,
+      "step": 1560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984530255653517e-05,
+      "loss": 2.168,
+      "step": 1565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984424534882824e-05,
+      "loss": 2.2225,
+      "step": 1570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984318454655099e-05,
+      "loss": 2.2263,
+      "step": 1575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984212014977992e-05,
+      "loss": 2.1751,
+      "step": 1580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984105215859177e-05,
+      "loss": 2.2001,
+      "step": 1585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98399805730636e-05,
+      "loss": 2.2273,
+      "step": 1590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983890539327266e-05,
+      "loss": 2.1849,
+      "step": 1595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983782661929649e-05,
+      "loss": 2.2115,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98367442512129e-05,
+      "loss": 2.1945,
+      "step": 1605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983565828909993e-05,
+      "loss": 2.1678,
+      "step": 1610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98345687330359e-05,
+      "loss": 2.1702,
+      "step": 1615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98334755830994e-05,
+      "loss": 2.2226,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983237883936925e-05,
+      "loss": 2.1825,
+      "step": 1625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983127850192453e-05,
+      "loss": 2.1838,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983017457084461e-05,
+      "loss": 2.2007,
+      "step": 1635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982906704620912e-05,
+      "loss": 2.2086,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982795592809789e-05,
+      "loss": 2.1763,
+      "step": 1645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982684121659104e-05,
+      "loss": 2.1939,
+      "step": 1650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982572291176902e-05,
+      "loss": 2.1988,
+      "step": 1655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982460101371243e-05,
+      "loss": 2.2159,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98234755225022e-05,
+      "loss": 2.2188,
+      "step": 1665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982234643821948e-05,
+      "loss": 2.1749,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98212137609457e-05,
+      "loss": 2.1928,
+      "step": 1675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982007749076255e-05,
+      "loss": 2.1866,
+      "step": 1680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981893762775196e-05,
+      "loss": 2.1737,
+      "step": 1685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981779417199613e-05,
+      "loss": 2.2033,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981664712357755e-05,
+      "loss": 2.1874,
+      "step": 1695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981549648257893e-05,
+      "loss": 2.2136,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981434224908323e-05,
+      "loss": 2.1775,
+      "step": 1705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98131844231737e-05,
+      "loss": 2.1935,
+      "step": 1710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981202300493384e-05,
+      "loss": 2.1892,
+      "step": 1715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.981085799444742e-05,
+      "loss": 2.1908,
+      "step": 1720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980968939179843e-05,
+      "loss": 2.1642,
+      "step": 1725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980851719707116e-05,
+      "loss": 2.1903,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980734141035015e-05,
+      "loss": 2.2671,
+      "step": 1735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980616203172017e-05,
+      "loss": 2.1318,
+      "step": 1740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980497906126629e-05,
+      "loss": 2.2091,
+      "step": 1745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980379249907382e-05,
+      "loss": 2.2291,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980260234522832e-05,
+      "loss": 2.1848,
+      "step": 1755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980140859981563e-05,
+      "loss": 2.1685,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.980021126292186e-05,
+      "loss": 2.2176,
+      "step": 1765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979901033463331e-05,
+      "loss": 2.1895,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97978058150366e-05,
+      "loss": 2.1474,
+      "step": 1775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97965977042186e-05,
+      "loss": 2.2125,
+      "step": 1780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979538600226646e-05,
+      "loss": 2.1732,
+      "step": 1785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979417070926754e-05,
+      "loss": 2.1646,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979295182530948e-05,
+      "loss": 2.1481,
+      "step": 1795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.979172935048018e-05,
+      "loss": 2.2126,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97905032848678e-05,
+      "loss": 2.2076,
+      "step": 1805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978927362856077e-05,
+      "loss": 2.1828,
+      "step": 1810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978804038164777e-05,
+      "loss": 2.1491,
+      "step": 1815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978680354421772e-05,
+      "loss": 2.1573,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978556311635982e-05,
+      "loss": 2.1997,
+      "step": 1825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978431909816353e-05,
+      "loss": 2.1763,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978307148971857e-05,
+      "loss": 2.1974,
+      "step": 1835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97818202911149e-05,
+      "loss": 2.1452,
+      "step": 1840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.978056550244276e-05,
+      "loss": 2.1494,
+      "step": 1845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977930712379265e-05,
+      "loss": 2.1462,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977804515525529e-05,
+      "loss": 2.154,
+      "step": 1855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977677959692169e-05,
+      "loss": 2.1765,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977551044888315e-05,
+      "loss": 2.216,
+      "step": 1865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977423771123118e-05,
+      "loss": 2.2046,
+      "step": 1870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977296138405756e-05,
+      "loss": 2.1547,
+      "step": 1875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.977168146745434e-05,
+      "loss": 2.2081,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97703979615138e-05,
+      "loss": 2.1872,
+      "step": 1885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976911086632854e-05,
+      "loss": 2.1962,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976782018199136e-05,
+      "loss": 2.1981,
+      "step": 1895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976652590859534e-05,
+      "loss": 2.1938,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976522804623381e-05,
+      "loss": 2.1785,
+      "step": 1905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976392659500037e-05,
+      "loss": 2.1273,
+      "step": 1910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976262155498887e-05,
+      "loss": 2.2476,
+      "step": 1915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976131292629346e-05,
+      "loss": 2.1255,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.976000070900849e-05,
+      "loss": 2.2096,
+      "step": 1925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975868490322856e-05,
+      "loss": 2.1124,
+      "step": 1930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97573655090486e-05,
+      "loss": 2.172,
+      "step": 1935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975604252656377e-05,
+      "loss": 2.1695,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975471595586943e-05,
+      "loss": 2.1509,
+      "step": 1945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975338579706128e-05,
+      "loss": 2.2007,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975205205023526e-05,
+      "loss": 2.1551,
+      "step": 1955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.975071471548752e-05,
+      "loss": 2.1442,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974937379291452e-05,
+      "loss": 2.242,
+      "step": 1965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974802928261295e-05,
+      "loss": 2.1443,
+      "step": 1970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974668118467979e-05,
+      "loss": 2.1681,
+      "step": 1975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974532949921225e-05,
+      "loss": 2.1305,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974397422630782e-05,
+      "loss": 2.2057,
+      "step": 1985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974261536606421e-05,
+      "loss": 2.1659,
+      "step": 1990
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.974125291857944e-05,
+      "loss": 2.2099,
+      "step": 1995
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973988688395175e-05,
+      "loss": 2.1634,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973851726227967e-05,
+      "loss": 2.1389,
+      "step": 2005
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973714405366194e-05,
+      "loss": 2.2261,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973576725819762e-05,
+      "loss": 2.1435,
+      "step": 2015
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973438687598599e-05,
+      "loss": 2.1905,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973300290712661e-05,
+      "loss": 2.1643,
+      "step": 2025
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973161535171927e-05,
+      "loss": 2.0964,
+      "step": 2030
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.973022420986402e-05,
+      "loss": 2.1616,
+      "step": 2035
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972882948166122e-05,
+      "loss": 2.1644,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.972743116721142e-05,
+      "loss": 2.1847,
+      "step": 2045
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.97260292666155e-05,
+      "loss": 2.1427,
+      "step": 2050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972462377997451e-05,
+      "loss": 2.1778,
+      "step": 2055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972321470738985e-05,
+      "loss": 2.1643,
+      "step": 2060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972180204896312e-05,
+      "loss": 2.1794,
+      "step": 2065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972038580479619e-05,
+      "loss": 2.1388,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97189659749912e-05,
+      "loss": 2.1595,
+      "step": 2075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971754255965055e-05,
+      "loss": 2.1442,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971611555887687e-05,
+      "loss": 2.1567,
+      "step": 2085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971468497277308e-05,
+      "loss": 2.1901,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971325080144237e-05,
+      "loss": 2.1716,
+      "step": 2095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971181304498813e-05,
+      "loss": 2.1283,
+      "step": 2100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971037170351407e-05,
+      "loss": 2.1248,
+      "step": 2105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970892677712413e-05,
+      "loss": 2.1701,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970747826592251e-05,
+      "loss": 2.1551,
+      "step": 2115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970602617001368e-05,
+      "loss": 2.152,
+      "step": 2120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970457048950232e-05,
+      "loss": 2.1531,
+      "step": 2125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970311122449347e-05,
+      "loss": 2.143,
+      "step": 2130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970164837509232e-05,
+      "loss": 2.1884,
+      "step": 2135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970018194140438e-05,
+      "loss": 2.195,
+      "step": 2140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969871192353539e-05,
+      "loss": 2.1641,
+      "step": 2145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969723832159139e-05,
+      "loss": 2.1544,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969576113567863e-05,
+      "loss": 2.194,
+      "step": 2155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969428036590364e-05,
+      "loss": 2.1973,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969279601237322e-05,
+      "loss": 2.1579,
+      "step": 2165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96913080751944e-05,
+      "loss": 2.1842,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96898165544745e-05,
+      "loss": 2.155,
+      "step": 2175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968832145032106e-05,
+      "loss": 2.1488,
+      "step": 2180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968682276284191e-05,
+      "loss": 2.1205,
+      "step": 2185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968532049214516e-05,
+      "loss": 2.1964,
+      "step": 2190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96838146383391e-05,
+      "loss": 2.1736,
+      "step": 2195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968230520153235e-05,
+      "loss": 2.1535,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968079218183377e-05,
+      "loss": 2.1015,
+      "step": 2205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967927557935246e-05,
+      "loss": 2.1237,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96777553941978e-05,
+      "loss": 2.1056,
+      "step": 2215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967623162647942e-05,
+      "loss": 2.1873,
+      "step": 2220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967470427630719e-05,
+      "loss": 2.125,
+      "step": 2225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967317334379129e-05,
+      "loss": 2.1026,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967163882904209e-05,
+      "loss": 2.1269,
+      "step": 2235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967010073217027e-05,
+      "loss": 2.1128,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966855905328676e-05,
+      "loss": 2.1471,
+      "step": 2245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966701379250272e-05,
+      "loss": 2.1784,
+      "step": 2250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96654649499296e-05,
+      "loss": 2.1544,
+      "step": 2255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966391252567911e-05,
+      "loss": 2.1268,
+      "step": 2260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966235651986319e-05,
+      "loss": 2.1488,
+      "step": 2265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966079693259403e-05,
+      "loss": 2.117,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965923376398414e-05,
+      "loss": 2.1698,
+      "step": 2275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965766701414623e-05,
+      "loss": 2.14,
+      "step": 2280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96560966831933e-05,
+      "loss": 2.1862,
+      "step": 2285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965452277123856e-05,
+      "loss": 2.2011,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965294527839557e-05,
+      "loss": 2.093,
+      "step": 2295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965136420477806e-05,
+      "loss": 2.144,
+      "step": 2300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964977955050006e-05,
+      "loss": 2.1561,
+      "step": 2305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964819131567582e-05,
+      "loss": 2.1045,
+      "step": 2310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964659950041993e-05,
+      "loss": 2.1619,
+      "step": 2315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964500410484714e-05,
+      "loss": 2.168,
+      "step": 2320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964340512907251e-05,
+      "loss": 2.1162,
+      "step": 2325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964180257321138e-05,
+      "loss": 2.1877,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.964019643737929e-05,
+      "loss": 2.1559,
+      "step": 2335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96385867216921e-05,
+      "loss": 2.1706,
+      "step": 2340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963697342626586e-05,
+      "loss": 2.1675,
+      "step": 2345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963535655121694e-05,
+      "loss": 2.1625,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963373609666192e-05,
+      "loss": 2.1434,
+      "step": 2355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963211206271769e-05,
+      "loss": 2.1458,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.963048444950134e-05,
+      "loss": 2.1973,
+      "step": 2365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962885325713028e-05,
+      "loss": 2.1271,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962721848572211e-05,
+      "loss": 2.1472,
+      "step": 2375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962558013539474e-05,
+      "loss": 2.1867,
+      "step": 2380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962393820626632e-05,
+      "loss": 2.1383,
+      "step": 2385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962229269845527e-05,
+      "loss": 2.069,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.962064361208023e-05,
+      "loss": 2.1885,
+      "step": 2395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961899094726016e-05,
+      "loss": 2.132,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961733470411422e-05,
+      "loss": 2.1795,
+      "step": 2405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961567488276187e-05,
+      "loss": 2.1244,
+      "step": 2410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961401148332278e-05,
+      "loss": 2.1192,
+      "step": 2415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961234450591693e-05,
+      "loss": 2.115,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.961067395066454e-05,
+      "loss": 2.1415,
+      "step": 2425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960899981768607e-05,
+      "loss": 2.1424,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960732210710226e-05,
+      "loss": 2.1006,
+      "step": 2435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.96056408190341e-05,
+      "loss": 2.1324,
+      "step": 2440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960395595360282e-05,
+      "loss": 2.1948,
+      "step": 2445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960226751092997e-05,
+      "loss": 2.0975,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.960057549113724e-05,
+      "loss": 2.1303,
+      "step": 2455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959887989434673e-05,
+      "loss": 2.1006,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959718072068068e-05,
+      "loss": 2.1923,
+      "step": 2465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959547797026163e-05,
+      "loss": 2.1518,
+      "step": 2470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95937716432124e-05,
+      "loss": 2.1224,
+      "step": 2475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9592061739656e-05,
+      "loss": 2.1161,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.959034825971577e-05,
+      "loss": 2.1312,
+      "step": 2485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958863120351527e-05,
+      "loss": 2.1401,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958691057117835e-05,
+      "loss": 2.1601,
+      "step": 2495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958518636282905e-05,
+      "loss": 2.1393,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958345857859177e-05,
+      "loss": 2.1581,
+      "step": 2505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.958172721859106e-05,
+      "loss": 2.1335,
+      "step": 2510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95799922829518e-05,
+      "loss": 2.1619,
+      "step": 2515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957825377179912e-05,
+      "loss": 2.1339,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957651168525837e-05,
+      "loss": 2.0766,
+      "step": 2525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957476602345519e-05,
+      "loss": 2.0608,
+      "step": 2530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95730167865155e-05,
+      "loss": 2.1515,
+      "step": 2535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.957126397456539e-05,
+      "loss": 2.1612,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956950758773131e-05,
+      "loss": 2.1514,
+      "step": 2545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956774762613991e-05,
+      "loss": 2.113,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95659840899181e-05,
+      "loss": 2.1598,
+      "step": 2555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956421697919309e-05,
+      "loss": 2.1765,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.95624462940923e-05,
+      "loss": 2.101,
+      "step": 2565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.956067203474342e-05,
+      "loss": 2.1009,
+      "step": 2570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955889420127441e-05,
+      "loss": 2.1385,
+      "step": 2575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955711279381348e-05,
+      "loss": 2.1381,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955532781248909e-05,
+      "loss": 2.0694,
+      "step": 2585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955353925742996e-05,
+      "loss": 2.178,
+      "step": 2590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955174712876509e-05,
+      "loss": 2.1152,
+      "step": 2595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954995142662372e-05,
+      "loss": 2.1133,
+      "step": 2600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954815215113535e-05,
+      "loss": 2.1903,
+      "step": 2605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954634930242973e-05,
+      "loss": 2.1287,
+      "step": 2610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954454288063686e-05,
+      "loss": 2.0863,
+      "step": 2615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954273288588703e-05,
+      "loss": 2.1838,
+      "step": 2620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.954091931831078e-05,
+      "loss": 2.1387,
+      "step": 2625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953910217803888e-05,
+      "loss": 2.1901,
+      "step": 2630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.953728146520237e-05,
+      "loss": 2.1405,
+      "step": 2635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953545717993258e-05,
+      "loss": 2.1159,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953362932236104e-05,
+      "loss": 2.1436,
+      "step": 2645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953179789261958e-05,
+      "loss": 2.1284,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952996289084028e-05,
+      "loss": 2.1216,
+      "step": 2655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952812431715547e-05,
+      "loss": 2.1364,
+      "step": 2660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952628217169773e-05,
+      "loss": 2.2046,
+      "step": 2665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952443645459992e-05,
+      "loss": 2.1742,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952258716599516e-05,
+      "loss": 2.1121,
+      "step": 2675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952073430601679e-05,
+      "loss": 2.1334,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951887787479843e-05,
+      "loss": 2.1537,
+      "step": 2685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951701787247397e-05,
+      "loss": 2.1606,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951515429917755e-05,
+      "loss": 2.1399,
+      "step": 2695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951328715504354e-05,
+      "loss": 2.1431,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951141644020662e-05,
+      "loss": 2.1468,
+      "step": 2705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950954215480168e-05,
+      "loss": 2.1318,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95076642989639e-05,
+      "loss": 2.1822,
+      "step": 2715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950578287282867e-05,
+      "loss": 2.1564,
+      "step": 2720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950389787653172e-05,
+      "loss": 2.1633,
+      "step": 2725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950200931020896e-05,
+      "loss": 2.1362,
+      "step": 2730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95001171739966e-05,
+      "loss": 2.1412,
+      "step": 2735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949822146803108e-05,
+      "loss": 2.1905,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949632219244911e-05,
+      "loss": 2.1356,
+      "step": 2745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949441934738766e-05,
+      "loss": 2.1061,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949251293298396e-05,
+      "loss": 2.1411,
+      "step": 2755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949060294937548e-05,
+      "loss": 2.1087,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94886893967e-05,
+      "loss": 2.1059,
+      "step": 2765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948677227509548e-05,
+      "loss": 2.0789,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948485158470018e-05,
+      "loss": 2.088,
+      "step": 2775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948292732565263e-05,
+      "loss": 2.1246,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948099949809158e-05,
+      "loss": 2.0963,
+      "step": 2785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947906810215606e-05,
+      "loss": 2.1017,
+      "step": 2790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947713313798537e-05,
+      "loss": 2.1449,
+      "step": 2795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947519460571904e-05,
+      "loss": 2.1365,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947325250549686e-05,
+      "loss": 2.0801,
+      "step": 2805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947130683745891e-05,
+      "loss": 2.1565,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946935760174549e-05,
+      "loss": 2.1282,
+      "step": 2815
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946740479849716e-05,
+      "loss": 2.1044,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94654484278548e-05,
+      "loss": 2.1527,
+      "step": 2825
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946348848995941e-05,
+      "loss": 2.128,
+      "step": 2830
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94615249849524e-05,
+      "loss": 2.1257,
+      "step": 2835
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945955791297536e-05,
+      "loss": 2.1591,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945758727417014e-05,
+      "loss": 2.1173,
+      "step": 2845
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945561306867883e-05,
+      "loss": 2.1048,
+      "step": 2850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945363529664385e-05,
+      "loss": 2.0935,
+      "step": 2855
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94516539582078e-05,
+      "loss": 2.1521,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944966905351355e-05,
+      "loss": 2.179,
+      "step": 2865
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944768058270429e-05,
+      "loss": 2.1949,
+      "step": 2870
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944568854592337e-05,
+      "loss": 2.0731,
+      "step": 2875
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944369294331448e-05,
+      "loss": 2.1413,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944169377502152e-05,
+      "loss": 2.1367,
+      "step": 2885
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943969104118869e-05,
+      "loss": 2.0686,
+      "step": 2890
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943768474196037e-05,
+      "loss": 2.0917,
+      "step": 2895
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94356748774813e-05,
+      "loss": 2.0917,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94336614478964e-05,
+      "loss": 2.1221,
+      "step": 2905
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.943164445335086e-05,
+      "loss": 2.1606,
+      "step": 2910
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942962389399014e-05,
+      "loss": 2.1821,
+      "step": 2915
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.942759976995997e-05,
+      "loss": 2.1088,
+      "step": 2920
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94255720814063e-05,
+      "loss": 2.1079,
+      "step": 2925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94235408284754e-05,
+      "loss": 2.1163,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.94215060113137e-05,
+      "loss": 2.1405,
+      "step": 2935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9419467630068e-05,
+      "loss": 2.1237,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941742568488524e-05,
+      "loss": 2.1176,
+      "step": 2945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941538017591272e-05,
+      "loss": 2.1693,
+      "step": 2950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941333110329794e-05,
+      "loss": 2.1085,
+      "step": 2955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941127846718868e-05,
+      "loss": 2.11,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940922226773297e-05,
+      "loss": 2.0851,
+      "step": 2965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940716250507906e-05,
+      "loss": 2.1118,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940509917937556e-05,
+      "loss": 2.1348,
+      "step": 2975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940303229077118e-05,
+      "loss": 2.0975,
+      "step": 2980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940096183941506e-05,
+      "loss": 2.1373,
+      "step": 2985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939888782545645e-05,
+      "loss": 2.1259,
+      "step": 2990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939681024904495e-05,
+      "loss": 2.1128,
+      "step": 2995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939472911033038e-05,
+      "loss": 2.1479,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939264440946283e-05,
+      "loss": 2.1264,
+      "step": 3005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939055614659265e-05,
+      "loss": 2.0299,
+      "step": 3010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938846432187041e-05,
+      "loss": 2.0895,
+      "step": 3015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938636893544697e-05,
+      "loss": 2.1338,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938426998747345e-05,
+      "loss": 2.0715,
+      "step": 3025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938216747810123e-05,
+      "loss": 2.1331,
+      "step": 3030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.938006140748191e-05,
+      "loss": 2.1247,
+      "step": 3035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937795177576739e-05,
+      "loss": 2.0845,
+      "step": 3040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937583858310979e-05,
+      "loss": 2.0935,
+      "step": 3045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937372182966154e-05,
+      "loss": 2.1038,
+      "step": 3050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937160151557525e-05,
+      "loss": 2.115,
+      "step": 3055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936947764100383e-05,
+      "loss": 2.116,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936735020610049e-05,
+      "loss": 2.1132,
+      "step": 3065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936521921101861e-05,
+      "loss": 2.1308,
+      "step": 3070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936308465591188e-05,
+      "loss": 2.1322,
+      "step": 3075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.936094654093425e-05,
+      "loss": 2.1447,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935880486623989e-05,
+      "loss": 2.104,
+      "step": 3085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935665963198329e-05,
+      "loss": 2.1237,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935451083831909e-05,
+      "loss": 2.1174,
+      "step": 3095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93523584854023e-05,
+      "loss": 2.145,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935020257338813e-05,
+      "loss": 2.1277,
+      "step": 3105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934804310243205e-05,
+      "loss": 2.0895,
+      "step": 3110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93458800726898e-05,
+      "loss": 2.1426,
+      "step": 3115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934371348431736e-05,
+      "loss": 2.1692,
+      "step": 3120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.934154333747099e-05,
+      "loss": 2.1011,
+      "step": 3125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933936963230719e-05,
+      "loss": 2.1114,
+      "step": 3130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.93371923689827e-05,
+      "loss": 2.1189,
+      "step": 3135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933501154765454e-05,
+      "loss": 2.0949,
+      "step": 3140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933282716848001e-05,
+      "loss": 2.0832,
+      "step": 3145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.933063923161661e-05,
+      "loss": 2.172,
+      "step": 3150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932844773722214e-05,
+      "loss": 2.1169,
+      "step": 3155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932625268545462e-05,
+      "loss": 2.1619,
+      "step": 3160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932405407647239e-05,
+      "loss": 2.1191,
+      "step": 3165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932185191043395e-05,
+      "loss": 2.139,
+      "step": 3170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931964618749817e-05,
+      "loss": 2.061,
+      "step": 3175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931743690782409e-05,
+      "loss": 2.1032,
+      "step": 3180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931522407157101e-05,
+      "loss": 2.1253,
+      "step": 3185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931300767889855e-05,
+      "loss": 2.1122,
+      "step": 3190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.931078772996654e-05,
+      "loss": 2.0913,
+      "step": 3195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930856422493505e-05,
+      "loss": 2.1008,
+      "step": 3200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930633716396446e-05,
+      "loss": 2.1259,
+      "step": 3205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930410654721536e-05,
+      "loss": 2.1013,
+      "step": 3210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.930187237484862e-05,
+      "loss": 2.1488,
+      "step": 3215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.929963464702535e-05,
+      "loss": 2.0659,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929739336390693e-05,
+      "loss": 2.1278,
+      "step": 3225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929514852565501e-05,
+      "loss": 2.0821,
+      "step": 3230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929290013243144e-05,
+      "loss": 2.1148,
+      "step": 3235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929064818439841e-05,
+      "loss": 2.1086,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928839268171831e-05,
+      "loss": 2.04,
+      "step": 3245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928613362455376e-05,
+      "loss": 2.1214,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928387101306773e-05,
+      "loss": 2.1616,
+      "step": 3255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928160484742335e-05,
+      "loss": 2.0804,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927933512778407e-05,
+      "loss": 2.1071,
+      "step": 3265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927706185431356e-05,
+      "loss": 2.0636,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927478502717578e-05,
+      "loss": 2.0651,
+      "step": 3275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927250464653488e-05,
+      "loss": 2.0628,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927022071255536e-05,
+      "loss": 2.0942,
+      "step": 3285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926793322540191e-05,
+      "loss": 2.1131,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926564218523948e-05,
+      "loss": 2.1172,
+      "step": 3295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926334759223332e-05,
+      "loss": 2.1516,
+      "step": 3300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926104944654888e-05,
+      "loss": 2.1073,
+      "step": 3305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925874774835192e-05,
+      "loss": 2.106,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925644249780841e-05,
+      "loss": 2.1323,
+      "step": 3315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92541336950846e-05,
+      "loss": 2.0967,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925182134034699e-05,
+      "loss": 2.126,
+      "step": 3325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924950543376233e-05,
+      "loss": 2.0924,
+      "step": 3330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924718597549766e-05,
+      "loss": 2.1606,
+      "step": 3335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924486296572023e-05,
+      "loss": 2.1428,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924253640459758e-05,
+      "loss": 2.1127,
+      "step": 3345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924020629229747e-05,
+      "loss": 2.1316,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923787262898795e-05,
+      "loss": 2.103,
+      "step": 3355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923553541483732e-05,
+      "loss": 2.0627,
+      "step": 3360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923319465001413e-05,
+      "loss": 2.1111,
+      "step": 3365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923085033468718e-05,
+      "loss": 2.1265,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922850246902553e-05,
+      "loss": 2.1644,
+      "step": 3375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922615105319853e-05,
+      "loss": 2.1097,
+      "step": 3380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92237960873757e-05,
+      "loss": 2.103,
+      "step": 3385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.92214375717269e-05,
+      "loss": 2.1819,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921907550642224e-05,
+      "loss": 2.1779,
+      "step": 3395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921670989163203e-05,
+      "loss": 2.1498,
+      "step": 3400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921434072752688e-05,
+      "loss": 2.1033,
+      "step": 3405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921196801427766e-05,
+      "loss": 2.13,
+      "step": 3410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920959175205544e-05,
+      "loss": 2.0894,
+      "step": 3415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920721194103162e-05,
+      "loss": 2.0644,
+      "step": 3420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920482858137782e-05,
+      "loss": 2.134,
+      "step": 3425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920244167326592e-05,
+      "loss": 2.0577,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920005121686801e-05,
+      "loss": 2.0884,
+      "step": 3435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919765721235654e-05,
+      "loss": 2.1157,
+      "step": 3440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919525965990415e-05,
+      "loss": 2.1892,
+      "step": 3445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91928585596837e-05,
+      "loss": 2.1232,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919045391186838e-05,
+      "loss": 2.1417,
+      "step": 3455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918804571663161e-05,
+      "loss": 2.0865,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918563397414704e-05,
+      "loss": 2.1401,
+      "step": 3465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91832186845886e-05,
+      "loss": 2.0901,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918079984813047e-05,
+      "loss": 2.1096,
+      "step": 3475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91783774649471e-05,
+      "loss": 2.1246,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917595153521316e-05,
+      "loss": 2.1007,
+      "step": 3485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917352205910361e-05,
+      "loss": 2.1226,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.917108903679366e-05,
+      "loss": 2.0895,
+      "step": 3495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916865246845876e-05,
+      "loss": 2.0825,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916621235427463e-05,
+      "loss": 2.0608,
+      "step": 3505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916376869441725e-05,
+      "loss": 2.1358,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.916132148906285e-05,
+      "loss": 2.1015,
+      "step": 3515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91588707383879e-05,
+      "loss": 2.0937,
+      "step": 3520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915641644256912e-05,
+      "loss": 2.0298,
+      "step": 3525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.915395860178354e-05,
+      "loss": 2.0844,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91514972162084e-05,
+      "loss": 2.0589,
+      "step": 3535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.91490322860212e-05,
+      "loss": 2.1388,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914656381139968e-05,
+      "loss": 2.1043,
+      "step": 3545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914409179252189e-05,
+      "loss": 2.1151,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.914161622956609e-05,
+      "loss": 2.0747,
+      "step": 3555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913913712271081e-05,
+      "loss": 2.1258,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913665447213483e-05,
+      "loss": 2.0782,
+      "step": 3565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913416827801719e-05,
+      "loss": 2.1114,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.913167854053719e-05,
+      "loss": 2.1045,
+      "step": 3575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912918525987437e-05,
+      "loss": 2.0996,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912668843620854e-05,
+      "loss": 2.1282,
+      "step": 3585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912418806971976e-05,
+      "loss": 2.0321,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.912168416058836e-05,
+      "loss": 2.0867,
+      "step": 3595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911917670899489e-05,
+      "loss": 2.1568,
+      "step": 3600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911666571512017e-05,
+      "loss": 2.1099,
+      "step": 3605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911415117914533e-05,
+      "loss": 2.1127,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.911163310125166e-05,
+      "loss": 2.0448,
+      "step": 3615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910911148162079e-05,
+      "loss": 2.0636,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910658632043454e-05,
+      "loss": 2.0925,
+      "step": 3625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910405761787502e-05,
+      "loss": 2.1188,
+      "step": 3630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.910152537412462e-05,
+      "loss": 2.1055,
+      "step": 3635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909898958936592e-05,
+      "loss": 2.1194,
+      "step": 3640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909645026378179e-05,
+      "loss": 2.1307,
+      "step": 3645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909390739755539e-05,
+      "loss": 2.0859,
+      "step": 3650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.909136099087007e-05,
+      "loss": 2.087,
+      "step": 3655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908881104390948e-05,
+      "loss": 2.1228,
+      "step": 3660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90862575568575e-05,
+      "loss": 2.1213,
+      "step": 3665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90837005298983e-05,
+      "loss": 2.0373,
+      "step": 3670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.908113996321626e-05,
+      "loss": 2.1124,
+      "step": 3675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907857585699606e-05,
+      "loss": 2.076,
+      "step": 3680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907600821142259e-05,
+      "loss": 2.0294,
+      "step": 3685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907343702668104e-05,
+      "loss": 2.1015,
+      "step": 3690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.907086230295681e-05,
+      "loss": 2.1164,
+      "step": 3695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90682840404356e-05,
+      "loss": 2.1444,
+      "step": 3700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906570223930334e-05,
+      "loss": 2.088,
+      "step": 3705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90631168997462e-05,
+      "loss": 2.0627,
+      "step": 3710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.906052802195064e-05,
+      "loss": 2.1493,
+      "step": 3715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905793560610339e-05,
+      "loss": 2.0805,
+      "step": 3720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905533965239133e-05,
+      "loss": 2.1042,
+      "step": 3725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905274016100174e-05,
+      "loss": 2.1208,
+      "step": 3730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.905013713212207e-05,
+      "loss": 2.1181,
+      "step": 3735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904753056594001e-05,
+      "loss": 2.1393,
+      "step": 3740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904492046264356e-05,
+      "loss": 2.1191,
+      "step": 3745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.904230682242095e-05,
+      "loss": 2.1018,
+      "step": 3750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903968964546065e-05,
+      "loss": 2.1048,
+      "step": 3755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903706893195142e-05,
+      "loss": 2.049,
+      "step": 3760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903444468208225e-05,
+      "loss": 2.078,
+      "step": 3765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.903181689604237e-05,
+      "loss": 2.0768,
+      "step": 3770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902918557402132e-05,
+      "loss": 2.137,
+      "step": 3775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902655071620884e-05,
+      "loss": 2.1563,
+      "step": 3780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902391232279494e-05,
+      "loss": 2.1422,
+      "step": 3785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.902127039396991e-05,
+      "loss": 2.0906,
+      "step": 3790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901862492992426e-05,
+      "loss": 2.0548,
+      "step": 3795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.901597593084878e-05,
+      "loss": 2.0966,
+      "step": 3800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.90133233969345e-05,
+      "loss": 2.1482,
+      "step": 3805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90106673283727e-05,
+      "loss": 2.0784,
+      "step": 3810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900800772535494e-05,
+      "loss": 2.1077,
+      "step": 3815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900534458807302e-05,
+      "loss": 2.0572,
+      "step": 3820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900267791671899e-05,
+      "loss": 2.0918,
+      "step": 3825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900000771148517e-05,
+      "loss": 2.1121,
+      "step": 3830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89973339725641e-05,
+      "loss": 2.0771,
+      "step": 3835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899465670014863e-05,
+      "loss": 2.1562,
+      "step": 3840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89919758944318e-05,
+      "loss": 2.0865,
+      "step": 3845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898929155560697e-05,
+      "loss": 2.054,
+      "step": 3850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898660368386771e-05,
+      "loss": 2.1253,
+      "step": 3855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898391227940787e-05,
+      "loss": 2.0875,
+      "step": 3860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898121734242152e-05,
+      "loss": 2.1321,
+      "step": 3865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897851887310303e-05,
+      "loss": 2.0586,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897581687164699e-05,
+      "loss": 2.1012,
+      "step": 3875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897311133824828e-05,
+      "loss": 2.0498,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897040227310198e-05,
+      "loss": 2.1135,
+      "step": 3885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896768967640349e-05,
+      "loss": 2.0536,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896497354834839e-05,
+      "loss": 2.1035,
+      "step": 3895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89622538891326e-05,
+      "loss": 2.0852,
+      "step": 3900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895953069895223e-05,
+      "loss": 2.0906,
+      "step": 3905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895680397800366e-05,
+      "loss": 2.0876,
+      "step": 3910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895407372648354e-05,
+      "loss": 2.0978,
+      "step": 3915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.895133994458876e-05,
+      "loss": 2.0506,
+      "step": 3920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894860263251648e-05,
+      "loss": 2.0509,
+      "step": 3925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89458617904641e-05,
+      "loss": 2.1256,
+      "step": 3930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894311741862927e-05,
+      "loss": 2.0696,
+      "step": 3935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89403695172099e-05,
+      "loss": 2.0845,
+      "step": 3940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893761808640418e-05,
+      "loss": 2.0803,
+      "step": 3945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.893486312641053e-05,
+      "loss": 2.0575,
+      "step": 3950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89321046374276e-05,
+      "loss": 2.0831,
+      "step": 3955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892934261965434e-05,
+      "loss": 2.0562,
+      "step": 3960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892657707328995e-05,
+      "loss": 2.0721,
+      "step": 3965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892380799853382e-05,
+      "loss": 2.055,
+      "step": 3970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89210353955857e-05,
+      "loss": 2.1443,
+      "step": 3975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891825926464551e-05,
+      "loss": 2.0979,
+      "step": 3980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891547960591347e-05,
+      "loss": 2.117,
+      "step": 3985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.891269641959001e-05,
+      "loss": 2.0555,
+      "step": 3990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890990970587588e-05,
+      "loss": 2.1012,
+      "step": 3995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890711946497202e-05,
+      "loss": 2.0747,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890432569707966e-05,
+      "loss": 2.0869,
+      "step": 4005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.890152840240028e-05,
+      "loss": 2.0957,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88987275811356e-05,
+      "loss": 2.0838,
+      "step": 4015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88959232334876e-05,
+      "loss": 2.114,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889311535965856e-05,
+      "loss": 2.07,
+      "step": 4025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889030395985091e-05,
+      "loss": 2.0671,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888748903426743e-05,
+      "loss": 2.0979,
+      "step": 4035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888467058311113e-05,
+      "loss": 2.1169,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888184860658525e-05,
+      "loss": 2.0982,
+      "step": 4045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88790231048933e-05,
+      "loss": 2.0506,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887619407823906e-05,
+      "loss": 2.1098,
+      "step": 4055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887336152682652e-05,
+      "loss": 2.0968,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.887052545085997e-05,
+      "loss": 2.1259,
+      "step": 4065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.886768585054393e-05,
+      "loss": 2.0693,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88648427260832e-05,
+      "loss": 2.1149,
+      "step": 4075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88619960776828e-05,
+      "loss": 2.0579,
+      "step": 4080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885914590554802e-05,
+      "loss": 2.1207,
+      "step": 4085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88562922098844e-05,
+      "loss": 2.0936,
+      "step": 4090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.885343499089774e-05,
+      "loss": 2.0415,
+      "step": 4095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.88505742487941e-05,
+      "loss": 2.0739,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884770998377977e-05,
+      "loss": 2.0252,
+      "step": 4105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884484219606132e-05,
+      "loss": 2.1207,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.884197088584555e-05,
+      "loss": 2.0914,
+      "step": 4115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883909605333954e-05,
+      "loss": 2.0389,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883621769875063e-05,
+      "loss": 2.0667,
+      "step": 4125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883333582228635e-05,
+      "loss": 2.1074,
+      "step": 4130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.883045042415456e-05,
+      "loss": 2.0857,
+      "step": 4135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882756150456335e-05,
+      "loss": 2.0628,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882466906372103e-05,
+      "loss": 2.0632,
+      "step": 4145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.882177310183622e-05,
+      "loss": 2.0689,
+      "step": 4150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881887361911775e-05,
+      "loss": 2.0947,
+      "step": 4155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881597061577472e-05,
+      "loss": 2.1169,
+      "step": 4160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881306409201649e-05,
+      "loss": 2.0678,
+      "step": 4165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.881015404805265e-05,
+      "loss": 2.0806,
+      "step": 4170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880724048409308e-05,
+      "loss": 2.0671,
+      "step": 4175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880432340034788e-05,
+      "loss": 2.0329,
+      "step": 4180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.880140279702743e-05,
+      "loss": 2.0929,
+      "step": 4185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879847867434235e-05,
+      "loss": 2.0295,
+      "step": 4190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879555103250353e-05,
+      "loss": 2.0796,
+      "step": 4195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.879261987172205e-05,
+      "loss": 2.0811,
+      "step": 4200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878968519220937e-05,
+      "loss": 2.0585,
+      "step": 4205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878674699417706e-05,
+      "loss": 2.1367,
+      "step": 4210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878380527783704e-05,
+      "loss": 2.0796,
+      "step": 4215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.878086004340146e-05,
+      "loss": 2.0414,
+      "step": 4220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87779112910827e-05,
+      "loss": 2.0953,
+      "step": 4225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877495902109343e-05,
+      "loss": 2.0815,
+      "step": 4230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.877200323364655e-05,
+      "loss": 2.0847,
+      "step": 4235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87690439289552e-05,
+      "loss": 2.0578,
+      "step": 4240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876608110723284e-05,
+      "loss": 2.0881,
+      "step": 4245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.876311476869309e-05,
+      "loss": 2.0536,
+      "step": 4250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87601449135499e-05,
+      "loss": 2.1104,
+      "step": 4255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875717154201742e-05,
+      "loss": 2.1024,
+      "step": 4260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875419465431011e-05,
+      "loss": 2.0731,
+      "step": 4265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.875121425064264e-05,
+      "loss": 2.1172,
+      "step": 4270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874823033122992e-05,
+      "loss": 2.0772,
+      "step": 4275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874524289628715e-05,
+      "loss": 2.0337,
+      "step": 4280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.874225194602979e-05,
+      "loss": 2.0579,
+      "step": 4285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873925748067355e-05,
+      "loss": 2.0984,
+      "step": 4290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873625950043431e-05,
+      "loss": 2.0921,
+      "step": 4295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873325800552834e-05,
+      "loss": 2.0985,
+      "step": 4300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.873025299617207e-05,
+      "loss": 2.0698,
+      "step": 4305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87272444725822e-05,
+      "loss": 2.0625,
+      "step": 4310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872423243497571e-05,
+      "loss": 2.0645,
+      "step": 4315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.872121688356981e-05,
+      "loss": 2.1228,
+      "step": 4320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871819781858198e-05,
+      "loss": 2.1235,
+      "step": 4325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871517524022993e-05,
+      "loss": 2.1366,
+      "step": 4330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.871214914873164e-05,
+      "loss": 2.0914,
+      "step": 4335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870911954430533e-05,
+      "loss": 2.0619,
+      "step": 4340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.87060864271695e-05,
+      "loss": 2.0766,
+      "step": 4345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870304979754287e-05,
+      "loss": 2.0988,
+      "step": 4350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.870000965564445e-05,
+      "loss": 2.0721,
+      "step": 4355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869696600169346e-05,
+      "loss": 2.0433,
+      "step": 4360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869391883590942e-05,
+      "loss": 2.1149,
+      "step": 4365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.869086815851204e-05,
+      "loss": 2.0379,
+      "step": 4370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86878139697214e-05,
+      "loss": 2.0848,
+      "step": 4375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.868475626975767e-05,
+      "loss": 2.1087,
+      "step": 4380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.86816950588414e-05,
+      "loss": 2.104,
+      "step": 4385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.867863033719335e-05,
+      "loss": 2.0602,
+      "step": 4390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867556210503454e-05,
+      "loss": 2.078,
+      "step": 4395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867249036258624e-05,
+      "loss": 2.0745,
+      "step": 4400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866941511006995e-05,
+      "loss": 2.1183,
+      "step": 4405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866633634770746e-05,
+      "loss": 2.1,
+      "step": 4410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866325407572082e-05,
+      "loss": 2.0555,
+      "step": 4415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866016829433227e-05,
+      "loss": 2.0842,
+      "step": 4420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865707900376435e-05,
+      "loss": 2.0326,
+      "step": 4425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865398620423987e-05,
+      "loss": 2.0918,
+      "step": 4430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865088989598188e-05,
+      "loss": 2.0953,
+      "step": 4435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864779007921363e-05,
+      "loss": 1.9992,
+      "step": 4440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864468675415871e-05,
+      "loss": 2.0443,
+      "step": 4445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864157992104091e-05,
+      "loss": 2.0972,
+      "step": 4450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863846958008425e-05,
+      "loss": 2.0473,
+      "step": 4455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863535573151306e-05,
+      "loss": 2.0255,
+      "step": 4460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863223837555191e-05,
+      "loss": 2.1009,
+      "step": 4465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862911751242559e-05,
+      "loss": 2.0741,
+      "step": 4470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.862599314235919e-05,
+      "loss": 2.0725,
+      "step": 4475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8622865265578e-05,
+      "loss": 2.0837,
+      "step": 4480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861973388230758e-05,
+      "loss": 2.0765,
+      "step": 4485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86165989927738e-05,
+      "loss": 2.0651,
+      "step": 4490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861346059720271e-05,
+      "loss": 2.0277,
+      "step": 4495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861031869582062e-05,
+      "loss": 2.1266,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860717328885415e-05,
+      "loss": 2.0748,
+      "step": 4505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86040243765301e-05,
+      "loss": 2.0839,
+      "step": 4510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860087195907558e-05,
+      "loss": 2.1268,
+      "step": 4515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.859771603671792e-05,
+      "loss": 2.1157,
+      "step": 4520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85945566096847e-05,
+      "loss": 2.0943,
+      "step": 4525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85913936782038e-05,
+      "loss": 2.0586,
+      "step": 4530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85882272425033e-05,
+      "loss": 2.0232,
+      "step": 4535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858505730281154e-05,
+      "loss": 2.0532,
+      "step": 4540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858188385935712e-05,
+      "loss": 2.1296,
+      "step": 4545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857870691236891e-05,
+      "loss": 2.079,
+      "step": 4550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857552646207603e-05,
+      "loss": 2.0738,
+      "step": 4555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857234250870781e-05,
+      "loss": 2.0469,
+      "step": 4560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856915505249391e-05,
+      "loss": 2.1151,
+      "step": 4565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856596409366416e-05,
+      "loss": 2.0731,
+      "step": 4570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.856276963244868e-05,
+      "loss": 2.1258,
+      "step": 4575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855957166907786e-05,
+      "loss": 2.1381,
+      "step": 4580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85563702037823e-05,
+      "loss": 2.0553,
+      "step": 4585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855316523679289e-05,
+      "loss": 2.1102,
+      "step": 4590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854995676834078e-05,
+      "loss": 2.0515,
+      "step": 4595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854674479865733e-05,
+      "loss": 2.0814,
+      "step": 4600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854352932797417e-05,
+      "loss": 2.0644,
+      "step": 4605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854031035652321e-05,
+      "loss": 2.066,
+      "step": 4610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853708788453657e-05,
+      "loss": 2.0653,
+      "step": 4615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853386191224665e-05,
+      "loss": 2.069,
+      "step": 4620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.853063243988607e-05,
+      "loss": 2.0239,
+      "step": 4625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852739946768777e-05,
+      "loss": 2.0639,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852416299588488e-05,
+      "loss": 2.0637,
+      "step": 4635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85209230247108e-05,
+      "loss": 2.0757,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851767955439915e-05,
+      "loss": 2.0617,
+      "step": 4645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.85144325851839e-05,
+      "loss": 2.0824,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.851118211729917e-05,
+      "loss": 2.0918,
+      "step": 4655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850792815097938e-05,
+      "loss": 2.0602,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850467068645919e-05,
+      "loss": 2.0783,
+      "step": 4665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.850140972397352e-05,
+      "loss": 2.064,
+      "step": 4670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849814526375753e-05,
+      "loss": 2.0618,
+      "step": 4675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849487730604665e-05,
+      "loss": 2.0647,
+      "step": 4680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.849160585107656e-05,
+      "loss": 2.0538,
+      "step": 4685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848833089908315e-05,
+      "loss": 2.0413,
+      "step": 4690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848505245030262e-05,
+      "loss": 2.0285,
+      "step": 4695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.848177050497141e-05,
+      "loss": 2.1046,
+      "step": 4700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847848506332619e-05,
+      "loss": 2.0656,
+      "step": 4705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847519612560387e-05,
+      "loss": 2.0782,
+      "step": 4710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.847190369204169e-05,
+      "loss": 2.0665,
+      "step": 4715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846860776287703e-05,
+      "loss": 2.0917,
+      "step": 4720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846530833834761e-05,
+      "loss": 2.0692,
+      "step": 4725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.846200541869135e-05,
+      "loss": 2.0483,
+      "step": 4730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845869900414649e-05,
+      "loss": 2.088,
+      "step": 4735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845538909495143e-05,
+      "loss": 2.047,
+      "step": 4740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.845207569134487e-05,
+      "loss": 2.0425,
+      "step": 4745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844875879356577e-05,
+      "loss": 2.0811,
+      "step": 4750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844543840185334e-05,
+      "loss": 2.0947,
+      "step": 4755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.844211451644703e-05,
+      "loss": 2.0803,
+      "step": 4760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843878713758654e-05,
+      "loss": 2.0746,
+      "step": 4765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.843545626551182e-05,
+      "loss": 2.0639,
+      "step": 4770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.84321219004631e-05,
+      "loss": 2.054,
+      "step": 4775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842878404268082e-05,
+      "loss": 2.0461,
+      "step": 4780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842544269240572e-05,
+      "loss": 2.0548,
+      "step": 4785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.842209784987873e-05,
+      "loss": 2.0379,
+      "step": 4790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841874951534109e-05,
+      "loss": 2.0298,
+      "step": 4795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841539768903426e-05,
+      "loss": 1.9793,
+      "step": 4800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.841204237119997e-05,
+      "loss": 2.0636,
+      "step": 4805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840868356208019e-05,
+      "loss": 2.0656,
+      "step": 4810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840532126191712e-05,
+      "loss": 2.1156,
+      "step": 4815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.840195547095327e-05,
+      "loss": 2.0524,
+      "step": 4820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839858618943136e-05,
+      "loss": 2.0767,
+      "step": 4825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839521341759437e-05,
+      "loss": 2.057,
+      "step": 4830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.839183715568551e-05,
+      "loss": 2.1058,
+      "step": 4835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838845740394826e-05,
+      "loss": 2.0753,
+      "step": 4840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83850741626264e-05,
+      "loss": 2.0309,
+      "step": 4845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838168743196387e-05,
+      "loss": 2.0672,
+      "step": 4850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837829721220493e-05,
+      "loss": 2.0773,
+      "step": 4855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837490350359406e-05,
+      "loss": 2.0419,
+      "step": 4860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.837150630637601e-05,
+      "loss": 2.0845,
+      "step": 4865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836810562079576e-05,
+      "loss": 2.0028,
+      "step": 4870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836470144709859e-05,
+      "loss": 2.0464,
+      "step": 4875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.836129378552994e-05,
+      "loss": 2.0264,
+      "step": 4880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835788263633558e-05,
+      "loss": 2.0601,
+      "step": 4885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835446799976153e-05,
+      "loss": 2.0791,
+      "step": 4890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.835104987605402e-05,
+      "loss": 2.1123,
+      "step": 4895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834762826545954e-05,
+      "loss": 2.0266,
+      "step": 4900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834420316822486e-05,
+      "loss": 2.1171,
+      "step": 4905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.834077458459698e-05,
+      "loss": 2.0514,
+      "step": 4910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833734251482319e-05,
+      "loss": 2.0762,
+      "step": 4915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.833390695915092e-05,
+      "loss": 2.0827,
+      "step": 4920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8330467917828e-05,
+      "loss": 2.076,
+      "step": 4925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832702539110241e-05,
+      "loss": 2.0987,
+      "step": 4930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832357937922241e-05,
+      "loss": 2.1042,
+      "step": 4935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.832012988243652e-05,
+      "loss": 2.0871,
+      "step": 4940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831667690099351e-05,
+      "loss": 2.0798,
+      "step": 4945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.831322043514237e-05,
+      "loss": 2.0846,
+      "step": 4950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83097604851324e-05,
+      "loss": 2.0933,
+      "step": 4955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.83062970512131e-05,
+      "loss": 2.0124,
+      "step": 4960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.830283013363423e-05,
+      "loss": 2.0195,
+      "step": 4965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829935973264584e-05,
+      "loss": 2.0492,
+      "step": 4970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829588584849817e-05,
+      "loss": 2.1006,
+      "step": 4975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.829240848144177e-05,
+      "loss": 2.0632,
+      "step": 4980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828892763172738e-05,
+      "loss": 2.0883,
+      "step": 4985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828544329960605e-05,
+      "loss": 2.0479,
+      "step": 4990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828195548532906e-05,
+      "loss": 2.0447,
+      "step": 4995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827846418914793e-05,
+      "loss": 2.0215,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827496941131445e-05,
+      "loss": 2.0578,
+      "step": 5005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82714711520806e-05,
+      "loss": 2.0895,
+      "step": 5010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826796941169874e-05,
+      "loss": 2.0461,
+      "step": 5015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826446419042132e-05,
+      "loss": 2.0568,
+      "step": 5020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826095548850118e-05,
+      "loss": 2.0794,
+      "step": 5025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825744330619135e-05,
+      "loss": 2.0633,
+      "step": 5030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825392764374509e-05,
+      "loss": 2.03,
+      "step": 5035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825040850141594e-05,
+      "loss": 2.0754,
+      "step": 5040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824688587945769e-05,
+      "loss": 2.0916,
+      "step": 5045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82433597781244e-05,
+      "loss": 2.0616,
+      "step": 5050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823983019767034e-05,
+      "loss": 2.0941,
+      "step": 5055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823629713835003e-05,
+      "loss": 2.0874,
+      "step": 5060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.823276060041828e-05,
+      "loss": 2.1104,
+      "step": 5065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822922058413013e-05,
+      "loss": 2.0586,
+      "step": 5070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822567708974089e-05,
+      "loss": 2.0807,
+      "step": 5075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822213011750607e-05,
+      "loss": 2.0754,
+      "step": 5080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821857966768149e-05,
+      "loss": 2.0627,
+      "step": 5085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821502574052316e-05,
+      "loss": 2.0579,
+      "step": 5090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.821146833628742e-05,
+      "loss": 2.0103,
+      "step": 5095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820790745523078e-05,
+      "loss": 2.0519,
+      "step": 5100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820434309761007e-05,
+      "loss": 2.0566,
+      "step": 5105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82007752636823e-05,
+      "loss": 2.0642,
+      "step": 5110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819720395370479e-05,
+      "loss": 2.0105,
+      "step": 5115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819362916793509e-05,
+      "loss": 2.0454,
+      "step": 5120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819005090663101e-05,
+      "loss": 2.1016,
+      "step": 5125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818646917005056e-05,
+      "loss": 2.0428,
+      "step": 5130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.818288395845207e-05,
+      "loss": 2.069,
+      "step": 5135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817929527209408e-05,
+      "loss": 2.0689,
+      "step": 5140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817570311123541e-05,
+      "loss": 2.0695,
+      "step": 5145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81721074761351e-05,
+      "loss": 2.0544,
+      "step": 5150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816850836705246e-05,
+      "loss": 2.0256,
+      "step": 5155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816490578424704e-05,
+      "loss": 2.039,
+      "step": 5160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.816129972797863e-05,
+      "loss": 2.0815,
+      "step": 5165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815769019850729e-05,
+      "loss": 2.0551,
+      "step": 5170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815407719609336e-05,
+      "loss": 2.0178,
+      "step": 5175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815046072099734e-05,
+      "loss": 2.0524,
+      "step": 5180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.814684077348006e-05,
+      "loss": 2.0682,
+      "step": 5185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81432173538026e-05,
+      "loss": 2.0964,
+      "step": 5190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813959046222622e-05,
+      "loss": 2.048,
+      "step": 5195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.813596009901253e-05,
+      "loss": 2.044,
+      "step": 5200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81323262644233e-05,
+      "loss": 2.014,
+      "step": 5205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812868895872059e-05,
+      "loss": 2.0314,
+      "step": 5210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812504818216673e-05,
+      "loss": 2.0811,
+      "step": 5215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812140393502425e-05,
+      "loss": 2.029,
+      "step": 5220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811775621755599e-05,
+      "loss": 2.0335,
+      "step": 5225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811410503002498e-05,
+      "loss": 2.0285,
+      "step": 5230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.811045037269455e-05,
+      "loss": 2.0544,
+      "step": 5235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810679224582825e-05,
+      "loss": 1.9827,
+      "step": 5240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810313064968988e-05,
+      "loss": 2.0322,
+      "step": 5245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809946558454353e-05,
+      "loss": 2.0752,
+      "step": 5250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809579705065348e-05,
+      "loss": 2.0908,
+      "step": 5255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.809212504828429e-05,
+      "loss": 2.0605,
+      "step": 5260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80884495777008e-05,
+      "loss": 2.0546,
+      "step": 5265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808477063916804e-05,
+      "loss": 2.0644,
+      "step": 5270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.808108823295135e-05,
+      "loss": 2.0247,
+      "step": 5275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807740235931625e-05,
+      "loss": 2.0637,
+      "step": 5280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.807371301852857e-05,
+      "loss": 2.0926,
+      "step": 5285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80700202108544e-05,
+      "loss": 2.0291,
+      "step": 5290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806632393656001e-05,
+      "loss": 2.0441,
+      "step": 5295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.806262419591198e-05,
+      "loss": 2.0867,
+      "step": 5300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805892098917711e-05,
+      "loss": 2.056,
+      "step": 5305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805521431662248e-05,
+      "loss": 2.1321,
+      "step": 5310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.805150417851536e-05,
+      "loss": 2.0388,
+      "step": 5315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804779057512337e-05,
+      "loss": 2.0863,
+      "step": 5320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804407350671428e-05,
+      "loss": 2.0555,
+      "step": 5325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.804035297355615e-05,
+      "loss": 2.022,
+      "step": 5330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80366289759173e-05,
+      "loss": 2.0551,
+      "step": 5335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.80329015140663e-05,
+      "loss": 2.0988,
+      "step": 5340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802917058827194e-05,
+      "loss": 2.0916,
+      "step": 5345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802543619880328e-05,
+      "loss": 2.0666,
+      "step": 5350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.802169834592964e-05,
+      "loss": 2.0263,
+      "step": 5355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801795702992061e-05,
+      "loss": 2.0461,
+      "step": 5360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801421225104592e-05,
+      "loss": 2.0731,
+      "step": 5365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.801046400957571e-05,
+      "loss": 2.0471,
+      "step": 5370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800671230578023e-05,
+      "loss": 2.0333,
+      "step": 5375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.800295713993009e-05,
+      "loss": 2.1004,
+      "step": 5380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799919851229605e-05,
+      "loss": 2.031,
+      "step": 5385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.799543642314917e-05,
+      "loss": 2.0498,
+      "step": 5390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79916708727608e-05,
+      "loss": 2.0887,
+      "step": 5395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798790186140245e-05,
+      "loss": 2.0673,
+      "step": 5400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798412938934596e-05,
+      "loss": 2.0937,
+      "step": 5405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.798035345686337e-05,
+      "loss": 2.061,
+      "step": 5410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797657406422698e-05,
+      "loss": 2.0916,
+      "step": 5415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.797279121170937e-05,
+      "loss": 2.0535,
+      "step": 5420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796900489958331e-05,
+      "loss": 2.0141,
+      "step": 5425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796521512812187e-05,
+      "loss": 2.0351,
+      "step": 5430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.796142189759837e-05,
+      "loss": 2.0464,
+      "step": 5435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795762520828633e-05,
+      "loss": 2.0471,
+      "step": 5440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795382506045958e-05,
+      "loss": 2.0657,
+      "step": 5445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.795002145439216e-05,
+      "loss": 2.0258,
+      "step": 5450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794621439035836e-05,
+      "loss": 2.0254,
+      "step": 5455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.794240386863276e-05,
+      "loss": 2.0419,
+      "step": 5460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793858988949013e-05,
+      "loss": 2.0801,
+      "step": 5465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793477245320553e-05,
+      "loss": 2.1151,
+      "step": 5470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.793095156005427e-05,
+      "loss": 2.0428,
+      "step": 5475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792712721031188e-05,
+      "loss": 2.0819,
+      "step": 5480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792329940425416e-05,
+      "loss": 2.0592,
+      "step": 5485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791946814215717e-05,
+      "loss": 2.0997,
+      "step": 5490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.79156334242972e-05,
+      "loss": 2.0208,
+      "step": 5495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.791179525095078e-05,
+      "loss": 1.9978,
+      "step": 5500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790795362239471e-05,
+      "loss": 2.0556,
+      "step": 5505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790410853890604e-05,
+      "loss": 2.0757,
+      "step": 5510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.790026000076208e-05,
+      "loss": 2.069,
+      "step": 5515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789640800824034e-05,
+      "loss": 2.0499,
+      "step": 5520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.789255256161862e-05,
+      "loss": 2.0797,
+      "step": 5525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788869366117496e-05,
+      "loss": 2.0414,
+      "step": 5530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788483130718764e-05,
+      "loss": 2.0831,
+      "step": 5535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.788096549993523e-05,
+      "loss": 2.0505,
+      "step": 5540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787709623969648e-05,
+      "loss": 1.9859,
+      "step": 5545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.787322352675046e-05,
+      "loss": 2.0628,
+      "step": 5550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78693473613764e-05,
+      "loss": 2.0875,
+      "step": 5555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78654677438539e-05,
+      "loss": 2.0296,
+      "step": 5560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.78615846744627e-05,
+      "loss": 2.0478,
+      "step": 5565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785769815348284e-05,
+      "loss": 2.0271,
+      "step": 5570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785380818119461e-05,
+      "loss": 2.0808,
+      "step": 5575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784991475787855e-05,
+      "loss": 2.0657,
+      "step": 5580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78460178838154e-05,
+      "loss": 2.0303,
+      "step": 5585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78421175592862e-05,
+      "loss": 2.0739,
+      "step": 5590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783821378457225e-05,
+      "loss": 2.072,
+      "step": 5595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783430655995507e-05,
+      "loss": 2.0592,
+      "step": 5600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783039588571641e-05,
+      "loss": 2.0503,
+      "step": 5605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782648176213832e-05,
+      "loss": 1.9888,
+      "step": 5610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782256418950304e-05,
+      "loss": 2.0445,
+      "step": 5615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78186431680931e-05,
+      "loss": 2.0723,
+      "step": 5620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78147186981913e-05,
+      "loss": 2.0383,
+      "step": 5625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.781079078008062e-05,
+      "loss": 2.0304,
+      "step": 5630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780685941404434e-05,
+      "loss": 2.0834,
+      "step": 5635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780292460036598e-05,
+      "loss": 2.0696,
+      "step": 5640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779898633932928e-05,
+      "loss": 2.1164,
+      "step": 5645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779504463121827e-05,
+      "loss": 2.0283,
+      "step": 5650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.779109947631721e-05,
+      "loss": 2.0388,
+      "step": 5655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778715087491063e-05,
+      "loss": 2.0417,
+      "step": 5660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778319882728325e-05,
+      "loss": 2.0141,
+      "step": 5665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777924333372008e-05,
+      "loss": 2.0723,
+      "step": 5670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77752843945064e-05,
+      "loss": 2.0372,
+      "step": 5675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.777132200992771e-05,
+      "loss": 2.1086,
+      "step": 5680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776735618026973e-05,
+      "loss": 2.0246,
+      "step": 5685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77633869058185e-05,
+      "loss": 2.0957,
+      "step": 5690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775941418686023e-05,
+      "loss": 2.0855,
+      "step": 5695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775543802368144e-05,
+      "loss": 2.0659,
+      "step": 5700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.775145841656887e-05,
+      "loss": 2.0406,
+      "step": 5705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774747536580953e-05,
+      "loss": 2.0568,
+      "step": 5710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774348887169064e-05,
+      "loss": 2.0292,
+      "step": 5715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773949893449969e-05,
+      "loss": 2.0571,
+      "step": 5720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773550555452442e-05,
+      "loss": 2.0433,
+      "step": 5725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.773150873205284e-05,
+      "loss": 2.0426,
+      "step": 5730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772750846737315e-05,
+      "loss": 2.0345,
+      "step": 5735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772350476077387e-05,
+      "loss": 2.0601,
+      "step": 5740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.77194976125437e-05,
+      "loss": 2.016,
+      "step": 5745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771548702297163e-05,
+      "loss": 2.0325,
+      "step": 5750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.771147299234688e-05,
+      "loss": 2.0413,
+      "step": 5755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770745552095897e-05,
+      "loss": 2.0332,
+      "step": 5760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770343460909757e-05,
+      "loss": 2.0194,
+      "step": 5765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769941025705266e-05,
+      "loss": 2.0416,
+      "step": 5770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769538246511448e-05,
+      "loss": 2.0764,
+      "step": 5775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.769135123357349e-05,
+      "loss": 2.0498,
+      "step": 5780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768731656272042e-05,
+      "loss": 1.9876,
+      "step": 5785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768327845284622e-05,
+      "loss": 2.048,
+      "step": 5790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76792369042421e-05,
+      "loss": 2.0698,
+      "step": 5795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767519191719951e-05,
+      "loss": 2.0807,
+      "step": 5800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767114349201019e-05,
+      "loss": 2.0654,
+      "step": 5805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766709162896608e-05,
+      "loss": 2.0314,
+      "step": 5810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.766303632835938e-05,
+      "loss": 2.0237,
+      "step": 5815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765897759048254e-05,
+      "loss": 2.0814,
+      "step": 5820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76549154156283e-05,
+      "loss": 2.0744,
+      "step": 5825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.765084980408954e-05,
+      "loss": 2.0833,
+      "step": 5830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76467807561595e-05,
+      "loss": 1.9777,
+      "step": 5835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.764270827213162e-05,
+      "loss": 2.053,
+      "step": 5840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763863235229958e-05,
+      "loss": 2.0446,
+      "step": 5845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763455299695734e-05,
+      "loss": 2.035,
+      "step": 5850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.763047020639906e-05,
+      "loss": 2.041,
+      "step": 5855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76263839809192e-05,
+      "loss": 2.0338,
+      "step": 5860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.76222943208124e-05,
+      "loss": 2.0117,
+      "step": 5865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761820122637364e-05,
+      "loss": 2.0409,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761410469789807e-05,
+      "loss": 2.0336,
+      "step": 5875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.761000473568114e-05,
+      "loss": 2.0689,
+      "step": 5880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760590134001849e-05,
+      "loss": 2.0281,
+      "step": 5885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760179451120606e-05,
+      "loss": 2.0268,
+      "step": 5890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759768424954001e-05,
+      "loss": 2.0482,
+      "step": 5895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.759357055531678e-05,
+      "loss": 2.0189,
+      "step": 5900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7589453428833e-05,
+      "loss": 2.0237,
+      "step": 5905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758533287038559e-05,
+      "loss": 2.0048,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.758120888027172e-05,
+      "loss": 2.0314,
+      "step": 5915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757708145878879e-05,
+      "loss": 2.0502,
+      "step": 5920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.757295060623446e-05,
+      "loss": 2.0154,
+      "step": 5925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756881632290662e-05,
+      "loss": 2.059,
+      "step": 5930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756467860910342e-05,
+      "loss": 2.0896,
+      "step": 5935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.756053746512326e-05,
+      "loss": 2.0187,
+      "step": 5940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755639289126477e-05,
+      "loss": 2.0415,
+      "step": 5945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.755224488782686e-05,
+      "loss": 2.0097,
+      "step": 5950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754809345510865e-05,
+      "loss": 1.9988,
+      "step": 5955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.754393859340955e-05,
+      "loss": 2.0566,
+      "step": 5960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753978030302916e-05,
+      "loss": 1.9779,
+      "step": 5965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753561858426738e-05,
+      "loss": 2.0447,
+      "step": 5970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.753145343742434e-05,
+      "loss": 2.0316,
+      "step": 5975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752728486280038e-05,
+      "loss": 2.0163,
+      "step": 5980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.752311286069618e-05,
+      "loss": 2.0232,
+      "step": 5985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751893743141256e-05,
+      "loss": 2.116,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751475857525066e-05,
+      "loss": 2.0156,
+      "step": 5995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.751057629251182e-05,
+      "loss": 2.0174,
+      "step": 6000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750639058349766e-05,
+      "loss": 2.0772,
+      "step": 6005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.750220144851006e-05,
+      "loss": 2.0689,
+      "step": 6010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749800888785107e-05,
+      "loss": 2.0447,
+      "step": 6015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.749381290182311e-05,
+      "loss": 1.9924,
+      "step": 6020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748961349072873e-05,
+      "loss": 2.0626,
+      "step": 6025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748541065487078e-05,
+      "loss": 2.0679,
+      "step": 6030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.748120439455237e-05,
+      "loss": 2.0341,
+      "step": 6035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747699471007683e-05,
+      "loss": 2.0262,
+      "step": 6040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.747278160174774e-05,
+      "loss": 2.0225,
+      "step": 6045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746856506986893e-05,
+      "loss": 1.996,
+      "step": 6050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74643451147445e-05,
+      "loss": 1.9946,
+      "step": 6055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.746012173667875e-05,
+      "loss": 2.066,
+      "step": 6060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745589493597627e-05,
+      "loss": 2.0456,
+      "step": 6065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.745166471294189e-05,
+      "loss": 2.0478,
+      "step": 6070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.744743106788065e-05,
+      "loss": 2.0155,
+      "step": 6075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74431940010979e-05,
+      "loss": 2.0399,
+      "step": 6080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743895351289914e-05,
+      "loss": 2.0545,
+      "step": 6085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743470960359025e-05,
+      "loss": 2.0111,
+      "step": 6090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.743046227347724e-05,
+      "loss": 2.0361,
+      "step": 6095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74262115228664e-05,
+      "loss": 2.015,
+      "step": 6100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.742195735206432e-05,
+      "loss": 2.0927,
+      "step": 6105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741769976137775e-05,
+      "loss": 2.0804,
+      "step": 6110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.741343875111377e-05,
+      "loss": 2.0644,
+      "step": 6115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740917432157966e-05,
+      "loss": 2.0162,
+      "step": 6120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740490647308292e-05,
+      "loss": 2.0437,
+      "step": 6125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.740063520593136e-05,
+      "loss": 2.0678,
+      "step": 6130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739636052043302e-05,
+      "loss": 2.0308,
+      "step": 6135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.739208241689612e-05,
+      "loss": 2.0177,
+      "step": 6140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738780089562924e-05,
+      "loss": 2.0395,
+      "step": 6145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.738351595694111e-05,
+      "loss": 2.0334,
+      "step": 6150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737922760114075e-05,
+      "loss": 2.0632,
+      "step": 6155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737493582853743e-05,
+      "loss": 2.0861,
+      "step": 6160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737064063944066e-05,
+      "loss": 2.0174,
+      "step": 6165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736634203416015e-05,
+      "loss": 2.0289,
+      "step": 6170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736204001300595e-05,
+      "loss": 2.063,
+      "step": 6175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735773457628829e-05,
+      "loss": 2.0307,
+      "step": 6180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735342572431764e-05,
+      "loss": 2.0248,
+      "step": 6185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734911345740475e-05,
+      "loss": 1.9808,
+      "step": 6190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734479777586061e-05,
+      "loss": 2.0007,
+      "step": 6195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734047867999645e-05,
+      "loss": 2.0086,
+      "step": 6200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733615617012375e-05,
+      "loss": 2.0301,
+      "step": 6205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733183024655421e-05,
+      "loss": 2.0822,
+      "step": 6210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732750090959981e-05,
+      "loss": 2.1218,
+      "step": 6215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.732316815957278e-05,
+      "loss": 2.047,
+      "step": 6220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731883199678556e-05,
+      "loss": 2.0822,
+      "step": 6225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731449242155087e-05,
+      "loss": 2.0126,
+      "step": 6230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.731014943418165e-05,
+      "loss": 2.0385,
+      "step": 6235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73058030349911e-05,
+      "loss": 2.0252,
+      "step": 6240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73014532242927e-05,
+      "loss": 2.0378,
+      "step": 6245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72971000024001e-05,
+      "loss": 2.0636,
+      "step": 6250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729274336962725e-05,
+      "loss": 2.0113,
+      "step": 6255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728838332628834e-05,
+      "loss": 2.0244,
+      "step": 6260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.728401987269778e-05,
+      "loss": 2.0474,
+      "step": 6265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727965300917027e-05,
+      "loss": 2.0238,
+      "step": 6270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727528273602072e-05,
+      "loss": 2.0476,
+      "step": 6275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727090905356429e-05,
+      "loss": 2.068,
+      "step": 6280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.726653196211639e-05,
+      "loss": 2.0768,
+      "step": 6285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72621514619927e-05,
+      "loss": 1.9921,
+      "step": 6290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725776755350912e-05,
+      "loss": 2.0531,
+      "step": 6295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725338023698178e-05,
+      "loss": 2.0199,
+      "step": 6300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72489895127271e-05,
+      "loss": 2.0504,
+      "step": 6305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724459538106169e-05,
+      "loss": 2.0161,
+      "step": 6310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.724019784230246e-05,
+      "loss": 1.9904,
+      "step": 6315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723579689676655e-05,
+      "loss": 2.0497,
+      "step": 6320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.723139254477132e-05,
+      "loss": 2.0145,
+      "step": 6325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722698478663441e-05,
+      "loss": 2.0086,
+      "step": 6330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.722257362267368e-05,
+      "loss": 2.0205,
+      "step": 6335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721815905320725e-05,
+      "loss": 2.0749,
+      "step": 6340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72137410785535e-05,
+      "loss": 2.0984,
+      "step": 6345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7209319699031e-05,
+      "loss": 2.0228,
+      "step": 6350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720489491495861e-05,
+      "loss": 2.0499,
+      "step": 6355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.720046672665546e-05,
+      "loss": 2.0125,
+      "step": 6360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719603513444087e-05,
+      "loss": 2.0304,
+      "step": 6365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719160013863441e-05,
+      "loss": 1.9992,
+      "step": 6370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718716173955596e-05,
+      "loss": 2.0739,
+      "step": 6375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.718271993752555e-05,
+      "loss": 2.0163,
+      "step": 6380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717827473286354e-05,
+      "loss": 1.9899,
+      "step": 6385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71738261258905e-05,
+      "loss": 2.0276,
+      "step": 6390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716937411692722e-05,
+      "loss": 2.0176,
+      "step": 6395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716491870629479e-05,
+      "loss": 2.0317,
+      "step": 6400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.716045989431451e-05,
+      "loss": 2.0101,
+      "step": 6405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71559976813079e-05,
+      "loss": 2.0028,
+      "step": 6410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71515320675968e-05,
+      "loss": 2.0536,
+      "step": 6415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714706305350325e-05,
+      "loss": 2.0348,
+      "step": 6420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.714259063934951e-05,
+      "loss": 2.0176,
+      "step": 6425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713811482545812e-05,
+      "loss": 1.9996,
+      "step": 6430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.713363561215188e-05,
+      "loss": 2.0419,
+      "step": 6435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712915299975377e-05,
+      "loss": 2.0701,
+      "step": 6440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.712466698858712e-05,
+      "loss": 2.0162,
+      "step": 6445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.71201775789754e-05,
+      "loss": 2.0254,
+      "step": 6450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711568477124238e-05,
+      "loss": 1.9776,
+      "step": 6455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.711118856571203e-05,
+      "loss": 2.0309,
+      "step": 6460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710668896270868e-05,
+      "loss": 2.0149,
+      "step": 6465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.710218596255673e-05,
+      "loss": 2.0528,
+      "step": 6470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7097679565581e-05,
+      "loss": 2.0709,
+      "step": 6475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.709316977210641e-05,
+      "loss": 1.9945,
+      "step": 6480
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.708865658245823e-05,
+      "loss": 2.0354,
+      "step": 6485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.70841399969619e-05,
+      "loss": 2.084,
+      "step": 6490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707962001594318e-05,
+      "loss": 2.0831,
+      "step": 6495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7075096639728e-05,
+      "loss": 2.0783,
+      "step": 6500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.707056986864257e-05,
+      "loss": 2.0602,
+      "step": 6505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706603970301336e-05,
+      "loss": 2.0087,
+      "step": 6510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.706150614316705e-05,
+      "loss": 2.0342,
+      "step": 6515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705696918943059e-05,
+      "loss": 1.9722,
+      "step": 6520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.705242884213118e-05,
+      "loss": 2.1146,
+      "step": 6525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704788510159623e-05,
+      "loss": 2.039,
+      "step": 6530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.704333796815342e-05,
+      "loss": 2.1147,
+      "step": 6535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703878744213068e-05,
+      "loss": 2.0703,
+      "step": 6540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.703423352385618e-05,
+      "loss": 2.0637,
+      "step": 6545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702967621365833e-05,
+      "loss": 2.0388,
+      "step": 6550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702511551186577e-05,
+      "loss": 1.9933,
+      "step": 6555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.702055141880741e-05,
+      "loss": 2.0045,
+      "step": 6560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701598393481242e-05,
+      "loss": 2.0194,
+      "step": 6565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.701141306021014e-05,
+      "loss": 2.0125,
+      "step": 6570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700683879533022e-05,
+      "loss": 2.0192,
+      "step": 6575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.700226114050255e-05,
+      "loss": 2.0277,
+      "step": 6580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699768009605724e-05,
+      "loss": 2.0486,
+      "step": 6585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.699309566232466e-05,
+      "loss": 2.0716,
+      "step": 6590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698850783963544e-05,
+      "loss": 2.0327,
+      "step": 6595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.698391662832041e-05,
+      "loss": 2.0081,
+      "step": 6600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69793220287107e-05,
+      "loss": 2.0239,
+      "step": 6605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697472404113762e-05,
+      "loss": 2.0896,
+      "step": 6610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.697012266593277e-05,
+      "loss": 2.0653,
+      "step": 6615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696551790342798e-05,
+      "loss": 2.0113,
+      "step": 6620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.696090975395534e-05,
+      "loss": 2.07,
+      "step": 6625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695629821784717e-05,
+      "loss": 2.0455,
+      "step": 6630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.695168329543602e-05,
+      "loss": 2.0135,
+      "step": 6635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694706498705472e-05,
+      "loss": 2.0197,
+      "step": 6640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.694244329303632e-05,
+      "loss": 2.0243,
+      "step": 6645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69378182137141e-05,
+      "loss": 2.0065,
+      "step": 6650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.693318974942163e-05,
+      "loss": 1.9883,
+      "step": 6655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.69285579004927e-05,
+      "loss": 2.0575,
+      "step": 6660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.692392266726129e-05,
+      "loss": 2.0505,
+      "step": 6665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691928405006172e-05,
+      "loss": 2.0729,
+      "step": 6670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.691464204922848e-05,
+      "loss": 2.0541,
+      "step": 6675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690999666509636e-05,
+      "loss": 2.0186,
+      "step": 6680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690534789800037e-05,
+      "loss": 1.9662,
+      "step": 6685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.690069574827575e-05,
+      "loss": 2.0729,
+      "step": 6690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689604021625798e-05,
+      "loss": 2.0653,
+      "step": 6695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.689138130228282e-05,
+      "loss": 2.0391,
+      "step": 6700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688671900668623e-05,
+      "loss": 1.9944,
+      "step": 6705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.688205332980445e-05,
+      "loss": 2.0134,
+      "step": 6710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687738427197394e-05,
+      "loss": 1.9427,
+      "step": 6715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.687271183353145e-05,
+      "loss": 2.0638,
+      "step": 6720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.686803601481389e-05,
+      "loss": 2.0479,
+      "step": 6725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.68633568161585e-05,
+      "loss": 2.0221,
+      "step": 6730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.685867423790269e-05,
+      "loss": 2.1123,
+      "step": 6735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68539882803842e-05,
+      "loss": 2.1121,
+      "step": 6740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68492989439409e-05,
+      "loss": 2.0229,
+      "step": 6745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6844606228911e-05,
+      "loss": 2.0437,
+      "step": 6750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683991013563291e-05,
+      "loss": 1.9907,
+      "step": 6755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683521066444532e-05,
+      "loss": 2.0228,
+      "step": 6760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68305078156871e-05,
+      "loss": 2.0317,
+      "step": 6765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682580158969743e-05,
+      "loss": 2.0833,
+      "step": 6770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682109198681568e-05,
+      "loss": 2.0152,
+      "step": 6775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.681637900738153e-05,
+      "loss": 2.0718,
+      "step": 6780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68116626517348e-05,
+      "loss": 2.0638,
+      "step": 6785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680694292021566e-05,
+      "loss": 2.0506,
+      "step": 6790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680221981316447e-05,
+      "loss": 2.0657,
+      "step": 6795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679749333092182e-05,
+      "loss": 2.0051,
+      "step": 6800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679276347382862e-05,
+      "loss": 2.07,
+      "step": 6805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67880302422259e-05,
+      "loss": 1.9961,
+      "step": 6810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678329363645505e-05,
+      "loss": 2.0782,
+      "step": 6815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677855365685763e-05,
+      "loss": 2.0072,
+      "step": 6820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.677381030377548e-05,
+      "loss": 2.0016,
+      "step": 6825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676906357755066e-05,
+      "loss": 2.0606,
+      "step": 6830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67643134785255e-05,
+      "loss": 2.0154,
+      "step": 6835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675956000704256e-05,
+      "loss": 2.0521,
+      "step": 6840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675480316344464e-05,
+      "loss": 2.0101,
+      "step": 6845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.675004294807476e-05,
+      "loss": 2.0013,
+      "step": 6850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.674527936127624e-05,
+      "loss": 2.074,
+      "step": 6855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67405124033926e-05,
+      "loss": 2.0187,
+      "step": 6860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673574207476762e-05,
+      "loss": 2.0546,
+      "step": 6865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673096837574529e-05,
+      "loss": 2.0629,
+      "step": 6870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.67261913066699e-05,
+      "loss": 2.0187,
+      "step": 6875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.672141086788595e-05,
+      "loss": 2.0095,
+      "step": 6880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671662705973816e-05,
+      "loss": 2.0192,
+      "step": 6885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671183988257155e-05,
+      "loss": 1.9865,
+      "step": 6890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670704933673136e-05,
+      "loss": 2.0227,
+      "step": 6895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.670225542256302e-05,
+      "loss": 2.0336,
+      "step": 6900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669745814041228e-05,
+      "loss": 2.0577,
+      "step": 6905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669265749062511e-05,
+      "loss": 2.0474,
+      "step": 6910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66878534735477e-05,
+      "loss": 2.0493,
+      "step": 6915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.668304608952649e-05,
+      "loss": 2.084,
+      "step": 6920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667823533890819e-05,
+      "loss": 2.0526,
+      "step": 6925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66734212220397e-05,
+      "loss": 2.0193,
+      "step": 6930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666860373926822e-05,
+      "loss": 2.0154,
+      "step": 6935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.666378289094116e-05,
+      "loss": 1.9944,
+      "step": 6940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665895867740617e-05,
+      "loss": 2.1148,
+      "step": 6945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.665413109901119e-05,
+      "loss": 2.0432,
+      "step": 6950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664930015610433e-05,
+      "loss": 1.9604,
+      "step": 6955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6644465849034e-05,
+      "loss": 2.0128,
+      "step": 6960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66396281781488e-05,
+      "loss": 2.0149,
+      "step": 6965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.663478714379765e-05,
+      "loss": 2.0122,
+      "step": 6970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662994274632961e-05,
+      "loss": 2.0835,
+      "step": 6975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.66250949860941e-05,
+      "loss": 2.0478,
+      "step": 6980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662024386344068e-05,
+      "loss": 2.038,
+      "step": 6985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661538937871922e-05,
+      "loss": 2.0768,
+      "step": 6990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.661053153227976e-05,
+      "loss": 1.9897,
+      "step": 6995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660567032447267e-05,
+      "loss": 2.008,
+      "step": 7000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.660080575564853e-05,
+      "loss": 2.0267,
+      "step": 7005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659593782615813e-05,
+      "loss": 1.9761,
+      "step": 7010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.659106653635253e-05,
+      "loss": 2.0462,
+      "step": 7015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658619188658301e-05,
+      "loss": 2.0733,
+      "step": 7020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.658131387720116e-05,
+      "loss": 1.9761,
+      "step": 7025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65764325085587e-05,
+      "loss": 2.0511,
+      "step": 7030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65715477810077e-05,
+      "loss": 2.0119,
+      "step": 7035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656665969490042e-05,
+      "loss": 2.0572,
+      "step": 7040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.656176825058935e-05,
+      "loss": 2.0757,
+      "step": 7045
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655687344842728e-05,
+      "loss": 1.9632,
+      "step": 7050
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.655197528876716e-05,
+      "loss": 2.0156,
+      "step": 7055
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654707377196226e-05,
+      "loss": 2.0838,
+      "step": 7060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.654216889836604e-05,
+      "loss": 2.0625,
+      "step": 7065
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653726066833222e-05,
+      "loss": 2.0261,
+      "step": 7070
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.653234908221476e-05,
+      "loss": 2.0299,
+      "step": 7075
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652743414036788e-05,
+      "loss": 2.0264,
+      "step": 7080
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.652251584314601e-05,
+      "loss": 2.0278,
+      "step": 7085
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651759419090385e-05,
+      "loss": 2.0329,
+      "step": 7090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.651266918399632e-05,
+      "loss": 2.0275,
+      "step": 7095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65077408227786e-05,
+      "loss": 2.0501,
+      "step": 7100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.65028091076061e-05,
+      "loss": 2.0456,
+      "step": 7105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649787403883447e-05,
+      "loss": 2.0356,
+      "step": 7110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.649293561681963e-05,
+      "loss": 1.9973,
+      "step": 7115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648799384191768e-05,
+      "loss": 2.028,
+      "step": 7120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.648304871448503e-05,
+      "loss": 1.9916,
+      "step": 7125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647810023487832e-05,
+      "loss": 2.0089,
+      "step": 7130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.647314840345438e-05,
+      "loss": 1.9644,
+      "step": 7135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646819322057032e-05,
+      "loss": 1.9946,
+      "step": 7140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646323468658351e-05,
+      "loss": 2.0344,
+      "step": 7145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645827280185153e-05,
+      "loss": 2.0377,
+      "step": 7150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.645330756673221e-05,
+      "loss": 2.0362,
+      "step": 7155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64483389815836e-05,
+      "loss": 2.0594,
+      "step": 7160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.644336704676406e-05,
+      "loss": 2.0564,
+      "step": 7165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643839176263211e-05,
+      "loss": 2.0357,
+      "step": 7170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643341312954658e-05,
+      "loss": 2.0751,
+      "step": 7175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.642843114786648e-05,
+      "loss": 2.0437,
+      "step": 7180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.64234458179511e-05,
+      "loss": 2.056,
+      "step": 7185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641845714015995e-05,
+      "loss": 2.0546,
+      "step": 7190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.641346511485283e-05,
+      "loss": 2.0099,
+      "step": 7195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640846974238972e-05,
+      "loss": 1.996,
+      "step": 7200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640347102313086e-05,
+      "loss": 2.0687,
+      "step": 7205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639846895743676e-05,
+      "loss": 2.0546,
+      "step": 7210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.639346354566811e-05,
+      "loss": 2.0343,
+      "step": 7215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638845478818595e-05,
+      "loss": 2.0668,
+      "step": 7220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.638344268535142e-05,
+      "loss": 1.9709,
+      "step": 7225
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6378427237526e-05,
+      "loss": 2.0546,
+      "step": 7230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637340844507139e-05,
+      "loss": 1.9704,
+      "step": 7235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636838630834952e-05,
+      "loss": 2.0798,
+      "step": 7240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.636336082772259e-05,
+      "loss": 2.0609,
+      "step": 7245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635833200355297e-05,
+      "loss": 2.0013,
+      "step": 7250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.635329983620336e-05,
+      "loss": 2.0287,
+      "step": 7255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634826432603662e-05,
+      "loss": 2.0571,
+      "step": 7260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.634322547341595e-05,
+      "loss": 2.0113,
+      "step": 7265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633818327870468e-05,
+      "loss": 2.0157,
+      "step": 7270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.633313774226646e-05,
+      "loss": 1.991,
+      "step": 7275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632808886446513e-05,
+      "loss": 2.0238,
+      "step": 7280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.632303664566485e-05,
+      "loss": 2.0315,
+      "step": 7285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63179810862299e-05,
+      "loss": 2.0006,
+      "step": 7290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.631292218652488e-05,
+      "loss": 2.0113,
+      "step": 7295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630785994691465e-05,
+      "loss": 2.0326,
+      "step": 7300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630279436776426e-05,
+      "loss": 2.004,
+      "step": 7305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629772544943902e-05,
+      "loss": 2.0755,
+      "step": 7310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.629265319230448e-05,
+      "loss": 2.0596,
+      "step": 7315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.628757759672642e-05,
+      "loss": 2.0208,
+      "step": 7320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62824986630709e-05,
+      "loss": 2.036,
+      "step": 7325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627741639170418e-05,
+      "loss": 2.023,
+      "step": 7330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627233078299276e-05,
+      "loss": 2.0062,
+      "step": 7335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62672418373034e-05,
+      "loss": 2.0812,
+      "step": 7340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626214955500313e-05,
+      "loss": 2.0355,
+      "step": 7345
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625705393645912e-05,
+      "loss": 2.0171,
+      "step": 7350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625195498203888e-05,
+      "loss": 2.0253,
+      "step": 7355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624685269211013e-05,
+      "loss": 2.0607,
+      "step": 7360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624174706704083e-05,
+      "loss": 2.0563,
+      "step": 7365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623663810719918e-05,
+      "loss": 2.0585,
+      "step": 7370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62315258129536e-05,
+      "loss": 1.9767,
+      "step": 7375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622641018467278e-05,
+      "loss": 2.0098,
+      "step": 7380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622129122272563e-05,
+      "loss": 2.0431,
+      "step": 7385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621616892748131e-05,
+      "loss": 2.0423,
+      "step": 7390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621104329930924e-05,
+      "loss": 1.992,
+      "step": 7395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620591433857904e-05,
+      "loss": 1.9847,
+      "step": 7400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620078204566059e-05,
+      "loss": 2.0033,
+      "step": 7405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.619564642092403e-05,
+      "loss": 2.0717,
+      "step": 7410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61905074647397e-05,
+      "loss": 1.9875,
+      "step": 7415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618536517747821e-05,
+      "loss": 2.0252,
+      "step": 7420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.618021955951039e-05,
+      "loss": 2.0313,
+      "step": 7425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617507061120734e-05,
+      "loss": 2.0212,
+      "step": 7430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616991833294038e-05,
+      "loss": 2.0536,
+      "step": 7435
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.616476272508106e-05,
+      "loss": 2.0013,
+      "step": 7440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615960378800119e-05,
+      "loss": 2.0582,
+      "step": 7445
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61544415220728e-05,
+      "loss": 2.0402,
+      "step": 7450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.61492759276682e-05,
+      "loss": 2.0223,
+      "step": 7455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.614410700515988e-05,
+      "loss": 2.0089,
+      "step": 7460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613893475492062e-05,
+      "loss": 2.0184,
+      "step": 7465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613375917732343e-05,
+      "loss": 2.0473,
+      "step": 7470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612858027274153e-05,
+      "loss": 2.0121,
+      "step": 7475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.612339804154841e-05,
+      "loss": 2.0001,
+      "step": 7480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611821248411782e-05,
+      "loss": 2.109,
+      "step": 7485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611302360082367e-05,
+      "loss": 2.0477,
+      "step": 7490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610783139204021e-05,
+      "loss": 2.0051,
+      "step": 7495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610263585814184e-05,
+      "loss": 2.0159,
+      "step": 7500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609743699950327e-05,
+      "loss": 2.0119,
+      "step": 7505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.609223481649943e-05,
+      "loss": 2.108,
+      "step": 7510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608702930950544e-05,
+      "loss": 2.0345,
+      "step": 7515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608182047889674e-05,
+      "loss": 2.0287,
+      "step": 7520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607660832504895e-05,
+      "loss": 2.0074,
+      "step": 7525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.607139284833798e-05,
+      "loss": 2.0184,
+      "step": 7530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606617404913989e-05,
+      "loss": 2.0425,
+      "step": 7535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.606095192783109e-05,
+      "loss": 2.0322,
+      "step": 7540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605572648478815e-05,
+      "loss": 2.022,
+      "step": 7545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605049772038794e-05,
+      "loss": 2.0391,
+      "step": 7550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.604526563500751e-05,
+      "loss": 2.0903,
+      "step": 7555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60400302290242e-05,
+      "loss": 1.971,
+      "step": 7560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603479150281554e-05,
+      "loss": 2.0536,
+      "step": 7565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602954945675933e-05,
+      "loss": 1.9992,
+      "step": 7570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602430409123363e-05,
+      "loss": 1.9994,
+      "step": 7575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.60190554066167e-05,
+      "loss": 2.0144,
+      "step": 7580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.601380340328704e-05,
+      "loss": 2.0373,
+      "step": 7585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600854808162344e-05,
+      "loss": 2.0636,
+      "step": 7590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600328944200486e-05,
+      "loss": 2.0335,
+      "step": 7595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599802748481055e-05,
+      "loss": 1.9738,
+      "step": 7600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599276221041997e-05,
+      "loss": 2.0678,
+      "step": 7605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598749361921285e-05,
+      "loss": 1.9855,
+      "step": 7610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59822217115691e-05,
+      "loss": 2.011,
+      "step": 7615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597694648786896e-05,
+      "loss": 2.0422,
+      "step": 7620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.597166794849281e-05,
+      "loss": 2.0393,
+      "step": 7625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.596638609382137e-05,
+      "loss": 2.0362,
+      "step": 7630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59611009242355e-05,
+      "loss": 2.0027,
+      "step": 7635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595581244011637e-05,
+      "loss": 1.9917,
+      "step": 7640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595052064184535e-05,
+      "loss": 2.0123,
+      "step": 7645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.594522552980408e-05,
+      "loss": 2.0641,
+      "step": 7650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593992710437442e-05,
+      "loss": 2.0223,
+      "step": 7655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.593462536593844e-05,
+      "loss": 1.9601,
+      "step": 7660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592932031487854e-05,
+      "loss": 2.0081,
+      "step": 7665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592401195157726e-05,
+      "loss": 2.048,
+      "step": 7670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59187002764174e-05,
+      "loss": 2.0383,
+      "step": 7675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.591338528978205e-05,
+      "loss": 2.0184,
+      "step": 7680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59080669920545e-05,
+      "loss": 2.0195,
+      "step": 7685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.59027453836183e-05,
+      "loss": 2.0035,
+      "step": 7690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589742046485718e-05,
+      "loss": 1.997,
+      "step": 7695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589209223615518e-05,
+      "loss": 2.0762,
+      "step": 7700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588676069789654e-05,
+      "loss": 2.0341,
+      "step": 7705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.588142585046578e-05,
+      "loss": 1.9911,
+      "step": 7710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58760876942476e-05,
+      "loss": 2.103,
+      "step": 7715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.587074622962698e-05,
+      "loss": 2.0506,
+      "step": 7720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.58654014569891e-05,
+      "loss": 2.0111,
+      "step": 7725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586005337671942e-05,
+      "loss": 1.9653,
+      "step": 7730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.585470198920363e-05,
+      "loss": 2.0373,
+      "step": 7735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584934729482766e-05,
+      "loss": 2.0156,
+      "step": 7740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.584398929397763e-05,
+      "loss": 1.983,
+      "step": 7745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583862798703997e-05,
+      "loss": 2.0093,
+      "step": 7750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583326337440131e-05,
+      "loss": 2.0569,
+      "step": 7755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582789545644852e-05,
+      "loss": 2.021,
+      "step": 7760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.582252423356872e-05,
+      "loss": 2.0511,
+      "step": 7765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581714970614927e-05,
+      "loss": 1.9782,
+      "step": 7770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581177187457773e-05,
+      "loss": 1.9815,
+      "step": 7775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580639073924196e-05,
+      "loss": 2.0441,
+      "step": 7780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.580100630052999e-05,
+      "loss": 2.0508,
+      "step": 7785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579561855883017e-05,
+      "loss": 1.9998,
+      "step": 7790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.579022751453101e-05,
+      "loss": 2.058,
+      "step": 7795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57848331680213e-05,
+      "loss": 1.9847,
+      "step": 7800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577943551969005e-05,
+      "loss": 2.1067,
+      "step": 7805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.577403456992653e-05,
+      "loss": 2.0542,
+      "step": 7810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.576863031912023e-05,
+      "loss": 1.9782,
+      "step": 7815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.57632227676609e-05,
+      "loss": 2.0087,
+      "step": 7820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575781191593846e-05,
+      "loss": 1.9946,
+      "step": 7825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.575239776434319e-05,
+      "loss": 2.0052,
+      "step": 7830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574698031326549e-05,
+      "loss": 2.011,
+      "step": 7835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.574155956309606e-05,
+      "loss": 2.0322,
+      "step": 7840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573613551422582e-05,
+      "loss": 2.0623,
+      "step": 7845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.573070816704593e-05,
+      "loss": 1.9428,
+      "step": 7850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572527752194777e-05,
+      "loss": 1.9933,
+      "step": 7855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571984357932302e-05,
+      "loss": 1.9933,
+      "step": 7860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571440633956353e-05,
+      "loss": 2.0083,
+      "step": 7865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570896580306141e-05,
+      "loss": 2.0464,
+      "step": 7870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.570352197020901e-05,
+      "loss": 2.0458,
+      "step": 7875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569807484139894e-05,
+      "loss": 2.0259,
+      "step": 7880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.569262441702398e-05,
+      "loss": 1.9954,
+      "step": 7885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.568717069747724e-05,
+      "loss": 1.988,
+      "step": 7890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.5681713683152e-05,
+      "loss": 2.0539,
+      "step": 7895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56762533744418e-05,
+      "loss": 1.9995,
+      "step": 7900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.56707897717404e-05,
+      "loss": 1.9898,
+      "step": 7905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.566532287544184e-05,
+      "loss": 1.9955,
+      "step": 7910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565985268594034e-05,
+      "loss": 2.0207,
+      "step": 7915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565437920363043e-05,
+      "loss": 2.0022,
+      "step": 7920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56489024289068e-05,
+      "loss": 2.0629,
+      "step": 7925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564342236216444e-05,
+      "loss": 2.0102,
+      "step": 7930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563793900379854e-05,
+      "loss": 2.0331,
+      "step": 7935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563245235420453e-05,
+      "loss": 1.9755,
+      "step": 7940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562696241377809e-05,
+      "loss": 2.0137,
+      "step": 7945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562146918291512e-05,
+      "loss": 2.0389,
+      "step": 7950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56159726620118e-05,
+      "loss": 2.0081,
+      "step": 7955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56104728514645e-05,
+      "loss": 1.972,
+      "step": 7960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560496975166986e-05,
+      "loss": 2.0009,
+      "step": 7965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559946336302473e-05,
+      "loss": 2.0008,
+      "step": 7970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559395368592619e-05,
+      "loss": 1.9982,
+      "step": 7975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558844072077158e-05,
+      "loss": 2.0202,
+      "step": 7980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558292446795852e-05,
+      "loss": 2.0543,
+      "step": 7985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557740492788476e-05,
+      "loss": 2.0071,
+      "step": 7990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.55718821009484e-05,
+      "loss": 2.0189,
+      "step": 7995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.556635598754768e-05,
+      "loss": 2.0027,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 58593,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.905411560256307e+19,
+  "trial_name": null,
+  "trial_params": null
+}