diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,77890 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 11.527111111111111,
+  "global_step": 64840,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997037037037034e-05,
+      "loss": 4.7865,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999407407407407e-05,
+      "loss": 4.1868,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999111111111111e-05,
+      "loss": 3.8968,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998814814814815e-05,
+      "loss": 3.7514,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998518518518518e-05,
+      "loss": 3.6851,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998222222222222e-05,
+      "loss": 3.5718,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997925925925926e-05,
+      "loss": 3.5858,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99762962962963e-05,
+      "loss": 3.484,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997333333333333e-05,
+      "loss": 3.5216,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997037037037037e-05,
+      "loss": 3.4322,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996740740740741e-05,
+      "loss": 3.3487,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996444444444445e-05,
+      "loss": 3.3356,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.996148148148148e-05,
+      "loss": 3.2931,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995851851851852e-05,
+      "loss": 3.4107,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.995555555555556e-05,
+      "loss": 3.3909,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9952592592592596e-05,
+      "loss": 3.2811,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994962962962963e-05,
+      "loss": 3.288,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994666666666667e-05,
+      "loss": 3.2796,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9943703703703706e-05,
+      "loss": 3.2775,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9940740740740745e-05,
+      "loss": 3.2935,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993777777777778e-05,
+      "loss": 3.2768,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9934814814814816e-05,
+      "loss": 3.3402,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9931851851851855e-05,
+      "loss": 3.178,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9928888888888893e-05,
+      "loss": 3.13,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9925925925925926e-05,
+      "loss": 3.1009,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9922962962962964e-05,
+      "loss": 3.2681,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992e-05,
+      "loss": 3.0927,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.991703703703704e-05,
+      "loss": 3.1378,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9914074074074074e-05,
+      "loss": 3.0502,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991111111111111e-05,
+      "loss": 3.1192,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990814814814815e-05,
+      "loss": 3.1041,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990518518518519e-05,
+      "loss": 3.0226,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990222222222222e-05,
+      "loss": 3.1867,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989925925925926e-05,
+      "loss": 3.0374,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9896296296296293e-05,
+      "loss": 3.0775,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989333333333334e-05,
+      "loss": 3.0081,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989037037037037e-05,
+      "loss": 3.0772,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988740740740741e-05,
+      "loss": 3.1312,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988444444444444e-05,
+      "loss": 2.8459,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.988148148148149e-05,
+      "loss": 3.1078,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987851851851852e-05,
+      "loss": 3.1878,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987555555555556e-05,
+      "loss": 3.0411,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.987259259259259e-05,
+      "loss": 2.9985,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9869629629629636e-05,
+      "loss": 3.0428,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986666666666667e-05,
+      "loss": 3.0759,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986370370370371e-05,
+      "loss": 2.9847,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.986074074074074e-05,
+      "loss": 3.0635,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985777777777778e-05,
+      "loss": 3.0223,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9854814814814817e-05,
+      "loss": 3.0469,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9851851851851855e-05,
+      "loss": 3.105,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.984888888888889e-05,
+      "loss": 3.0966,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9845925925925926e-05,
+      "loss": 3.0835,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9842962962962965e-05,
+      "loss": 3.0889,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 2.9536,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9837037037037036e-05,
+      "loss": 3.1165,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9834074074074075e-05,
+      "loss": 2.9886,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9831111111111114e-05,
+      "loss": 3.1402,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982814814814815e-05,
+      "loss": 3.0506,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9825185185185184e-05,
+      "loss": 3.0962,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.982222222222222e-05,
+      "loss": 2.9656,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.981925925925926e-05,
+      "loss": 3.018,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.98162962962963e-05,
+      "loss": 2.9314,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981333333333333e-05,
+      "loss": 3.0717,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.981037037037037e-05,
+      "loss": 3.0789,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980740740740741e-05,
+      "loss": 3.0124,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980444444444445e-05,
+      "loss": 2.937,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980148148148148e-05,
+      "loss": 2.8069,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979851851851852e-05,
+      "loss": 2.832,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.979555555555556e-05,
+      "loss": 3.0651,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.97925925925926e-05,
+      "loss": 3.015,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978962962962963e-05,
+      "loss": 2.921,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978666666666667e-05,
+      "loss": 2.946,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.978370370370371e-05,
+      "loss": 2.8124,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9780740740740746e-05,
+      "loss": 2.9505,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977777777777778e-05,
+      "loss": 2.8616,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.977481481481482e-05,
+      "loss": 2.8996,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9771851851851856e-05,
+      "loss": 2.8545,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9768888888888895e-05,
+      "loss": 3.0045,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976592592592593e-05,
+      "loss": 2.8424,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9762962962962966e-05,
+      "loss": 2.8732,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.976e-05,
+      "loss": 2.7691,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9757037037037044e-05,
+      "loss": 2.7451,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9754074074074076e-05,
+      "loss": 2.8599,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9751111111111114e-05,
+      "loss": 2.7836,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9748148148148146e-05,
+      "loss": 2.7931,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.974518518518519e-05,
+      "loss": 2.9787,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9742222222222224e-05,
+      "loss": 2.9116,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973925925925926e-05,
+      "loss": 2.761,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9736296296296295e-05,
+      "loss": 2.8602,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973333333333334e-05,
+      "loss": 2.9382,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.973037037037037e-05,
+      "loss": 2.8331,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972740740740741e-05,
+      "loss": 2.8605,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9724444444444443e-05,
+      "loss": 2.9158,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.972148148148148e-05,
+      "loss": 2.8463,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.971851851851852e-05,
+      "loss": 2.8619,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.971555555555556e-05,
+      "loss": 2.8954,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.971259259259259e-05,
+      "loss": 2.8213,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970962962962963e-05,
+      "loss": 2.9462,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970666666666667e-05,
+      "loss": 2.8364,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970370370370371e-05,
+      "loss": 2.8938,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.970074074074074e-05,
+      "loss": 2.763,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.969777777777778e-05,
+      "loss": 2.9042,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.969481481481482e-05,
+      "loss": 2.7493,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.969185185185186e-05,
+      "loss": 2.9707,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968888888888889e-05,
+      "loss": 2.8097,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.968592592592593e-05,
+      "loss": 2.9142,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.968296296296297e-05,
+      "loss": 2.8978,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 2.8335,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.967703703703704e-05,
+      "loss": 2.8292,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9674074074074076e-05,
+      "loss": 2.8864,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9671111111111115e-05,
+      "loss": 2.7604,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9668148148148154e-05,
+      "loss": 2.8053,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9665185185185186e-05,
+      "loss": 2.8569,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9662222222222225e-05,
+      "loss": 2.8642,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9659259259259264e-05,
+      "loss": 2.7163,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.96562962962963e-05,
+      "loss": 2.7935,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9653333333333335e-05,
+      "loss": 2.8108,
+      "step": 585
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.965037037037037e-05,
+      "loss": 2.8263,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964740740740741e-05,
+      "loss": 2.8264,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964444444444445e-05,
+      "loss": 2.7342,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.964148148148148e-05,
+      "loss": 2.8448,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963851851851852e-05,
+      "loss": 2.791,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.963555555555556e-05,
+      "loss": 2.8689,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.96325925925926e-05,
+      "loss": 2.6708,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962962962962963e-05,
+      "loss": 2.6226,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962666666666667e-05,
+      "loss": 2.9335,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.96237037037037e-05,
+      "loss": 2.8356,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.962074074074075e-05,
+      "loss": 2.7369,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.961777777777778e-05,
+      "loss": 2.6848,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.961481481481482e-05,
+      "loss": 2.7344,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.961185185185185e-05,
+      "loss": 2.8622,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9608888888888897e-05,
+      "loss": 2.7942,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.960592592592593e-05,
+      "loss": 2.6875,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.960296296296297e-05,
+      "loss": 2.7193,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.96e-05,
+      "loss": 2.7213,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9597037037037045e-05,
+      "loss": 2.755,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.959407407407408e-05,
+      "loss": 2.8535,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9591111111111116e-05,
+      "loss": 2.7465,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958814814814815e-05,
+      "loss": 2.8645,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.958518518518519e-05,
+      "loss": 2.8154,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9582222222222226e-05,
+      "loss": 2.8004,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9579259259259264e-05,
+      "loss": 2.786,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9576296296296296e-05,
+      "loss": 2.6265,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9573333333333335e-05,
+      "loss": 2.8455,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9570370370370374e-05,
+      "loss": 2.6883,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.956740740740741e-05,
+      "loss": 2.8027,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9564444444444445e-05,
+      "loss": 2.7642,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9561481481481484e-05,
+      "loss": 2.7519,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.955851851851852e-05,
+      "loss": 2.6603,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.955555555555556e-05,
+      "loss": 2.7165,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.9552592592592593e-05,
+      "loss": 2.8298,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.954962962962963e-05,
+      "loss": 2.8128,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.954666666666667e-05,
+      "loss": 2.7578,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.954370370370371e-05,
+      "loss": 2.7739,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.954074074074074e-05,
+      "loss": 2.642,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9537777777777774e-05,
+      "loss": 2.7469,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.953481481481482e-05,
+      "loss": 2.8179,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.953185185185185e-05,
+      "loss": 2.6917,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.952888888888889e-05,
+      "loss": 2.6922,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.952592592592592e-05,
+      "loss": 2.7306,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.952296296296297e-05,
+      "loss": 2.7722,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.952e-05,
+      "loss": 2.8362,
+      "step": 810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.951703703703704e-05,
+      "loss": 2.7178,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.951407407407407e-05,
+      "loss": 2.7951,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.951111111111112e-05,
+      "loss": 2.6786,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.950814814814815e-05,
+      "loss": 2.8212,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.950518518518519e-05,
+      "loss": 2.5981,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.950222222222222e-05,
+      "loss": 2.8504,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9499259259259265e-05,
+      "loss": 2.7314,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.94962962962963e-05,
+      "loss": 2.8478,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9493333333333336e-05,
+      "loss": 2.8575,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.949037037037037e-05,
+      "loss": 2.7991,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.948740740740741e-05,
+      "loss": 2.7548,
+      "step": 865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9484444444444446e-05,
+      "loss": 2.7882,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9481481481481485e-05,
+      "loss": 2.6827,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.947851851851852e-05,
+      "loss": 2.6902,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9475555555555555e-05,
+      "loss": 2.7625,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9472592592592594e-05,
+      "loss": 2.6774,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946962962962963e-05,
+      "loss": 2.7684,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9466666666666665e-05,
+      "loss": 2.7199,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9463703703703704e-05,
+      "loss": 2.6763,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.946074074074074e-05,
+      "loss": 2.7746,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945777777777778e-05,
+      "loss": 2.7869,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9454814814814814e-05,
+      "loss": 2.7418,
+      "step": 920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.945185185185185e-05,
+      "loss": 2.8152,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.944888888888889e-05,
+      "loss": 2.7189,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.944592592592593e-05,
+      "loss": 2.8392,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.944296296296296e-05,
+      "loss": 2.7173,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.944e-05,
+      "loss": 2.7719,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.943703703703704e-05,
+      "loss": 2.7683,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.943407407407408e-05,
+      "loss": 2.8069,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.943111111111111e-05,
+      "loss": 2.7951,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.942814814814815e-05,
+      "loss": 2.7212,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.942518518518519e-05,
+      "loss": 2.8281,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.942222222222223e-05,
+      "loss": 2.9378,
+      "step": 975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.941925925925926e-05,
+      "loss": 2.7481,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.94162962962963e-05,
+      "loss": 2.8349,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.941333333333334e-05,
+      "loss": 2.6851,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9410370370370376e-05,
+      "loss": 2.7403,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.940740740740741e-05,
+      "loss": 2.7812,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9404444444444447e-05,
+      "loss": 2.6925,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.940148148148148e-05,
+      "loss": 2.7102,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9398518518518524e-05,
+      "loss": 2.6798,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9395555555555556e-05,
+      "loss": 2.7171,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9392592592592595e-05,
+      "loss": 2.709,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.938962962962963e-05,
+      "loss": 2.7912,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.938666666666667e-05,
+      "loss": 2.6622,
+      "step": 1035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9383703703703705e-05,
+      "loss": 2.7107,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9380740740740744e-05,
+      "loss": 2.7652,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9377777777777776e-05,
+      "loss": 2.7961,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.937481481481482e-05,
+      "loss": 2.5857,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.937185185185185e-05,
+      "loss": 2.5863,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.936888888888889e-05,
+      "loss": 2.6799,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9365925925925924e-05,
+      "loss": 2.6358,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.936296296296297e-05,
+      "loss": 2.768,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.936e-05,
+      "loss": 2.886,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.935703703703704e-05,
+      "loss": 2.5491,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.935407407407407e-05,
+      "loss": 2.7212,
+      "step": 1090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.935111111111111e-05,
+      "loss": 2.7452,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934814814814815e-05,
+      "loss": 2.6479,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934518518518519e-05,
+      "loss": 2.7943,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.934222222222222e-05,
+      "loss": 2.6701,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.933925925925926e-05,
+      "loss": 2.6863,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.93362962962963e-05,
+      "loss": 2.6787,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.933333333333334e-05,
+      "loss": 2.771,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.933037037037037e-05,
+      "loss": 2.5961,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.932740740740741e-05,
+      "loss": 2.7753,
+      "step": 1135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.932444444444445e-05,
+      "loss": 2.5641,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9321481481481486e-05,
+      "loss": 2.685,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.931851851851852e-05,
+      "loss": 2.7166,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.931555555555556e-05,
+      "loss": 2.746,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9312592592592596e-05,
+      "loss": 2.6803,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9309629629629635e-05,
+      "loss": 2.6282,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.930666666666667e-05,
+      "loss": 2.6186,
+      "step": 1170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9303703703703705e-05,
+      "loss": 2.7369,
+      "step": 1175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9300740740740744e-05,
+      "loss": 2.6946,
+      "step": 1180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.929777777777778e-05,
+      "loss": 2.6654,
+      "step": 1185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9294814814814815e-05,
+      "loss": 2.6052,
+      "step": 1190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9291851851851854e-05,
+      "loss": 2.7831,
+      "step": 1195
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.928888888888889e-05,
+      "loss": 2.7227,
+      "step": 1200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.928592592592593e-05,
+      "loss": 2.7201,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9282962962962964e-05,
+      "loss": 2.6648,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.928e-05,
+      "loss": 2.6865,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.927703703703704e-05,
+      "loss": 2.664,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.927407407407408e-05,
+      "loss": 2.5916,
+      "step": 1225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.927111111111111e-05,
+      "loss": 2.6063,
+      "step": 1230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.926814814814815e-05,
+      "loss": 2.696,
+      "step": 1235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.926518518518518e-05,
+      "loss": 2.7361,
+      "step": 1240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.926222222222223e-05,
+      "loss": 2.6426,
+      "step": 1245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.925925925925926e-05,
+      "loss": 2.6998,
+      "step": 1250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.92562962962963e-05,
+      "loss": 2.7315,
+      "step": 1255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.925333333333333e-05,
+      "loss": 2.6739,
+      "step": 1260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.925037037037038e-05,
+      "loss": 2.6776,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.924740740740741e-05,
+      "loss": 2.7521,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.924444444444445e-05,
+      "loss": 2.6265,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.924148148148148e-05,
+      "loss": 2.774,
+      "step": 1280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9238518518518526e-05,
+      "loss": 2.6793,
+      "step": 1285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.923555555555556e-05,
+      "loss": 2.6703,
+      "step": 1290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9232592592592597e-05,
+      "loss": 2.6451,
+      "step": 1295
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.922962962962963e-05,
+      "loss": 2.6367,
+      "step": 1300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9226666666666674e-05,
+      "loss": 2.5895,
+      "step": 1305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9223703703703706e-05,
+      "loss": 2.5834,
+      "step": 1310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9220740740740745e-05,
+      "loss": 2.7616,
+      "step": 1315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.921777777777778e-05,
+      "loss": 2.6617,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9214814814814816e-05,
+      "loss": 2.7151,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9211851851851855e-05,
+      "loss": 2.6578,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9208888888888894e-05,
+      "loss": 2.5476,
+      "step": 1335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9205925925925926e-05,
+      "loss": 2.7617,
+      "step": 1340
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9202962962962964e-05,
+      "loss": 2.5194,
+      "step": 1345
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.92e-05,
+      "loss": 2.6243,
+      "step": 1350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.919703703703704e-05,
+      "loss": 2.542,
+      "step": 1355
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9194074074074074e-05,
+      "loss": 2.5248,
+      "step": 1360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.919111111111111e-05,
+      "loss": 2.5072,
+      "step": 1365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918814814814815e-05,
+      "loss": 2.6479,
+      "step": 1370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.918518518518519e-05,
+      "loss": 2.6212,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.918222222222222e-05,
+      "loss": 2.6752,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917925925925926e-05,
+      "loss": 2.6133,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.91762962962963e-05,
+      "loss": 2.5788,
+      "step": 1390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917333333333334e-05,
+      "loss": 2.5973,
+      "step": 1395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.917037037037037e-05,
+      "loss": 2.6873,
+      "step": 1400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.916740740740741e-05,
+      "loss": 2.5677,
+      "step": 1405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.916444444444445e-05,
+      "loss": 2.6275,
+      "step": 1410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.916148148148149e-05,
+      "loss": 2.6584,
+      "step": 1415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.915851851851852e-05,
+      "loss": 2.5752,
+      "step": 1420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.915555555555556e-05,
+      "loss": 2.5627,
+      "step": 1425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.91525925925926e-05,
+      "loss": 2.5348,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9149629629629636e-05,
+      "loss": 2.7585,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.914666666666667e-05,
+      "loss": 2.5969,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.914370370370371e-05,
+      "loss": 2.6326,
+      "step": 1445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9140740740740746e-05,
+      "loss": 2.5628,
+      "step": 1450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9137777777777785e-05,
+      "loss": 2.6056,
+      "step": 1455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.913481481481482e-05,
+      "loss": 2.6705,
+      "step": 1460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9131851851851856e-05,
+      "loss": 2.6942,
+      "step": 1465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.912888888888889e-05,
+      "loss": 2.6487,
+      "step": 1470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.912592592592593e-05,
+      "loss": 2.563,
+      "step": 1475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9122962962962965e-05,
+      "loss": 2.7142,
+      "step": 1480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 2.703,
+      "step": 1485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9117037037037036e-05,
+      "loss": 2.5956,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.911407407407408e-05,
+      "loss": 2.6318,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9111111111111114e-05,
+      "loss": 2.5268,
+      "step": 1500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.910814814814815e-05,
+      "loss": 2.6568,
+      "step": 1505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9105185185185185e-05,
+      "loss": 2.6327,
+      "step": 1510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.910222222222223e-05,
+      "loss": 2.6212,
+      "step": 1515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.909925925925926e-05,
+      "loss": 2.6328,
+      "step": 1520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.90962962962963e-05,
+      "loss": 2.5819,
+      "step": 1525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.909333333333333e-05,
+      "loss": 2.7559,
+      "step": 1530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.909037037037038e-05,
+      "loss": 2.6101,
+      "step": 1535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.908740740740741e-05,
+      "loss": 2.522,
+      "step": 1540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.908444444444445e-05,
+      "loss": 2.6315,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.908148148148148e-05,
+      "loss": 2.613,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.907851851851852e-05,
+      "loss": 2.6335,
+      "step": 1555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.907555555555556e-05,
+      "loss": 2.5468,
+      "step": 1560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.907259259259259e-05,
+      "loss": 2.7266,
+      "step": 1565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906962962962963e-05,
+      "loss": 2.6311,
+      "step": 1570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906666666666667e-05,
+      "loss": 2.5856,
+      "step": 1575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906370370370371e-05,
+      "loss": 2.6343,
+      "step": 1580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.906074074074074e-05,
+      "loss": 2.6214,
+      "step": 1585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.905777777777778e-05,
+      "loss": 2.6503,
+      "step": 1590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.905481481481482e-05,
+      "loss": 2.6477,
+      "step": 1595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9051851851851856e-05,
+      "loss": 2.6671,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.904888888888889e-05,
+      "loss": 2.6796,
+      "step": 1605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.904592592592593e-05,
+      "loss": 2.4912,
+      "step": 1610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9042962962962966e-05,
+      "loss": 2.5484,
+      "step": 1615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 2.7285,
+      "step": 1620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.903703703703704e-05,
+      "loss": 2.5839,
+      "step": 1625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9034074074074076e-05,
+      "loss": 2.6494,
+      "step": 1630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.903111111111111e-05,
+      "loss": 2.6646,
+      "step": 1635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.902814814814815e-05,
+      "loss": 2.7042,
+      "step": 1640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9025185185185185e-05,
+      "loss": 2.5844,
+      "step": 1645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9022222222222224e-05,
+      "loss": 2.5931,
+      "step": 1650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9019259259259256e-05,
+      "loss": 2.538,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.90162962962963e-05,
+      "loss": 2.6606,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9013333333333334e-05,
+      "loss": 2.5271,
+      "step": 1665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.901037037037037e-05,
+      "loss": 2.6466,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9007407407407405e-05,
+      "loss": 2.5798,
+      "step": 1675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.900444444444445e-05,
+      "loss": 2.5151,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.900148148148148e-05,
+      "loss": 2.5624,
+      "step": 1685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.899851851851852e-05,
+      "loss": 2.6799,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.899555555555555e-05,
+      "loss": 2.6345,
+      "step": 1695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.89925925925926e-05,
+      "loss": 2.5328,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.898962962962963e-05,
+      "loss": 2.5292,
+      "step": 1705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.898666666666667e-05,
+      "loss": 2.7206,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.89837037037037e-05,
+      "loss": 2.5661,
+      "step": 1715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.898074074074074e-05,
+      "loss": 2.6318,
+      "step": 1720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.897777777777778e-05,
+      "loss": 2.5574,
+      "step": 1725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.897481481481482e-05,
+      "loss": 2.541,
+      "step": 1730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.897185185185185e-05,
+      "loss": 2.5739,
+      "step": 1735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.896888888888889e-05,
+      "loss": 2.6641,
+      "step": 1740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.896592592592593e-05,
+      "loss": 2.543,
+      "step": 1745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.896296296296297e-05,
+      "loss": 2.7357,
+      "step": 1750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.896e-05,
+      "loss": 2.4066,
+      "step": 1755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.895703703703704e-05,
+      "loss": 2.5714,
+      "step": 1760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8954074074074076e-05,
+      "loss": 2.6102,
+      "step": 1765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.8951111111111115e-05,
+      "loss": 2.6916,
+      "step": 1770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.894814814814815e-05,
+      "loss": 2.5464,
+      "step": 1775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8945185185185186e-05,
+      "loss": 2.4816,
+      "step": 1780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8942222222222225e-05,
+      "loss": 2.5859,
+      "step": 1785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8939259259259264e-05,
+      "loss": 2.6595,
+      "step": 1790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8936296296296296e-05,
+      "loss": 2.5294,
+      "step": 1795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8933333333333335e-05,
+      "loss": 2.6837,
+      "step": 1800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8930370370370373e-05,
+      "loss": 2.6714,
+      "step": 1805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.892740740740741e-05,
+      "loss": 2.5805,
+      "step": 1810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8924444444444444e-05,
+      "loss": 2.682,
+      "step": 1815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.892148148148148e-05,
+      "loss": 2.6585,
+      "step": 1820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.891851851851852e-05,
+      "loss": 2.5288,
+      "step": 1825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.891555555555556e-05,
+      "loss": 2.46,
+      "step": 1830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.891259259259259e-05,
+      "loss": 2.5487,
+      "step": 1835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.890962962962963e-05,
+      "loss": 2.5528,
+      "step": 1840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.890666666666667e-05,
+      "loss": 2.7303,
+      "step": 1845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.890370370370371e-05,
+      "loss": 2.6666,
+      "step": 1850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.890074074074074e-05,
+      "loss": 2.5087,
+      "step": 1855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.889777777777778e-05,
+      "loss": 2.5803,
+      "step": 1860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.889481481481481e-05,
+      "loss": 2.6161,
+      "step": 1865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.889185185185186e-05,
+      "loss": 2.5119,
+      "step": 1870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 2.5282,
+      "step": 1875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.888592592592593e-05,
+      "loss": 2.5958,
+      "step": 1880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.888296296296296e-05,
+      "loss": 2.3793,
+      "step": 1885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8880000000000006e-05,
+      "loss": 2.6117,
+      "step": 1890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.887703703703704e-05,
+      "loss": 2.6272,
+      "step": 1895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.887407407407408e-05,
+      "loss": 2.5193,
+      "step": 1900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.887111111111111e-05,
+      "loss": 2.5786,
+      "step": 1905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8868148148148155e-05,
+      "loss": 2.5585,
+      "step": 1910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.886518518518519e-05,
+      "loss": 2.5675,
+      "step": 1915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8862222222222226e-05,
+      "loss": 2.6321,
+      "step": 1920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.885925925925926e-05,
+      "loss": 2.5583,
+      "step": 1925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.88562962962963e-05,
+      "loss": 2.6734,
+      "step": 1930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8853333333333335e-05,
+      "loss": 2.515,
+      "step": 1935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.8850370370370374e-05,
+      "loss": 2.5814,
+      "step": 1940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8847407407407406e-05,
+      "loss": 2.5155,
+      "step": 1945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8844444444444445e-05,
+      "loss": 2.5582,
+      "step": 1950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8841481481481484e-05,
+      "loss": 2.4236,
+      "step": 1955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.883851851851852e-05,
+      "loss": 2.5744,
+      "step": 1960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8835555555555555e-05,
+      "loss": 2.6236,
+      "step": 1965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.8832592592592594e-05,
+      "loss": 2.5213,
+      "step": 1970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.882962962962963e-05,
+      "loss": 2.4923,
+      "step": 1975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.882666666666667e-05,
+      "loss": 2.6818,
+      "step": 1980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.88237037037037e-05,
+      "loss": 2.5147,
+      "step": 1985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.882074074074074e-05,
+      "loss": 2.6033,
+      "step": 1990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.881777777777778e-05,
+      "loss": 2.5704,
+      "step": 1995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.881481481481482e-05,
+      "loss": 2.649,
+      "step": 2000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.881185185185185e-05,
+      "loss": 2.584,
+      "step": 2005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.880888888888889e-05,
+      "loss": 2.5052,
+      "step": 2010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.880592592592593e-05,
+      "loss": 2.6428,
+      "step": 2015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.880296296296297e-05,
+      "loss": 2.7282,
+      "step": 2020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.88e-05,
+      "loss": 2.5777,
+      "step": 2025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.879703703703704e-05,
+      "loss": 2.6538,
+      "step": 2030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.879407407407408e-05,
+      "loss": 2.647,
+      "step": 2035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.879111111111112e-05,
+      "loss": 2.6602,
+      "step": 2040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.878814814814815e-05,
+      "loss": 2.5872,
+      "step": 2045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.878518518518519e-05,
+      "loss": 2.6631,
+      "step": 2050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8782222222222226e-05,
+      "loss": 2.6267,
+      "step": 2055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8779259259259265e-05,
+      "loss": 2.5958,
+      "step": 2060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.87762962962963e-05,
+      "loss": 2.6223,
+      "step": 2065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8773333333333336e-05,
+      "loss": 2.6239,
+      "step": 2070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8770370370370375e-05,
+      "loss": 2.55,
+      "step": 2075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8767407407407414e-05,
+      "loss": 2.5085,
+      "step": 2080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8764444444444446e-05,
+      "loss": 2.626,
+      "step": 2085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8761481481481485e-05,
+      "loss": 2.398,
+      "step": 2090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.875851851851852e-05,
+      "loss": 2.5995,
+      "step": 2095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.875555555555556e-05,
+      "loss": 2.4188,
+      "step": 2100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.8752592592592594e-05,
+      "loss": 2.6131,
+      "step": 2105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.874962962962963e-05,
+      "loss": 2.6237,
+      "step": 2110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8746666666666665e-05,
+      "loss": 2.3157,
+      "step": 2115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.874370370370371e-05,
+      "loss": 2.555,
+      "step": 2120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.874074074074074e-05,
+      "loss": 2.7092,
+      "step": 2125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.873777777777778e-05,
+      "loss": 2.435,
+      "step": 2130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.8734814814814814e-05,
+      "loss": 2.6015,
+      "step": 2135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.873185185185186e-05,
+      "loss": 2.5091,
+      "step": 2140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.872888888888889e-05,
+      "loss": 2.5709,
+      "step": 2145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.872592592592593e-05,
+      "loss": 2.5891,
+      "step": 2150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.872296296296296e-05,
+      "loss": 2.5093,
+      "step": 2155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 2.6391,
+      "step": 2160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.871703703703704e-05,
+      "loss": 2.6089,
+      "step": 2165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871407407407408e-05,
+      "loss": 2.6249,
+      "step": 2170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.871111111111111e-05,
+      "loss": 2.5036,
+      "step": 2175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.870814814814815e-05,
+      "loss": 2.536,
+      "step": 2180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.870518518518519e-05,
+      "loss": 2.6776,
+      "step": 2185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.870222222222223e-05,
+      "loss": 2.5875,
+      "step": 2190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.869925925925926e-05,
+      "loss": 2.6116,
+      "step": 2195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.86962962962963e-05,
+      "loss": 2.4586,
+      "step": 2200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.869333333333334e-05,
+      "loss": 2.5113,
+      "step": 2205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.8690370370370376e-05,
+      "loss": 2.4781,
+      "step": 2210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.868740740740741e-05,
+      "loss": 2.4665,
+      "step": 2215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.868444444444445e-05,
+      "loss": 2.6852,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8681481481481485e-05,
+      "loss": 2.5498,
+      "step": 2225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8678518518518524e-05,
+      "loss": 2.5123,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8675555555555556e-05,
+      "loss": 2.4898,
+      "step": 2235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8672592592592595e-05,
+      "loss": 2.5807,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8669629629629634e-05,
+      "loss": 2.4785,
+      "step": 2245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.866666666666667e-05,
+      "loss": 2.5469,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8663703703703705e-05,
+      "loss": 2.6372,
+      "step": 2255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.8660740740740744e-05,
+      "loss": 2.566,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865777777777778e-05,
+      "loss": 2.5668,
+      "step": 2265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865481481481482e-05,
+      "loss": 2.5924,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.865185185185185e-05,
+      "loss": 2.5431,
+      "step": 2275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.864888888888889e-05,
+      "loss": 2.4167,
+      "step": 2280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.864592592592593e-05,
+      "loss": 2.6605,
+      "step": 2285
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.864296296296297e-05,
+      "loss": 2.4347,
+      "step": 2290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.864e-05,
+      "loss": 2.7029,
+      "step": 2295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.863703703703704e-05,
+      "loss": 2.6861,
+      "step": 2300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.863407407407408e-05,
+      "loss": 2.5973,
+      "step": 2305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.863111111111112e-05,
+      "loss": 2.5206,
+      "step": 2310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.862814814814815e-05,
+      "loss": 2.5048,
+      "step": 2315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.862518518518519e-05,
+      "loss": 2.5822,
+      "step": 2320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.862222222222222e-05,
+      "loss": 2.4442,
+      "step": 2325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.861925925925926e-05,
+      "loss": 2.6353,
+      "step": 2330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.86162962962963e-05,
+      "loss": 2.5626,
+      "step": 2335
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.861333333333333e-05,
+      "loss": 2.6273,
+      "step": 2340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.861037037037037e-05,
+      "loss": 2.5765,
+      "step": 2345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.860740740740741e-05,
+      "loss": 2.5701,
+      "step": 2350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.860444444444445e-05,
+      "loss": 2.6121,
+      "step": 2355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.860148148148148e-05,
+      "loss": 2.5889,
+      "step": 2360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.859851851851852e-05,
+      "loss": 2.4516,
+      "step": 2365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.859555555555556e-05,
+      "loss": 2.5998,
+      "step": 2370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8592592592592596e-05,
+      "loss": 2.5884,
+      "step": 2375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.858962962962963e-05,
+      "loss": 2.4922,
+      "step": 2380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.858666666666667e-05,
+      "loss": 2.6406,
+      "step": 2385
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.8583703703703706e-05,
+      "loss": 2.5813,
+      "step": 2390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8580740740740744e-05,
+      "loss": 2.5106,
+      "step": 2395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8577777777777776e-05,
+      "loss": 2.7063,
+      "step": 2400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8574814814814815e-05,
+      "loss": 2.56,
+      "step": 2405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8571851851851854e-05,
+      "loss": 2.6228,
+      "step": 2410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.856888888888889e-05,
+      "loss": 2.4815,
+      "step": 2415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8565925925925925e-05,
+      "loss": 2.5204,
+      "step": 2420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8562962962962964e-05,
+      "loss": 2.5944,
+      "step": 2425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.856e-05,
+      "loss": 2.5486,
+      "step": 2430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.855703703703704e-05,
+      "loss": 2.5694,
+      "step": 2435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8554074074074073e-05,
+      "loss": 2.4542,
+      "step": 2440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.855111111111111e-05,
+      "loss": 2.6608,
+      "step": 2445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.854814814814815e-05,
+      "loss": 2.5552,
+      "step": 2450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.854518518518519e-05,
+      "loss": 2.4182,
+      "step": 2455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.854222222222222e-05,
+      "loss": 2.7799,
+      "step": 2460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853925925925926e-05,
+      "loss": 2.5867,
+      "step": 2465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.85362962962963e-05,
+      "loss": 2.4412,
+      "step": 2470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853333333333334e-05,
+      "loss": 2.547,
+      "step": 2475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853037037037037e-05,
+      "loss": 2.474,
+      "step": 2480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.852740740740741e-05,
+      "loss": 2.4617,
+      "step": 2485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.852444444444444e-05,
+      "loss": 2.4407,
+      "step": 2490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.852148148148149e-05,
+      "loss": 2.4253,
+      "step": 2495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.851851851851852e-05,
+      "loss": 2.4985,
+      "step": 2500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.851555555555556e-05,
+      "loss": 2.6185,
+      "step": 2505
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.851259259259259e-05,
+      "loss": 2.4942,
+      "step": 2510
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8509629629629636e-05,
+      "loss": 2.5665,
+      "step": 2515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.850666666666667e-05,
+      "loss": 2.5831,
+      "step": 2520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8503703703703706e-05,
+      "loss": 2.5389,
+      "step": 2525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.850074074074074e-05,
+      "loss": 2.5858,
+      "step": 2530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8497777777777784e-05,
+      "loss": 2.6014,
+      "step": 2535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8494814814814816e-05,
+      "loss": 2.5959,
+      "step": 2540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8491851851851855e-05,
+      "loss": 2.4836,
+      "step": 2545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.848888888888889e-05,
+      "loss": 2.5167,
+      "step": 2550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.8485925925925926e-05,
+      "loss": 2.4104,
+      "step": 2555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8482962962962965e-05,
+      "loss": 2.697,
+      "step": 2560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 2.5179,
+      "step": 2565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8477037037037035e-05,
+      "loss": 2.5655,
+      "step": 2570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8474074074074074e-05,
+      "loss": 2.7141,
+      "step": 2575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.847111111111111e-05,
+      "loss": 2.5745,
+      "step": 2580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.846814814814815e-05,
+      "loss": 2.5983,
+      "step": 2585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.8465185185185184e-05,
+      "loss": 2.5184,
+      "step": 2590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.846222222222222e-05,
+      "loss": 2.5778,
+      "step": 2595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.845925925925926e-05,
+      "loss": 2.5206,
+      "step": 2600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.84562962962963e-05,
+      "loss": 2.4711,
+      "step": 2605
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.845333333333333e-05,
+      "loss": 2.4716,
+      "step": 2610
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.845037037037037e-05,
+      "loss": 2.4848,
+      "step": 2615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.844740740740741e-05,
+      "loss": 2.654,
+      "step": 2620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.844444444444445e-05,
+      "loss": 2.5274,
+      "step": 2625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.844148148148148e-05,
+      "loss": 2.4585,
+      "step": 2630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.843851851851852e-05,
+      "loss": 2.5476,
+      "step": 2635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.843555555555556e-05,
+      "loss": 2.393,
+      "step": 2640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.84325925925926e-05,
+      "loss": 2.4644,
+      "step": 2645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.842962962962963e-05,
+      "loss": 2.6004,
+      "step": 2650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.842666666666667e-05,
+      "loss": 2.4683,
+      "step": 2655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.842370370370371e-05,
+      "loss": 2.5443,
+      "step": 2660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.8420740740740746e-05,
+      "loss": 2.5539,
+      "step": 2665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.841777777777778e-05,
+      "loss": 2.5629,
+      "step": 2670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.841481481481482e-05,
+      "loss": 2.5614,
+      "step": 2675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8411851851851856e-05,
+      "loss": 2.569,
+      "step": 2680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8408888888888894e-05,
+      "loss": 2.4735,
+      "step": 2685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8405925925925927e-05,
+      "loss": 2.6352,
+      "step": 2690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8402962962962965e-05,
+      "loss": 2.4947,
+      "step": 2695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 2.6446,
+      "step": 2700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.839703703703704e-05,
+      "loss": 2.4544,
+      "step": 2705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8394074074074075e-05,
+      "loss": 2.5697,
+      "step": 2710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8391111111111114e-05,
+      "loss": 2.4467,
+      "step": 2715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.8388148148148146e-05,
+      "loss": 2.5275,
+      "step": 2720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.838518518518519e-05,
+      "loss": 2.5394,
+      "step": 2725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8382222222222224e-05,
+      "loss": 2.4864,
+      "step": 2730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.837925925925926e-05,
+      "loss": 2.5427,
+      "step": 2735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.8376296296296294e-05,
+      "loss": 2.5405,
+      "step": 2740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.837333333333334e-05,
+      "loss": 2.4612,
+      "step": 2745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.837037037037037e-05,
+      "loss": 2.4184,
+      "step": 2750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.836740740740741e-05,
+      "loss": 2.5487,
+      "step": 2755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.836444444444444e-05,
+      "loss": 2.4494,
+      "step": 2760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.836148148148149e-05,
+      "loss": 2.5665,
+      "step": 2765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.835851851851852e-05,
+      "loss": 2.5655,
+      "step": 2770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.835555555555556e-05,
+      "loss": 2.5634,
+      "step": 2775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.835259259259259e-05,
+      "loss": 2.6474,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.834962962962963e-05,
+      "loss": 2.5541,
+      "step": 2785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.834666666666667e-05,
+      "loss": 2.5239,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.834370370370371e-05,
+      "loss": 2.3692,
+      "step": 2795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.834074074074074e-05,
+      "loss": 2.4673,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.833777777777778e-05,
+      "loss": 2.5017,
+      "step": 2805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.833481481481482e-05,
+      "loss": 2.5517,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8331851851851856e-05,
+      "loss": 2.5868,
+      "step": 2815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.832888888888889e-05,
+      "loss": 2.5741,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.832592592592593e-05,
+      "loss": 2.6339,
+      "step": 2825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8322962962962966e-05,
+      "loss": 2.5491,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 2.5256,
+      "step": 2835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.831703703703704e-05,
+      "loss": 2.5913,
+      "step": 2840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8314074074074076e-05,
+      "loss": 2.5661,
+      "step": 2845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8311111111111115e-05,
+      "loss": 2.3706,
+      "step": 2850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8308148148148153e-05,
+      "loss": 2.5229,
+      "step": 2855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8305185185185185e-05,
+      "loss": 2.4967,
+      "step": 2860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8302222222222224e-05,
+      "loss": 2.3973,
+      "step": 2865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.829925925925926e-05,
+      "loss": 2.6179,
+      "step": 2870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.82962962962963e-05,
+      "loss": 2.3491,
+      "step": 2875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8293333333333334e-05,
+      "loss": 2.62,
+      "step": 2880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.829037037037037e-05,
+      "loss": 2.5978,
+      "step": 2885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.828740740740741e-05,
+      "loss": 2.4692,
+      "step": 2890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.828444444444445e-05,
+      "loss": 2.5157,
+      "step": 2895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.828148148148148e-05,
+      "loss": 2.4219,
+      "step": 2900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827851851851852e-05,
+      "loss": 2.6283,
+      "step": 2905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.827555555555556e-05,
+      "loss": 2.5258,
+      "step": 2910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.82725925925926e-05,
+      "loss": 2.5934,
+      "step": 2915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826962962962963e-05,
+      "loss": 2.5742,
+      "step": 2920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826666666666667e-05,
+      "loss": 2.6123,
+      "step": 2925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826370370370371e-05,
+      "loss": 2.5939,
+      "step": 2930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.826074074074075e-05,
+      "loss": 2.49,
+      "step": 2935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.825777777777778e-05,
+      "loss": 2.3457,
+      "step": 2940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.825481481481482e-05,
+      "loss": 2.4655,
+      "step": 2945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.825185185185185e-05,
+      "loss": 2.5718,
+      "step": 2950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8248888888888896e-05,
+      "loss": 2.4498,
+      "step": 2955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824592592592593e-05,
+      "loss": 2.5131,
+      "step": 2960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824296296296297e-05,
+      "loss": 2.3037,
+      "step": 2965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.824e-05,
+      "loss": 2.5468,
+      "step": 2970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8237037037037045e-05,
+      "loss": 2.5251,
+      "step": 2975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8234074074074077e-05,
+      "loss": 2.5881,
+      "step": 2980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8231111111111115e-05,
+      "loss": 2.478,
+      "step": 2985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822814814814815e-05,
+      "loss": 2.5001,
+      "step": 2990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822518518518519e-05,
+      "loss": 2.5638,
+      "step": 2995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8222222222222225e-05,
+      "loss": 2.4833,
+      "step": 3000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8219259259259264e-05,
+      "loss": 2.482,
+      "step": 3005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8216296296296296e-05,
+      "loss": 2.5302,
+      "step": 3010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8213333333333335e-05,
+      "loss": 2.5613,
+      "step": 3015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8210370370370374e-05,
+      "loss": 2.527,
+      "step": 3020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820740740740741e-05,
+      "loss": 2.5483,
+      "step": 3025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8204444444444444e-05,
+      "loss": 2.5656,
+      "step": 3030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820148148148148e-05,
+      "loss": 2.5175,
+      "step": 3035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819851851851852e-05,
+      "loss": 2.4651,
+      "step": 3040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819555555555556e-05,
+      "loss": 2.5214,
+      "step": 3045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819259259259259e-05,
+      "loss": 2.5157,
+      "step": 3050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818962962962963e-05,
+      "loss": 2.5873,
+      "step": 3055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818666666666667e-05,
+      "loss": 2.573,
+      "step": 3060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818370370370371e-05,
+      "loss": 2.5124,
+      "step": 3065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.818074074074074e-05,
+      "loss": 2.4542,
+      "step": 3070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.817777777777778e-05,
+      "loss": 2.5165,
+      "step": 3075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.817481481481482e-05,
+      "loss": 2.4215,
+      "step": 3080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.817185185185186e-05,
+      "loss": 2.5509,
+      "step": 3085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.816888888888889e-05,
+      "loss": 2.4778,
+      "step": 3090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.816592592592593e-05,
+      "loss": 2.4294,
+      "step": 3095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.816296296296297e-05,
+      "loss": 2.5117,
+      "step": 3100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.816e-05,
+      "loss": 2.6064,
+      "step": 3105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.815703703703704e-05,
+      "loss": 2.3965,
+      "step": 3110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.815407407407407e-05,
+      "loss": 2.634,
+      "step": 3115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.8151111111111116e-05,
+      "loss": 2.5048,
+      "step": 3120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 2.4657,
+      "step": 3125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.814518518518519e-05,
+      "loss": 2.5087,
+      "step": 3130
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.814222222222222e-05,
+      "loss": 2.4964,
+      "step": 3135
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8139259259259265e-05,
+      "loss": 2.5269,
+      "step": 3140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.81362962962963e-05,
+      "loss": 2.4399,
+      "step": 3145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8133333333333336e-05,
+      "loss": 2.5173,
+      "step": 3150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.813037037037037e-05,
+      "loss": 2.5539,
+      "step": 3155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.812740740740741e-05,
+      "loss": 2.5669,
+      "step": 3160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8124444444444445e-05,
+      "loss": 2.3558,
+      "step": 3165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8121481481481484e-05,
+      "loss": 2.604,
+      "step": 3170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8118518518518516e-05,
+      "loss": 2.366,
+      "step": 3175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8115555555555555e-05,
+      "loss": 2.584,
+      "step": 3180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8112592592592594e-05,
+      "loss": 2.6561,
+      "step": 3185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.810962962962963e-05,
+      "loss": 2.4862,
+      "step": 3190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8106666666666665e-05,
+      "loss": 2.5852,
+      "step": 3195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.8103703703703703e-05,
+      "loss": 2.5441,
+      "step": 3200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.810074074074074e-05,
+      "loss": 2.6158,
+      "step": 3205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.809777777777778e-05,
+      "loss": 2.464,
+      "step": 3210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.809481481481481e-05,
+      "loss": 2.5,
+      "step": 3215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.809185185185185e-05,
+      "loss": 2.5206,
+      "step": 3220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.808888888888889e-05,
+      "loss": 2.5357,
+      "step": 3225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.808592592592593e-05,
+      "loss": 2.4031,
+      "step": 3230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.808296296296296e-05,
+      "loss": 2.531,
+      "step": 3235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.808e-05,
+      "loss": 2.4502,
+      "step": 3240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.807703703703704e-05,
+      "loss": 2.5183,
+      "step": 3245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.807407407407408e-05,
+      "loss": 2.4353,
+      "step": 3250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.807111111111111e-05,
+      "loss": 2.4385,
+      "step": 3255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.806814814814815e-05,
+      "loss": 2.3361,
+      "step": 3260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.806518518518519e-05,
+      "loss": 2.4356,
+      "step": 3265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8062222222222227e-05,
+      "loss": 2.4506,
+      "step": 3270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.805925925925926e-05,
+      "loss": 2.4999,
+      "step": 3275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.80562962962963e-05,
+      "loss": 2.5567,
+      "step": 3280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8053333333333336e-05,
+      "loss": 2.5722,
+      "step": 3285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.8050370370370375e-05,
+      "loss": 2.4391,
+      "step": 3290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.804740740740741e-05,
+      "loss": 2.4832,
+      "step": 3295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8044444444444446e-05,
+      "loss": 2.457,
+      "step": 3300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8041481481481485e-05,
+      "loss": 2.4073,
+      "step": 3305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8038518518518524e-05,
+      "loss": 2.6618,
+      "step": 3310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8035555555555556e-05,
+      "loss": 2.5364,
+      "step": 3315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8032592592592595e-05,
+      "loss": 2.6259,
+      "step": 3320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.802962962962963e-05,
+      "loss": 2.5101,
+      "step": 3325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.802666666666667e-05,
+      "loss": 2.5214,
+      "step": 3330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8023703703703704e-05,
+      "loss": 2.622,
+      "step": 3335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.802074074074074e-05,
+      "loss": 2.6139,
+      "step": 3340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8017777777777775e-05,
+      "loss": 2.4534,
+      "step": 3345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.801481481481482e-05,
+      "loss": 2.5654,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.801185185185185e-05,
+      "loss": 2.5499,
+      "step": 3355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.800888888888889e-05,
+      "loss": 2.4748,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8005925925925924e-05,
+      "loss": 2.5733,
+      "step": 3365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.800296296296297e-05,
+      "loss": 2.4487,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8e-05,
+      "loss": 2.4621,
+      "step": 3375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799703703703704e-05,
+      "loss": 2.5535,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799407407407407e-05,
+      "loss": 2.4086,
+      "step": 3385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.799111111111112e-05,
+      "loss": 2.5602,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.798814814814815e-05,
+      "loss": 2.4961,
+      "step": 3395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.798518518518519e-05,
+      "loss": 2.471,
+      "step": 3400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.798222222222222e-05,
+      "loss": 2.3396,
+      "step": 3405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.797925925925926e-05,
+      "loss": 2.4646,
+      "step": 3410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.79762962962963e-05,
+      "loss": 2.449,
+      "step": 3415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.797333333333334e-05,
+      "loss": 2.5317,
+      "step": 3420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.797037037037037e-05,
+      "loss": 2.5041,
+      "step": 3425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.796740740740741e-05,
+      "loss": 2.4711,
+      "step": 3430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.796444444444445e-05,
+      "loss": 2.4812,
+      "step": 3435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7961481481481486e-05,
+      "loss": 2.4339,
+      "step": 3440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.795851851851852e-05,
+      "loss": 2.4439,
+      "step": 3445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7955555555555556e-05,
+      "loss": 2.4874,
+      "step": 3450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.7952592592592595e-05,
+      "loss": 2.5121,
+      "step": 3455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7949629629629634e-05,
+      "loss": 2.6459,
+      "step": 3460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7946666666666666e-05,
+      "loss": 2.6186,
+      "step": 3465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7943703703703705e-05,
+      "loss": 2.3716,
+      "step": 3470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7940740740740744e-05,
+      "loss": 2.507,
+      "step": 3475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.793777777777778e-05,
+      "loss": 2.3918,
+      "step": 3480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7934814814814815e-05,
+      "loss": 2.3803,
+      "step": 3485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.7931851851851853e-05,
+      "loss": 2.4445,
+      "step": 3490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.792888888888889e-05,
+      "loss": 2.4617,
+      "step": 3495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.792592592592593e-05,
+      "loss": 2.5057,
+      "step": 3500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.792296296296296e-05,
+      "loss": 2.5112,
+      "step": 3505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.792e-05,
+      "loss": 2.4289,
+      "step": 3510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.791703703703704e-05,
+      "loss": 2.5129,
+      "step": 3515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.791407407407408e-05,
+      "loss": 2.4718,
+      "step": 3520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.791111111111111e-05,
+      "loss": 2.4584,
+      "step": 3525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.790814814814815e-05,
+      "loss": 2.4819,
+      "step": 3530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.790518518518519e-05,
+      "loss": 2.5925,
+      "step": 3535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.790222222222223e-05,
+      "loss": 2.4326,
+      "step": 3540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.789925925925926e-05,
+      "loss": 2.4423,
+      "step": 3545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.78962962962963e-05,
+      "loss": 2.5685,
+      "step": 3550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.789333333333334e-05,
+      "loss": 2.4943,
+      "step": 3555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.789037037037038e-05,
+      "loss": 2.6227,
+      "step": 3560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.788740740740741e-05,
+      "loss": 2.373,
+      "step": 3565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.788444444444445e-05,
+      "loss": 2.3402,
+      "step": 3570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.788148148148148e-05,
+      "loss": 2.5824,
+      "step": 3575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7878518518518525e-05,
+      "loss": 2.4371,
+      "step": 3580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.787555555555556e-05,
+      "loss": 2.3421,
+      "step": 3585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7872592592592596e-05,
+      "loss": 2.333,
+      "step": 3590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.786962962962963e-05,
+      "loss": 2.4595,
+      "step": 3595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7866666666666674e-05,
+      "loss": 2.4442,
+      "step": 3600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7863703703703706e-05,
+      "loss": 2.3995,
+      "step": 3605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7860740740740745e-05,
+      "loss": 2.4855,
+      "step": 3610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7857777777777777e-05,
+      "loss": 2.5998,
+      "step": 3615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.785481481481482e-05,
+      "loss": 2.4413,
+      "step": 3620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.7851851851851854e-05,
+      "loss": 2.5083,
+      "step": 3625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.784888888888889e-05,
+      "loss": 2.4819,
+      "step": 3630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7845925925925925e-05,
+      "loss": 2.5796,
+      "step": 3635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7842962962962964e-05,
+      "loss": 2.458,
+      "step": 3640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.784e-05,
+      "loss": 2.5684,
+      "step": 3645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.783703703703704e-05,
+      "loss": 2.3808,
+      "step": 3650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.7834074074074074e-05,
+      "loss": 2.4314,
+      "step": 3655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.783111111111111e-05,
+      "loss": 2.2501,
+      "step": 3660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.782814814814815e-05,
+      "loss": 2.5552,
+      "step": 3665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.782518518518519e-05,
+      "loss": 2.506,
+      "step": 3670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.782222222222222e-05,
+      "loss": 2.5063,
+      "step": 3675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.781925925925926e-05,
+      "loss": 2.4438,
+      "step": 3680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.78162962962963e-05,
+      "loss": 2.4133,
+      "step": 3685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.781333333333334e-05,
+      "loss": 2.5003,
+      "step": 3690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.781037037037037e-05,
+      "loss": 2.4661,
+      "step": 3695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.780740740740741e-05,
+      "loss": 2.4338,
+      "step": 3700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.780444444444445e-05,
+      "loss": 2.6581,
+      "step": 3705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.780148148148149e-05,
+      "loss": 2.5527,
+      "step": 3710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.779851851851852e-05,
+      "loss": 2.4698,
+      "step": 3715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.779555555555556e-05,
+      "loss": 2.4999,
+      "step": 3720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.77925925925926e-05,
+      "loss": 2.5276,
+      "step": 3725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7789629629629636e-05,
+      "loss": 2.4254,
+      "step": 3730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.778666666666667e-05,
+      "loss": 2.5416,
+      "step": 3735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.7783703703703706e-05,
+      "loss": 2.3298,
+      "step": 3740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7780740740740745e-05,
+      "loss": 2.5605,
+      "step": 3745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 2.5469,
+      "step": 3750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7774814814814816e-05,
+      "loss": 2.5102,
+      "step": 3755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7771851851851855e-05,
+      "loss": 2.4916,
+      "step": 3760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7768888888888894e-05,
+      "loss": 2.4455,
+      "step": 3765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.776592592592593e-05,
+      "loss": 2.4469,
+      "step": 3770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7762962962962965e-05,
+      "loss": 2.5556,
+      "step": 3775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 2.498,
+      "step": 3780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.775703703703704e-05,
+      "loss": 2.4771,
+      "step": 3785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.775407407407408e-05,
+      "loss": 2.4732,
+      "step": 3790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.775111111111111e-05,
+      "loss": 2.4418,
+      "step": 3795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774814814814815e-05,
+      "loss": 2.3981,
+      "step": 3800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7745185185185184e-05,
+      "loss": 2.4527,
+      "step": 3805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.774222222222223e-05,
+      "loss": 2.5408,
+      "step": 3810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773925925925926e-05,
+      "loss": 2.372,
+      "step": 3815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.77362962962963e-05,
+      "loss": 2.3876,
+      "step": 3820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773333333333333e-05,
+      "loss": 2.3631,
+      "step": 3825
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773037037037038e-05,
+      "loss": 2.3707,
+      "step": 3830
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772740740740741e-05,
+      "loss": 2.3632,
+      "step": 3835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772444444444445e-05,
+      "loss": 2.4309,
+      "step": 3840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772148148148148e-05,
+      "loss": 2.4249,
+      "step": 3845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771851851851853e-05,
+      "loss": 2.5575,
+      "step": 3850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.771555555555556e-05,
+      "loss": 2.4422,
+      "step": 3855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.77125925925926e-05,
+      "loss": 2.4859,
+      "step": 3860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770962962962963e-05,
+      "loss": 2.4379,
+      "step": 3865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770666666666667e-05,
+      "loss": 2.5409,
+      "step": 3870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770370370370371e-05,
+      "loss": 2.4814,
+      "step": 3875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.770074074074074e-05,
+      "loss": 2.5844,
+      "step": 3880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.769777777777778e-05,
+      "loss": 2.49,
+      "step": 3885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.769481481481482e-05,
+      "loss": 2.4241,
+      "step": 3890
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7691851851851856e-05,
+      "loss": 2.4844,
+      "step": 3895
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768888888888889e-05,
+      "loss": 2.4101,
+      "step": 3900
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768592592592593e-05,
+      "loss": 2.4722,
+      "step": 3905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7682962962962965e-05,
+      "loss": 2.3843,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7680000000000004e-05,
+      "loss": 2.3913,
+      "step": 3915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7677037037037036e-05,
+      "loss": 2.6144,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7674074074074075e-05,
+      "loss": 2.5613,
+      "step": 3925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7671111111111114e-05,
+      "loss": 2.4305,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.766814814814815e-05,
+      "loss": 2.4208,
+      "step": 3935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7665185185185185e-05,
+      "loss": 2.4655,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7662222222222224e-05,
+      "loss": 2.3634,
+      "step": 3945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.7659259259259256e-05,
+      "loss": 2.4413,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.76562962962963e-05,
+      "loss": 2.5993,
+      "step": 3955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.765333333333333e-05,
+      "loss": 2.4721,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.765037037037037e-05,
+      "loss": 2.5053,
+      "step": 3965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.7647407407407404e-05,
+      "loss": 2.5659,
+      "step": 3970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.764444444444445e-05,
+      "loss": 2.4318,
+      "step": 3975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.764148148148148e-05,
+      "loss": 2.4853,
+      "step": 3980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.763851851851852e-05,
+      "loss": 2.3437,
+      "step": 3985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.763555555555555e-05,
+      "loss": 2.4864,
+      "step": 3990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.76325925925926e-05,
+      "loss": 2.3076,
+      "step": 3995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.762962962962963e-05,
+      "loss": 2.5107,
+      "step": 4000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.762666666666667e-05,
+      "loss": 2.3943,
+      "step": 4005
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.76237037037037e-05,
+      "loss": 2.5734,
+      "step": 4010
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.762074074074075e-05,
+      "loss": 2.5178,
+      "step": 4015
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.761777777777778e-05,
+      "loss": 2.3443,
+      "step": 4020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.761481481481482e-05,
+      "loss": 2.4504,
+      "step": 4025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.761244444444445e-05,
+      "loss": 2.5104,
+      "step": 4030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.760948148148148e-05,
+      "loss": 2.5564,
+      "step": 4035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7606518518518525e-05,
+      "loss": 2.4881,
+      "step": 4040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.760355555555556e-05,
+      "loss": 2.4379,
+      "step": 4045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7600592592592596e-05,
+      "loss": 2.5927,
+      "step": 4050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.759762962962963e-05,
+      "loss": 2.434,
+      "step": 4055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7594666666666674e-05,
+      "loss": 2.4309,
+      "step": 4060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7591703703703706e-05,
+      "loss": 2.4294,
+      "step": 4065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7588740740740744e-05,
+      "loss": 2.4577,
+      "step": 4070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.7585777777777776e-05,
+      "loss": 2.3474,
+      "step": 4075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.758281481481482e-05,
+      "loss": 2.4289,
+      "step": 4080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7579851851851854e-05,
+      "loss": 2.5407,
+      "step": 4085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.757688888888889e-05,
+      "loss": 2.3306,
+      "step": 4090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7573925925925925e-05,
+      "loss": 2.6044,
+      "step": 4095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7570962962962964e-05,
+      "loss": 2.3461,
+      "step": 4100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7568e-05,
+      "loss": 2.4643,
+      "step": 4105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.756503703703704e-05,
+      "loss": 2.4313,
+      "step": 4110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.7562074074074074e-05,
+      "loss": 2.3669,
+      "step": 4115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.755911111111111e-05,
+      "loss": 2.5477,
+      "step": 4120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.755614814814815e-05,
+      "loss": 2.512,
+      "step": 4125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.755318518518519e-05,
+      "loss": 2.342,
+      "step": 4130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.755022222222222e-05,
+      "loss": 2.6974,
+      "step": 4135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.754725925925926e-05,
+      "loss": 2.3741,
+      "step": 4140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.75442962962963e-05,
+      "loss": 2.4483,
+      "step": 4145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.754133333333334e-05,
+      "loss": 2.514,
+      "step": 4150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.753837037037037e-05,
+      "loss": 2.4418,
+      "step": 4155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.753540740740741e-05,
+      "loss": 2.5164,
+      "step": 4160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.753244444444445e-05,
+      "loss": 2.3921,
+      "step": 4165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.752948148148149e-05,
+      "loss": 2.5978,
+      "step": 4170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.752651851851852e-05,
+      "loss": 2.2988,
+      "step": 4175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.752355555555556e-05,
+      "loss": 2.5154,
+      "step": 4180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.75205925925926e-05,
+      "loss": 2.4891,
+      "step": 4185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.7517629629629636e-05,
+      "loss": 2.3978,
+      "step": 4190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.751466666666667e-05,
+      "loss": 2.5693,
+      "step": 4195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7511703703703706e-05,
+      "loss": 2.5582,
+      "step": 4200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7508740740740745e-05,
+      "loss": 2.5992,
+      "step": 4205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7505777777777784e-05,
+      "loss": 2.5169,
+      "step": 4210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7502814814814816e-05,
+      "loss": 2.4072,
+      "step": 4215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7499851851851855e-05,
+      "loss": 2.6555,
+      "step": 4220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7496888888888894e-05,
+      "loss": 2.4927,
+      "step": 4225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.749392592592593e-05,
+      "loss": 2.4171,
+      "step": 4230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7490962962962965e-05,
+      "loss": 2.5243,
+      "step": 4235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.7488000000000003e-05,
+      "loss": 2.5331,
+      "step": 4240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.748503703703704e-05,
+      "loss": 2.5703,
+      "step": 4245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.748207407407408e-05,
+      "loss": 2.5743,
+      "step": 4250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.747911111111111e-05,
+      "loss": 2.2319,
+      "step": 4255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.747614814814815e-05,
+      "loss": 2.4138,
+      "step": 4260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7473185185185184e-05,
+      "loss": 2.2671,
+      "step": 4265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.747022222222223e-05,
+      "loss": 2.551,
+      "step": 4270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.746725925925926e-05,
+      "loss": 2.3967,
+      "step": 4275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.74642962962963e-05,
+      "loss": 2.5029,
+      "step": 4280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.746133333333333e-05,
+      "loss": 2.5147,
+      "step": 4285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.745837037037038e-05,
+      "loss": 2.5134,
+      "step": 4290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.745540740740741e-05,
+      "loss": 2.5816,
+      "step": 4295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.745244444444445e-05,
+      "loss": 2.423,
+      "step": 4300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.744948148148148e-05,
+      "loss": 2.4048,
+      "step": 4305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7446518518518527e-05,
+      "loss": 2.4497,
+      "step": 4310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.744355555555556e-05,
+      "loss": 2.5144,
+      "step": 4315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.74405925925926e-05,
+      "loss": 2.4839,
+      "step": 4320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743762962962963e-05,
+      "loss": 2.4039,
+      "step": 4325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743466666666667e-05,
+      "loss": 2.5328,
+      "step": 4330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.743170370370371e-05,
+      "loss": 2.4651,
+      "step": 4335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742874074074074e-05,
+      "loss": 2.4951,
+      "step": 4340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742577777777778e-05,
+      "loss": 2.3968,
+      "step": 4345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742281481481482e-05,
+      "loss": 2.4837,
+      "step": 4350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7419851851851856e-05,
+      "loss": 2.3474,
+      "step": 4355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.741688888888889e-05,
+      "loss": 2.4129,
+      "step": 4360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7413925925925927e-05,
+      "loss": 2.54,
+      "step": 4365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7410962962962965e-05,
+      "loss": 2.3878,
+      "step": 4370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7408000000000004e-05,
+      "loss": 2.444,
+      "step": 4375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7405037037037036e-05,
+      "loss": 2.4356,
+      "step": 4380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7402074074074075e-05,
+      "loss": 2.5181,
+      "step": 4385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7399111111111114e-05,
+      "loss": 2.465,
+      "step": 4390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.739614814814815e-05,
+      "loss": 2.4831,
+      "step": 4395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7393185185185185e-05,
+      "loss": 2.5451,
+      "step": 4400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7390222222222224e-05,
+      "loss": 2.4309,
+      "step": 4405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.738725925925926e-05,
+      "loss": 2.3851,
+      "step": 4410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.73842962962963e-05,
+      "loss": 2.2616,
+      "step": 4415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.738133333333333e-05,
+      "loss": 2.3663,
+      "step": 4420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.737837037037037e-05,
+      "loss": 2.448,
+      "step": 4425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7375407407407404e-05,
+      "loss": 2.4346,
+      "step": 4430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.737244444444445e-05,
+      "loss": 2.3232,
+      "step": 4435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.736948148148148e-05,
+      "loss": 2.3551,
+      "step": 4440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.736651851851852e-05,
+      "loss": 2.4111,
+      "step": 4445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.736355555555555e-05,
+      "loss": 2.5738,
+      "step": 4450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.73605925925926e-05,
+      "loss": 2.5933,
+      "step": 4455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.735762962962963e-05,
+      "loss": 2.4173,
+      "step": 4460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.735466666666667e-05,
+      "loss": 2.5098,
+      "step": 4465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.73517037037037e-05,
+      "loss": 2.4746,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.734874074074075e-05,
+      "loss": 2.4557,
+      "step": 4475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.734577777777778e-05,
+      "loss": 2.564,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.734281481481482e-05,
+      "loss": 2.5229,
+      "step": 4485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.733985185185185e-05,
+      "loss": 2.545,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.733688888888889e-05,
+      "loss": 2.4309,
+      "step": 4495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.733392592592593e-05,
+      "loss": 2.406,
+      "step": 4500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7330962962962966e-05,
+      "loss": 2.4441,
+      "step": 4505
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7328e-05,
+      "loss": 2.4585,
+      "step": 4510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.732503703703704e-05,
+      "loss": 2.4851,
+      "step": 4515
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7322074074074076e-05,
+      "loss": 2.4068,
+      "step": 4520
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7319111111111115e-05,
+      "loss": 2.4573,
+      "step": 4525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.731614814814815e-05,
+      "loss": 2.3281,
+      "step": 4530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7313185185185185e-05,
+      "loss": 2.4596,
+      "step": 4535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7310222222222224e-05,
+      "loss": 2.3728,
+      "step": 4540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.730725925925926e-05,
+      "loss": 2.4934,
+      "step": 4545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7304296296296295e-05,
+      "loss": 2.4534,
+      "step": 4550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7301333333333334e-05,
+      "loss": 2.4465,
+      "step": 4555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.729837037037037e-05,
+      "loss": 2.5807,
+      "step": 4560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.729540740740741e-05,
+      "loss": 2.467,
+      "step": 4565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7292444444444444e-05,
+      "loss": 2.5156,
+      "step": 4570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.728948148148148e-05,
+      "loss": 2.4047,
+      "step": 4575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.728651851851852e-05,
+      "loss": 2.4475,
+      "step": 4580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.728355555555556e-05,
+      "loss": 2.4209,
+      "step": 4585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.728059259259259e-05,
+      "loss": 2.4052,
+      "step": 4590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.727762962962963e-05,
+      "loss": 2.498,
+      "step": 4595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.727466666666667e-05,
+      "loss": 2.2793,
+      "step": 4600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.727170370370371e-05,
+      "loss": 2.5308,
+      "step": 4605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.726874074074074e-05,
+      "loss": 2.4118,
+      "step": 4610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.726577777777778e-05,
+      "loss": 2.3792,
+      "step": 4615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.726281481481482e-05,
+      "loss": 2.4621,
+      "step": 4620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.725985185185186e-05,
+      "loss": 2.5024,
+      "step": 4625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.725688888888889e-05,
+      "loss": 2.3679,
+      "step": 4630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.725392592592593e-05,
+      "loss": 2.1793,
+      "step": 4635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.725096296296297e-05,
+      "loss": 2.4323,
+      "step": 4640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7248000000000006e-05,
+      "loss": 2.4791,
+      "step": 4645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.724503703703704e-05,
+      "loss": 2.3062,
+      "step": 4650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7242074074074077e-05,
+      "loss": 2.4608,
+      "step": 4655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.723911111111111e-05,
+      "loss": 2.4645,
+      "step": 4660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7236148148148154e-05,
+      "loss": 2.4516,
+      "step": 4665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7233185185185186e-05,
+      "loss": 2.323,
+      "step": 4670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7230222222222225e-05,
+      "loss": 2.4933,
+      "step": 4675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.722725925925926e-05,
+      "loss": 2.4274,
+      "step": 4680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.72242962962963e-05,
+      "loss": 2.4182,
+      "step": 4685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7221333333333335e-05,
+      "loss": 2.3939,
+      "step": 4690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7218370370370374e-05,
+      "loss": 2.5969,
+      "step": 4695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7215407407407406e-05,
+      "loss": 2.4727,
+      "step": 4700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.721244444444445e-05,
+      "loss": 2.5203,
+      "step": 4705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.720948148148148e-05,
+      "loss": 2.4477,
+      "step": 4710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.720651851851852e-05,
+      "loss": 2.5023,
+      "step": 4715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7203555555555554e-05,
+      "loss": 2.5075,
+      "step": 4720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.720059259259259e-05,
+      "loss": 2.4749,
+      "step": 4725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.719762962962963e-05,
+      "loss": 2.4227,
+      "step": 4730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.719466666666667e-05,
+      "loss": 2.3535,
+      "step": 4735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.71917037037037e-05,
+      "loss": 2.5053,
+      "step": 4740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.718874074074074e-05,
+      "loss": 2.2861,
+      "step": 4745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.718577777777778e-05,
+      "loss": 2.435,
+      "step": 4750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.718281481481482e-05,
+      "loss": 2.3364,
+      "step": 4755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.717985185185185e-05,
+      "loss": 2.444,
+      "step": 4760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.717688888888889e-05,
+      "loss": 2.4585,
+      "step": 4765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.717392592592593e-05,
+      "loss": 2.3898,
+      "step": 4770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.717096296296297e-05,
+      "loss": 2.4509,
+      "step": 4775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7168e-05,
+      "loss": 2.4245,
+      "step": 4780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.716503703703704e-05,
+      "loss": 2.5839,
+      "step": 4785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.716207407407408e-05,
+      "loss": 2.3871,
+      "step": 4790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7159111111111116e-05,
+      "loss": 2.2953,
+      "step": 4795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.715614814814815e-05,
+      "loss": 2.3996,
+      "step": 4800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.715318518518519e-05,
+      "loss": 2.3379,
+      "step": 4805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7150222222222226e-05,
+      "loss": 2.4048,
+      "step": 4810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7147259259259265e-05,
+      "loss": 2.3484,
+      "step": 4815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.71442962962963e-05,
+      "loss": 2.5595,
+      "step": 4820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7141333333333336e-05,
+      "loss": 2.4604,
+      "step": 4825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7138370370370374e-05,
+      "loss": 2.4274,
+      "step": 4830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.713540740740741e-05,
+      "loss": 2.3699,
+      "step": 4835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7132444444444445e-05,
+      "loss": 2.3518,
+      "step": 4840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7129481481481484e-05,
+      "loss": 2.4632,
+      "step": 4845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.712651851851852e-05,
+      "loss": 2.3874,
+      "step": 4850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.712355555555556e-05,
+      "loss": 2.3997,
+      "step": 4855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7120592592592594e-05,
+      "loss": 2.5184,
+      "step": 4860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.711762962962963e-05,
+      "loss": 2.4227,
+      "step": 4865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.711466666666667e-05,
+      "loss": 2.4983,
+      "step": 4870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.711170370370371e-05,
+      "loss": 2.4304,
+      "step": 4875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.710874074074074e-05,
+      "loss": 2.4887,
+      "step": 4880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.710577777777778e-05,
+      "loss": 2.5103,
+      "step": 4885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.710281481481481e-05,
+      "loss": 2.3891,
+      "step": 4890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.709985185185186e-05,
+      "loss": 2.4116,
+      "step": 4895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.709688888888889e-05,
+      "loss": 2.4161,
+      "step": 4900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.709392592592593e-05,
+      "loss": 2.581,
+      "step": 4905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.709096296296296e-05,
+      "loss": 2.3946,
+      "step": 4910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.708800000000001e-05,
+      "loss": 2.3597,
+      "step": 4915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.708503703703704e-05,
+      "loss": 2.3345,
+      "step": 4920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.708207407407408e-05,
+      "loss": 2.3779,
+      "step": 4925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.707911111111111e-05,
+      "loss": 2.4762,
+      "step": 4930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7076148148148156e-05,
+      "loss": 2.4625,
+      "step": 4935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.707318518518519e-05,
+      "loss": 2.4128,
+      "step": 4940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.707022222222223e-05,
+      "loss": 2.4835,
+      "step": 4945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.706725925925926e-05,
+      "loss": 2.3698,
+      "step": 4950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.70642962962963e-05,
+      "loss": 2.4273,
+      "step": 4955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7061333333333336e-05,
+      "loss": 2.7041,
+      "step": 4960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7058370370370375e-05,
+      "loss": 2.3832,
+      "step": 4965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.705540740740741e-05,
+      "loss": 2.371,
+      "step": 4970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7052444444444446e-05,
+      "loss": 2.3734,
+      "step": 4975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7049481481481485e-05,
+      "loss": 2.414,
+      "step": 4980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7046518518518524e-05,
+      "loss": 2.5053,
+      "step": 4985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7043555555555556e-05,
+      "loss": 2.4202,
+      "step": 4990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7040592592592595e-05,
+      "loss": 2.3025,
+      "step": 4995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.703762962962963e-05,
+      "loss": 2.4485,
+      "step": 5000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.703466666666667e-05,
+      "loss": 2.4395,
+      "step": 5005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7031703703703704e-05,
+      "loss": 2.4186,
+      "step": 5010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.702874074074074e-05,
+      "loss": 2.4557,
+      "step": 5015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.702577777777778e-05,
+      "loss": 2.5432,
+      "step": 5020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.702281481481482e-05,
+      "loss": 2.3154,
+      "step": 5025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.701985185185185e-05,
+      "loss": 2.4949,
+      "step": 5030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701688888888889e-05,
+      "loss": 2.3995,
+      "step": 5035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701392592592593e-05,
+      "loss": 2.4746,
+      "step": 5040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701096296296297e-05,
+      "loss": 2.4757,
+      "step": 5045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7008e-05,
+      "loss": 2.4692,
+      "step": 5050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700503703703704e-05,
+      "loss": 2.3452,
+      "step": 5055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700207407407408e-05,
+      "loss": 2.5242,
+      "step": 5060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699911111111112e-05,
+      "loss": 2.4604,
+      "step": 5065
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699614814814815e-05,
+      "loss": 2.3832,
+      "step": 5070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699318518518519e-05,
+      "loss": 2.3376,
+      "step": 5075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.699022222222223e-05,
+      "loss": 2.4679,
+      "step": 5080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6987259259259266e-05,
+      "loss": 2.3359,
+      "step": 5085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.69842962962963e-05,
+      "loss": 2.3866,
+      "step": 5090
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.698133333333334e-05,
+      "loss": 2.3574,
+      "step": 5095
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6978370370370376e-05,
+      "loss": 2.4682,
+      "step": 5100
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6975407407407415e-05,
+      "loss": 2.4025,
+      "step": 5105
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.697244444444445e-05,
+      "loss": 2.2214,
+      "step": 5110
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696948148148148e-05,
+      "loss": 2.4172,
+      "step": 5115
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696651851851852e-05,
+      "loss": 2.3537,
+      "step": 5120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6963555555555556e-05,
+      "loss": 2.4709,
+      "step": 5125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6960592592592595e-05,
+      "loss": 2.2421,
+      "step": 5130
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.695762962962963e-05,
+      "loss": 2.4343,
+      "step": 5135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6954666666666666e-05,
+      "loss": 2.4245,
+      "step": 5140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6951703703703705e-05,
+      "loss": 2.3552,
+      "step": 5145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6948740740740744e-05,
+      "loss": 2.4264,
+      "step": 5150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6945777777777776e-05,
+      "loss": 2.3034,
+      "step": 5155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6942814814814815e-05,
+      "loss": 2.4818,
+      "step": 5160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6939851851851853e-05,
+      "loss": 2.4698,
+      "step": 5165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693688888888889e-05,
+      "loss": 2.3307,
+      "step": 5170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6933925925925924e-05,
+      "loss": 2.4316,
+      "step": 5175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693096296296296e-05,
+      "loss": 2.4789,
+      "step": 5180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6928e-05,
+      "loss": 2.4746,
+      "step": 5185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692503703703704e-05,
+      "loss": 2.3931,
+      "step": 5190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692207407407407e-05,
+      "loss": 2.3844,
+      "step": 5195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691911111111111e-05,
+      "loss": 2.4915,
+      "step": 5200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691614814814815e-05,
+      "loss": 2.3076,
+      "step": 5205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691318518518519e-05,
+      "loss": 2.508,
+      "step": 5210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691022222222222e-05,
+      "loss": 2.369,
+      "step": 5215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690725925925926e-05,
+      "loss": 2.3622,
+      "step": 5220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.69042962962963e-05,
+      "loss": 2.5399,
+      "step": 5225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690133333333334e-05,
+      "loss": 2.3469,
+      "step": 5230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689837037037037e-05,
+      "loss": 2.3712,
+      "step": 5235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689540740740741e-05,
+      "loss": 2.4749,
+      "step": 5240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689244444444445e-05,
+      "loss": 2.4072,
+      "step": 5245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6889481481481486e-05,
+      "loss": 2.327,
+      "step": 5250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688651851851852e-05,
+      "loss": 2.4527,
+      "step": 5255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.688355555555556e-05,
+      "loss": 2.3353,
+      "step": 5260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.688059259259259e-05,
+      "loss": 2.3977,
+      "step": 5265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6877629629629635e-05,
+      "loss": 2.5054,
+      "step": 5270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.687466666666667e-05,
+      "loss": 2.4468,
+      "step": 5275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6871703703703706e-05,
+      "loss": 2.3576,
+      "step": 5280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.686874074074074e-05,
+      "loss": 2.3727,
+      "step": 5285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.686577777777778e-05,
+      "loss": 2.3217,
+      "step": 5290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6862814814814815e-05,
+      "loss": 2.3375,
+      "step": 5295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6859851851851854e-05,
+      "loss": 2.3502,
+      "step": 5300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6856888888888886e-05,
+      "loss": 2.4503,
+      "step": 5305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.685392592592593e-05,
+      "loss": 2.3362,
+      "step": 5310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6850962962962964e-05,
+      "loss": 2.4725,
+      "step": 5315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6848e-05,
+      "loss": 2.5309,
+      "step": 5320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6845037037037035e-05,
+      "loss": 2.3586,
+      "step": 5325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.684207407407408e-05,
+      "loss": 2.3396,
+      "step": 5330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.683911111111111e-05,
+      "loss": 2.4974,
+      "step": 5335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.683614814814815e-05,
+      "loss": 2.3944,
+      "step": 5340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.683318518518518e-05,
+      "loss": 2.3822,
+      "step": 5345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.683022222222222e-05,
+      "loss": 2.536,
+      "step": 5350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.682725925925926e-05,
+      "loss": 2.4728,
+      "step": 5355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.68242962962963e-05,
+      "loss": 2.5047,
+      "step": 5360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.682133333333333e-05,
+      "loss": 2.3307,
+      "step": 5365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.681837037037037e-05,
+      "loss": 2.3908,
+      "step": 5370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.681540740740741e-05,
+      "loss": 2.4032,
+      "step": 5375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.681244444444445e-05,
+      "loss": 2.3613,
+      "step": 5380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.680948148148148e-05,
+      "loss": 2.4218,
+      "step": 5385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.680651851851852e-05,
+      "loss": 2.2862,
+      "step": 5390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.680355555555556e-05,
+      "loss": 2.2378,
+      "step": 5395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.68005925925926e-05,
+      "loss": 2.4057,
+      "step": 5400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.679762962962963e-05,
+      "loss": 2.3884,
+      "step": 5405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.679466666666667e-05,
+      "loss": 2.5152,
+      "step": 5410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6791703703703706e-05,
+      "loss": 2.4994,
+      "step": 5415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6788740740740745e-05,
+      "loss": 2.4554,
+      "step": 5420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.678577777777778e-05,
+      "loss": 2.5549,
+      "step": 5425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6782814814814816e-05,
+      "loss": 2.4477,
+      "step": 5430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6779851851851855e-05,
+      "loss": 2.401,
+      "step": 5435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6776888888888894e-05,
+      "loss": 2.4176,
+      "step": 5440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6773925925925926e-05,
+      "loss": 2.539,
+      "step": 5445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6770962962962965e-05,
+      "loss": 2.44,
+      "step": 5450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6768000000000004e-05,
+      "loss": 2.4583,
+      "step": 5455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.676503703703704e-05,
+      "loss": 2.294,
+      "step": 5460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6762074074074074e-05,
+      "loss": 2.3093,
+      "step": 5465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.675911111111111e-05,
+      "loss": 2.2936,
+      "step": 5470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.675614814814815e-05,
+      "loss": 2.4079,
+      "step": 5475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.675318518518519e-05,
+      "loss": 2.4775,
+      "step": 5480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.675022222222222e-05,
+      "loss": 2.3401,
+      "step": 5485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.674725925925926e-05,
+      "loss": 2.3485,
+      "step": 5490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6744296296296294e-05,
+      "loss": 2.2853,
+      "step": 5495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.674133333333334e-05,
+      "loss": 2.3732,
+      "step": 5500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.673837037037037e-05,
+      "loss": 2.4763,
+      "step": 5505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.673540740740741e-05,
+      "loss": 2.3342,
+      "step": 5510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.673244444444444e-05,
+      "loss": 2.3536,
+      "step": 5515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.672948148148149e-05,
+      "loss": 2.4202,
+      "step": 5520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.672651851851852e-05,
+      "loss": 2.4043,
+      "step": 5525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.672355555555556e-05,
+      "loss": 2.4102,
+      "step": 5530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.672059259259259e-05,
+      "loss": 2.3792,
+      "step": 5535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6717629629629636e-05,
+      "loss": 2.3936,
+      "step": 5540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.671466666666667e-05,
+      "loss": 2.5566,
+      "step": 5545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.671170370370371e-05,
+      "loss": 2.4846,
+      "step": 5550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.670874074074074e-05,
+      "loss": 2.4058,
+      "step": 5555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6705777777777785e-05,
+      "loss": 2.5478,
+      "step": 5560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.670281481481482e-05,
+      "loss": 2.3242,
+      "step": 5565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6699851851851856e-05,
+      "loss": 2.4553,
+      "step": 5570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.669688888888889e-05,
+      "loss": 2.4632,
+      "step": 5575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.669392592592593e-05,
+      "loss": 2.4154,
+      "step": 5580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6690962962962965e-05,
+      "loss": 2.4006,
+      "step": 5585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6688000000000004e-05,
+      "loss": 2.4437,
+      "step": 5590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6685037037037036e-05,
+      "loss": 2.4406,
+      "step": 5595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6682074074074075e-05,
+      "loss": 2.315,
+      "step": 5600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6679111111111114e-05,
+      "loss": 2.4427,
+      "step": 5605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.667614814814815e-05,
+      "loss": 2.3768,
+      "step": 5610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6673185185185185e-05,
+      "loss": 2.4391,
+      "step": 5615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6670222222222224e-05,
+      "loss": 2.5285,
+      "step": 5620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.666725925925926e-05,
+      "loss": 2.4759,
+      "step": 5625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.66642962962963e-05,
+      "loss": 2.4138,
+      "step": 5630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.666133333333333e-05,
+      "loss": 2.3682,
+      "step": 5635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.665837037037037e-05,
+      "loss": 2.2183,
+      "step": 5640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.665540740740741e-05,
+      "loss": 2.2696,
+      "step": 5645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.665244444444445e-05,
+      "loss": 2.2216,
+      "step": 5650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.664948148148148e-05,
+      "loss": 2.2421,
+      "step": 5655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.664651851851852e-05,
+      "loss": 2.2556,
+      "step": 5660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.664355555555556e-05,
+      "loss": 2.4091,
+      "step": 5665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.66405925925926e-05,
+      "loss": 2.302,
+      "step": 5670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.663762962962963e-05,
+      "loss": 2.2694,
+      "step": 5675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.663466666666667e-05,
+      "loss": 2.1418,
+      "step": 5680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.663170370370371e-05,
+      "loss": 2.3009,
+      "step": 5685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.662874074074075e-05,
+      "loss": 2.2199,
+      "step": 5690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.662577777777778e-05,
+      "loss": 2.1113,
+      "step": 5695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.662281481481482e-05,
+      "loss": 2.1406,
+      "step": 5700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6619851851851857e-05,
+      "loss": 2.3233,
+      "step": 5705
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6616888888888895e-05,
+      "loss": 2.2806,
+      "step": 5710
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.661392592592593e-05,
+      "loss": 2.241,
+      "step": 5715
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6610962962962966e-05,
+      "loss": 2.2779,
+      "step": 5720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6608e-05,
+      "loss": 2.3018,
+      "step": 5725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6605037037037044e-05,
+      "loss": 2.2499,
+      "step": 5730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6602074074074076e-05,
+      "loss": 2.3864,
+      "step": 5735
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6599111111111115e-05,
+      "loss": 2.321,
+      "step": 5740
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.659614814814815e-05,
+      "loss": 2.0841,
+      "step": 5745
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.659318518518519e-05,
+      "loss": 2.3812,
+      "step": 5750
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6590222222222224e-05,
+      "loss": 2.3174,
+      "step": 5755
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.658725925925926e-05,
+      "loss": 2.2251,
+      "step": 5760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6584296296296295e-05,
+      "loss": 2.2463,
+      "step": 5765
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.658133333333334e-05,
+      "loss": 2.1952,
+      "step": 5770
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.657837037037037e-05,
+      "loss": 2.3955,
+      "step": 5775
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.657540740740741e-05,
+      "loss": 2.1974,
+      "step": 5780
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6572444444444444e-05,
+      "loss": 2.2574,
+      "step": 5785
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.656948148148149e-05,
+      "loss": 2.3413,
+      "step": 5790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.656651851851852e-05,
+      "loss": 2.2958,
+      "step": 5795
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.656355555555556e-05,
+      "loss": 2.2036,
+      "step": 5800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.656059259259259e-05,
+      "loss": 2.3152,
+      "step": 5805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.655762962962963e-05,
+      "loss": 2.1738,
+      "step": 5810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.655466666666667e-05,
+      "loss": 2.2618,
+      "step": 5815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.655170370370371e-05,
+      "loss": 2.2715,
+      "step": 5820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654874074074074e-05,
+      "loss": 2.4127,
+      "step": 5825
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654577777777778e-05,
+      "loss": 2.3517,
+      "step": 5830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.654281481481482e-05,
+      "loss": 2.2022,
+      "step": 5835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.653985185185186e-05,
+      "loss": 2.4314,
+      "step": 5840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.653688888888889e-05,
+      "loss": 2.1655,
+      "step": 5845
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.653392592592593e-05,
+      "loss": 2.3021,
+      "step": 5850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.653096296296297e-05,
+      "loss": 2.4702,
+      "step": 5855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6528000000000006e-05,
+      "loss": 2.1836,
+      "step": 5860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.652503703703704e-05,
+      "loss": 2.2377,
+      "step": 5865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.652207407407408e-05,
+      "loss": 2.3459,
+      "step": 5870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6519111111111116e-05,
+      "loss": 2.2778,
+      "step": 5875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6516148148148154e-05,
+      "loss": 2.2239,
+      "step": 5880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6513185185185186e-05,
+      "loss": 2.3235,
+      "step": 5885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.651022222222222e-05,
+      "loss": 2.2257,
+      "step": 5890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6507259259259264e-05,
+      "loss": 2.1825,
+      "step": 5895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6504296296296296e-05,
+      "loss": 2.2704,
+      "step": 5900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6501333333333335e-05,
+      "loss": 2.2846,
+      "step": 5905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649837037037037e-05,
+      "loss": 2.353,
+      "step": 5910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.649540740740741e-05,
+      "loss": 2.3631,
+      "step": 5915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6492444444444445e-05,
+      "loss": 2.3174,
+      "step": 5920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6489481481481483e-05,
+      "loss": 2.1961,
+      "step": 5925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.6486518518518515e-05,
+      "loss": 2.2129,
+      "step": 5930
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.648355555555556e-05,
+      "loss": 2.1397,
+      "step": 5935
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.648059259259259e-05,
+      "loss": 2.2563,
+      "step": 5940
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.647762962962963e-05,
+      "loss": 2.2357,
+      "step": 5945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.6474666666666664e-05,
+      "loss": 2.3086,
+      "step": 5950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.647170370370371e-05,
+      "loss": 2.3182,
+      "step": 5955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.646874074074074e-05,
+      "loss": 2.2669,
+      "step": 5960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.646577777777778e-05,
+      "loss": 2.3287,
+      "step": 5965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.646281481481481e-05,
+      "loss": 2.3493,
+      "step": 5970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.645985185185185e-05,
+      "loss": 2.3483,
+      "step": 5975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.645688888888889e-05,
+      "loss": 2.1827,
+      "step": 5980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.645392592592593e-05,
+      "loss": 2.3309,
+      "step": 5985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.645096296296296e-05,
+      "loss": 2.2168,
+      "step": 5990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6448e-05,
+      "loss": 2.4169,
+      "step": 5995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.644503703703704e-05,
+      "loss": 2.2138,
+      "step": 6000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.644207407407408e-05,
+      "loss": 2.2737,
+      "step": 6005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.643911111111111e-05,
+      "loss": 2.3454,
+      "step": 6010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.643614814814815e-05,
+      "loss": 2.2545,
+      "step": 6015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.643318518518519e-05,
+      "loss": 2.2839,
+      "step": 6020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6430222222222226e-05,
+      "loss": 2.1243,
+      "step": 6025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.642725925925926e-05,
+      "loss": 2.3119,
+      "step": 6030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.64242962962963e-05,
+      "loss": 2.3298,
+      "step": 6035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6421333333333336e-05,
+      "loss": 2.1839,
+      "step": 6040
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.6418370370370374e-05,
+      "loss": 2.3518,
+      "step": 6045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6416000000000004e-05,
+      "loss": 2.2573,
+      "step": 6050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6413037037037036e-05,
+      "loss": 2.322,
+      "step": 6055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6410074074074075e-05,
+      "loss": 2.3098,
+      "step": 6060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6407111111111114e-05,
+      "loss": 2.2697,
+      "step": 6065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.640414814814815e-05,
+      "loss": 2.2009,
+      "step": 6070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6401185185185185e-05,
+      "loss": 2.2627,
+      "step": 6075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.6398222222222224e-05,
+      "loss": 2.4355,
+      "step": 6080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.639525925925926e-05,
+      "loss": 2.3458,
+      "step": 6085
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.63922962962963e-05,
+      "loss": 2.3021,
+      "step": 6090
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.638933333333333e-05,
+      "loss": 2.3315,
+      "step": 6095
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.638637037037037e-05,
+      "loss": 2.3391,
+      "step": 6100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.638340740740741e-05,
+      "loss": 2.2762,
+      "step": 6105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.638044444444445e-05,
+      "loss": 2.3654,
+      "step": 6110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.637748148148148e-05,
+      "loss": 2.2251,
+      "step": 6115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.637451851851852e-05,
+      "loss": 2.3115,
+      "step": 6120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.637155555555556e-05,
+      "loss": 2.1566,
+      "step": 6125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.63685925925926e-05,
+      "loss": 2.1691,
+      "step": 6130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.636562962962963e-05,
+      "loss": 2.1287,
+      "step": 6135
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.636266666666667e-05,
+      "loss": 2.2591,
+      "step": 6140
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.635970370370371e-05,
+      "loss": 2.204,
+      "step": 6145
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.635674074074075e-05,
+      "loss": 2.208,
+      "step": 6150
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.635377777777778e-05,
+      "loss": 2.2976,
+      "step": 6155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.635081481481482e-05,
+      "loss": 2.2446,
+      "step": 6160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6347851851851856e-05,
+      "loss": 2.1389,
+      "step": 6165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6344888888888895e-05,
+      "loss": 2.2893,
+      "step": 6170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.634192592592593e-05,
+      "loss": 2.2795,
+      "step": 6175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6338962962962966e-05,
+      "loss": 2.2333,
+      "step": 6180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6336000000000005e-05,
+      "loss": 2.1521,
+      "step": 6185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6333037037037044e-05,
+      "loss": 2.3204,
+      "step": 6190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6330074074074076e-05,
+      "loss": 2.2406,
+      "step": 6195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6327111111111115e-05,
+      "loss": 2.3181,
+      "step": 6200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.632414814814815e-05,
+      "loss": 2.3463,
+      "step": 6205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.632118518518519e-05,
+      "loss": 2.4011,
+      "step": 6210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6318222222222224e-05,
+      "loss": 2.2294,
+      "step": 6215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.631525925925926e-05,
+      "loss": 2.2613,
+      "step": 6220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.6312296296296295e-05,
+      "loss": 2.3396,
+      "step": 6225
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.630933333333334e-05,
+      "loss": 2.2936,
+      "step": 6230
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.630637037037037e-05,
+      "loss": 2.2127,
+      "step": 6235
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.630340740740741e-05,
+      "loss": 2.2135,
+      "step": 6240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.6300444444444444e-05,
+      "loss": 2.2102,
+      "step": 6245
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.629748148148149e-05,
+      "loss": 2.2245,
+      "step": 6250
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.629451851851852e-05,
+      "loss": 2.2822,
+      "step": 6255
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.629155555555556e-05,
+      "loss": 2.2077,
+      "step": 6260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.628859259259259e-05,
+      "loss": 2.3078,
+      "step": 6265
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.628562962962963e-05,
+      "loss": 2.1341,
+      "step": 6270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.628266666666667e-05,
+      "loss": 2.2508,
+      "step": 6275
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.627970370370371e-05,
+      "loss": 2.2801,
+      "step": 6280
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.627674074074074e-05,
+      "loss": 2.2972,
+      "step": 6285
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.627377777777778e-05,
+      "loss": 2.3508,
+      "step": 6290
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.627081481481482e-05,
+      "loss": 2.2515,
+      "step": 6295
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.626785185185186e-05,
+      "loss": 2.2735,
+      "step": 6300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.626488888888889e-05,
+      "loss": 2.287,
+      "step": 6305
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.626192592592593e-05,
+      "loss": 2.1915,
+      "step": 6310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.625896296296297e-05,
+      "loss": 2.3413,
+      "step": 6315
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.6256000000000006e-05,
+      "loss": 2.1968,
+      "step": 6320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.625303703703704e-05,
+      "loss": 2.2362,
+      "step": 6325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6250074074074077e-05,
+      "loss": 2.214,
+      "step": 6330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6247111111111115e-05,
+      "loss": 2.1207,
+      "step": 6335
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6244148148148154e-05,
+      "loss": 2.25,
+      "step": 6340
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6241185185185186e-05,
+      "loss": 2.3796,
+      "step": 6345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.623822222222222e-05,
+      "loss": 2.3142,
+      "step": 6350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6235259259259264e-05,
+      "loss": 2.2502,
+      "step": 6355
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6232296296296296e-05,
+      "loss": 2.2969,
+      "step": 6360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6229333333333335e-05,
+      "loss": 2.1859,
+      "step": 6365
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.622637037037037e-05,
+      "loss": 2.3306,
+      "step": 6370
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.622340740740741e-05,
+      "loss": 2.2796,
+      "step": 6375
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.6220444444444444e-05,
+      "loss": 2.2645,
+      "step": 6380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.621748148148148e-05,
+      "loss": 2.2048,
+      "step": 6385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.6214518518518515e-05,
+      "loss": 2.3097,
+      "step": 6390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.621155555555556e-05,
+      "loss": 2.2794,
+      "step": 6395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.620859259259259e-05,
+      "loss": 2.2973,
+      "step": 6400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.620562962962963e-05,
+      "loss": 2.2797,
+      "step": 6405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.6202666666666664e-05,
+      "loss": 2.2649,
+      "step": 6410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.619970370370371e-05,
+      "loss": 2.27,
+      "step": 6415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.619674074074074e-05,
+      "loss": 2.1985,
+      "step": 6420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.619377777777778e-05,
+      "loss": 2.3526,
+      "step": 6425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.619081481481481e-05,
+      "loss": 2.2487,
+      "step": 6430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.618785185185185e-05,
+      "loss": 2.251,
+      "step": 6435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.618488888888889e-05,
+      "loss": 2.2163,
+      "step": 6440
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.618192592592593e-05,
+      "loss": 2.3746,
+      "step": 6445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.617896296296296e-05,
+      "loss": 2.2269,
+      "step": 6450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.6176e-05,
+      "loss": 2.2837,
+      "step": 6455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.617303703703704e-05,
+      "loss": 2.2553,
+      "step": 6460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.617007407407408e-05,
+      "loss": 2.2647,
+      "step": 6465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.616711111111111e-05,
+      "loss": 2.2199,
+      "step": 6470
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.616414814814815e-05,
+      "loss": 2.2478,
+      "step": 6475
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.616118518518519e-05,
+      "loss": 2.183,
+      "step": 6480
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.6158222222222226e-05,
+      "loss": 2.2811,
+      "step": 6485
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.615525925925926e-05,
+      "loss": 2.2932,
+      "step": 6490
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.61522962962963e-05,
+      "loss": 2.2424,
+      "step": 6495
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6149333333333336e-05,
+      "loss": 2.3681,
+      "step": 6500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6146370370370374e-05,
+      "loss": 2.287,
+      "step": 6505
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6143407407407406e-05,
+      "loss": 2.17,
+      "step": 6510
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6140444444444445e-05,
+      "loss": 2.3137,
+      "step": 6515
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6137481481481484e-05,
+      "loss": 2.282,
+      "step": 6520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.613451851851852e-05,
+      "loss": 2.3593,
+      "step": 6525
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6131555555555555e-05,
+      "loss": 2.2093,
+      "step": 6530
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6128592592592594e-05,
+      "loss": 2.4281,
+      "step": 6535
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.612562962962963e-05,
+      "loss": 2.2629,
+      "step": 6540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.612266666666667e-05,
+      "loss": 2.2337,
+      "step": 6545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.6119703703703703e-05,
+      "loss": 2.3051,
+      "step": 6550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.611674074074074e-05,
+      "loss": 2.2408,
+      "step": 6555
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.611377777777778e-05,
+      "loss": 2.3408,
+      "step": 6560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.611081481481482e-05,
+      "loss": 2.0711,
+      "step": 6565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.610785185185185e-05,
+      "loss": 2.3387,
+      "step": 6570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.610488888888889e-05,
+      "loss": 2.2371,
+      "step": 6575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.610192592592592e-05,
+      "loss": 2.3484,
+      "step": 6580
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.609896296296297e-05,
+      "loss": 2.345,
+      "step": 6585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.6096e-05,
+      "loss": 2.1856,
+      "step": 6590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.609303703703704e-05,
+      "loss": 2.2727,
+      "step": 6595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.609007407407407e-05,
+      "loss": 2.2317,
+      "step": 6600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.608711111111112e-05,
+      "loss": 2.229,
+      "step": 6605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.608414814814815e-05,
+      "loss": 2.3958,
+      "step": 6610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.608118518518519e-05,
+      "loss": 2.247,
+      "step": 6615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.607822222222222e-05,
+      "loss": 2.2758,
+      "step": 6620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6075259259259265e-05,
+      "loss": 2.2069,
+      "step": 6625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.60722962962963e-05,
+      "loss": 2.3289,
+      "step": 6630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6069333333333336e-05,
+      "loss": 2.3894,
+      "step": 6635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6066962962962966e-05,
+      "loss": 2.1465,
+      "step": 6640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6064000000000005e-05,
+      "loss": 2.0789,
+      "step": 6645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6061037037037044e-05,
+      "loss": 2.3641,
+      "step": 6650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6058074074074076e-05,
+      "loss": 2.2511,
+      "step": 6655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6055111111111115e-05,
+      "loss": 2.2238,
+      "step": 6660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.6052148148148147e-05,
+      "loss": 2.1556,
+      "step": 6665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.604918518518519e-05,
+      "loss": 2.2347,
+      "step": 6670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.6046222222222224e-05,
+      "loss": 2.2219,
+      "step": 6675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.604325925925926e-05,
+      "loss": 2.291,
+      "step": 6680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.6040296296296295e-05,
+      "loss": 2.3262,
+      "step": 6685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.603733333333334e-05,
+      "loss": 2.4022,
+      "step": 6690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.603437037037037e-05,
+      "loss": 2.2977,
+      "step": 6695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.603140740740741e-05,
+      "loss": 2.1813,
+      "step": 6700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.6028444444444444e-05,
+      "loss": 2.3603,
+      "step": 6705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.602548148148149e-05,
+      "loss": 2.3723,
+      "step": 6710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.602251851851852e-05,
+      "loss": 2.4143,
+      "step": 6715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.601955555555556e-05,
+      "loss": 2.3454,
+      "step": 6720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.601659259259259e-05,
+      "loss": 2.248,
+      "step": 6725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.601362962962963e-05,
+      "loss": 2.3829,
+      "step": 6730
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.601066666666667e-05,
+      "loss": 2.1406,
+      "step": 6735
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.600770370370371e-05,
+      "loss": 2.2528,
+      "step": 6740
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.600474074074074e-05,
+      "loss": 2.2589,
+      "step": 6745
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.600177777777778e-05,
+      "loss": 2.2276,
+      "step": 6750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.599881481481482e-05,
+      "loss": 2.2346,
+      "step": 6755
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.599585185185186e-05,
+      "loss": 2.23,
+      "step": 6760
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.599288888888889e-05,
+      "loss": 2.3288,
+      "step": 6765
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.598992592592593e-05,
+      "loss": 2.3982,
+      "step": 6770
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.598696296296297e-05,
+      "loss": 2.2338,
+      "step": 6775
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5984000000000006e-05,
+      "loss": 2.2898,
+      "step": 6780
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.598103703703704e-05,
+      "loss": 2.2546,
+      "step": 6785
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5978074074074076e-05,
+      "loss": 2.3104,
+      "step": 6790
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5975111111111115e-05,
+      "loss": 2.2883,
+      "step": 6795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5972148148148154e-05,
+      "loss": 2.1726,
+      "step": 6800
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5969185185185186e-05,
+      "loss": 2.2614,
+      "step": 6805
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5966222222222225e-05,
+      "loss": 2.3599,
+      "step": 6810
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5963259259259264e-05,
+      "loss": 2.3086,
+      "step": 6815
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5960296296296296e-05,
+      "loss": 2.2824,
+      "step": 6820
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5957333333333335e-05,
+      "loss": 2.3178,
+      "step": 6825
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.595437037037037e-05,
+      "loss": 2.2981,
+      "step": 6830
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.595140740740741e-05,
+      "loss": 2.3144,
+      "step": 6835
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.5948444444444444e-05,
+      "loss": 2.3403,
+      "step": 6840
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.594548148148148e-05,
+      "loss": 2.2833,
+      "step": 6845
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.5942518518518515e-05,
+      "loss": 2.3044,
+      "step": 6850
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.593955555555556e-05,
+      "loss": 2.2435,
+      "step": 6855
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.593659259259259e-05,
+      "loss": 2.1663,
+      "step": 6860
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.593362962962963e-05,
+      "loss": 2.2313,
+      "step": 6865
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.5930666666666664e-05,
+      "loss": 2.1784,
+      "step": 6870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.592770370370371e-05,
+      "loss": 2.1213,
+      "step": 6875
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.592474074074074e-05,
+      "loss": 2.404,
+      "step": 6880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.592177777777778e-05,
+      "loss": 2.2114,
+      "step": 6885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.591881481481481e-05,
+      "loss": 2.2593,
+      "step": 6890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.591585185185185e-05,
+      "loss": 2.3533,
+      "step": 6895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.591288888888889e-05,
+      "loss": 2.3332,
+      "step": 6900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.590992592592593e-05,
+      "loss": 2.2287,
+      "step": 6905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.590696296296296e-05,
+      "loss": 2.1564,
+      "step": 6910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.5904e-05,
+      "loss": 2.3331,
+      "step": 6915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.590103703703704e-05,
+      "loss": 2.3208,
+      "step": 6920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.589807407407408e-05,
+      "loss": 2.2578,
+      "step": 6925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.589511111111111e-05,
+      "loss": 2.1348,
+      "step": 6930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.589214814814815e-05,
+      "loss": 2.3644,
+      "step": 6935
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.588918518518519e-05,
+      "loss": 2.2255,
+      "step": 6940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.5886222222222226e-05,
+      "loss": 2.283,
+      "step": 6945
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.588325925925926e-05,
+      "loss": 2.2609,
+      "step": 6950
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.58802962962963e-05,
+      "loss": 2.2377,
+      "step": 6955
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5877333333333335e-05,
+      "loss": 2.2686,
+      "step": 6960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5874370370370374e-05,
+      "loss": 2.3053,
+      "step": 6965
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5871407407407406e-05,
+      "loss": 2.2412,
+      "step": 6970
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5868444444444445e-05,
+      "loss": 2.2785,
+      "step": 6975
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5865481481481484e-05,
+      "loss": 2.2444,
+      "step": 6980
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.586251851851852e-05,
+      "loss": 2.1793,
+      "step": 6985
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5859555555555555e-05,
+      "loss": 2.2404,
+      "step": 6990
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.5856592592592594e-05,
+      "loss": 2.198,
+      "step": 6995
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.585362962962963e-05,
+      "loss": 2.1602,
+      "step": 7000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.585066666666667e-05,
+      "loss": 2.2226,
+      "step": 7005
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.58477037037037e-05,
+      "loss": 2.3,
+      "step": 7010
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.584474074074074e-05,
+      "loss": 2.1955,
+      "step": 7015
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.584177777777778e-05,
+      "loss": 2.2604,
+      "step": 7020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.583881481481482e-05,
+      "loss": 2.3292,
+      "step": 7025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.583585185185185e-05,
+      "loss": 2.3145,
+      "step": 7030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.583288888888889e-05,
+      "loss": 2.2457,
+      "step": 7035
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.582992592592593e-05,
+      "loss": 2.3012,
+      "step": 7040
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.582696296296297e-05,
+      "loss": 2.2856,
+      "step": 7045
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.5824e-05,
+      "loss": 2.27,
+      "step": 7050
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.582103703703704e-05,
+      "loss": 2.2098,
+      "step": 7055
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.581807407407407e-05,
+      "loss": 2.3036,
+      "step": 7060
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.581511111111112e-05,
+      "loss": 2.2739,
+      "step": 7065
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.581214814814815e-05,
+      "loss": 2.2305,
+      "step": 7070
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.580918518518519e-05,
+      "loss": 2.1138,
+      "step": 7075
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.580622222222222e-05,
+      "loss": 2.3126,
+      "step": 7080
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5803259259259265e-05,
+      "loss": 2.3705,
+      "step": 7085
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.58002962962963e-05,
+      "loss": 2.3436,
+      "step": 7090
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5797333333333336e-05,
+      "loss": 2.2504,
+      "step": 7095
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.579437037037037e-05,
+      "loss": 2.2131,
+      "step": 7100
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5791407407407414e-05,
+      "loss": 2.2504,
+      "step": 7105
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5788444444444446e-05,
+      "loss": 2.2675,
+      "step": 7110
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5785481481481485e-05,
+      "loss": 2.3812,
+      "step": 7115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.578251851851852e-05,
+      "loss": 2.243,
+      "step": 7120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.5779555555555556e-05,
+      "loss": 2.2187,
+      "step": 7125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.5776592592592594e-05,
+      "loss": 2.2334,
+      "step": 7130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.577362962962963e-05,
+      "loss": 2.2459,
+      "step": 7135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.5770666666666665e-05,
+      "loss": 2.3018,
+      "step": 7140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.5767703703703704e-05,
+      "loss": 2.3779,
+      "step": 7145
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.576474074074074e-05,
+      "loss": 2.2553,
+      "step": 7150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.576177777777778e-05,
+      "loss": 2.2515,
+      "step": 7155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.5758814814814814e-05,
+      "loss": 2.1698,
+      "step": 7160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.575585185185185e-05,
+      "loss": 2.2536,
+      "step": 7165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.575288888888889e-05,
+      "loss": 2.3004,
+      "step": 7170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.574992592592593e-05,
+      "loss": 2.2575,
+      "step": 7175
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.574696296296296e-05,
+      "loss": 2.3674,
+      "step": 7180
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.5744e-05,
+      "loss": 2.2229,
+      "step": 7185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.574103703703704e-05,
+      "loss": 2.1819,
+      "step": 7190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.573807407407408e-05,
+      "loss": 2.3481,
+      "step": 7195
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.573511111111111e-05,
+      "loss": 2.183,
+      "step": 7200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.573214814814815e-05,
+      "loss": 2.2309,
+      "step": 7205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.572918518518519e-05,
+      "loss": 2.181,
+      "step": 7210
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.572622222222223e-05,
+      "loss": 2.1743,
+      "step": 7215
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.572325925925926e-05,
+      "loss": 2.2209,
+      "step": 7220
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.57202962962963e-05,
+      "loss": 2.2276,
+      "step": 7225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.571733333333334e-05,
+      "loss": 2.2843,
+      "step": 7230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5714370370370376e-05,
+      "loss": 2.2599,
+      "step": 7235
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.571140740740741e-05,
+      "loss": 2.2667,
+      "step": 7240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.570844444444445e-05,
+      "loss": 2.2352,
+      "step": 7245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5705481481481485e-05,
+      "loss": 2.2337,
+      "step": 7250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5702518518518524e-05,
+      "loss": 2.2564,
+      "step": 7255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5699555555555556e-05,
+      "loss": 2.2869,
+      "step": 7260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5696592592592595e-05,
+      "loss": 2.3509,
+      "step": 7265
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5693629629629634e-05,
+      "loss": 2.3603,
+      "step": 7270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.569066666666667e-05,
+      "loss": 2.2507,
+      "step": 7275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5687703703703705e-05,
+      "loss": 2.3355,
+      "step": 7280
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.5684740740740744e-05,
+      "loss": 2.2302,
+      "step": 7285
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.5681777777777776e-05,
+      "loss": 2.1399,
+      "step": 7290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.567881481481482e-05,
+      "loss": 2.1978,
+      "step": 7295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.567585185185185e-05,
+      "loss": 2.3155,
+      "step": 7300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.567288888888889e-05,
+      "loss": 2.2613,
+      "step": 7305
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.5669925925925924e-05,
+      "loss": 2.2463,
+      "step": 7310
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.566696296296297e-05,
+      "loss": 2.3753,
+      "step": 7315
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.5664e-05,
+      "loss": 2.3009,
+      "step": 7320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.566103703703704e-05,
+      "loss": 2.2551,
+      "step": 7325
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.565807407407407e-05,
+      "loss": 2.1589,
+      "step": 7330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.565511111111112e-05,
+      "loss": 2.2831,
+      "step": 7335
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.565214814814815e-05,
+      "loss": 2.2871,
+      "step": 7340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.564918518518519e-05,
+      "loss": 2.2423,
+      "step": 7345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.564622222222222e-05,
+      "loss": 2.2549,
+      "step": 7350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.564325925925926e-05,
+      "loss": 2.3964,
+      "step": 7355
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.56402962962963e-05,
+      "loss": 2.2096,
+      "step": 7360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.563733333333334e-05,
+      "loss": 2.3593,
+      "step": 7365
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.563437037037037e-05,
+      "loss": 2.186,
+      "step": 7370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.563140740740741e-05,
+      "loss": 2.2026,
+      "step": 7375
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.562844444444445e-05,
+      "loss": 2.3176,
+      "step": 7380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.5625481481481486e-05,
+      "loss": 2.2689,
+      "step": 7385
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.562251851851852e-05,
+      "loss": 2.2994,
+      "step": 7390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.561955555555556e-05,
+      "loss": 2.32,
+      "step": 7395
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5616592592592596e-05,
+      "loss": 2.2417,
+      "step": 7400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5613629629629635e-05,
+      "loss": 2.1608,
+      "step": 7405
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.561066666666667e-05,
+      "loss": 2.2433,
+      "step": 7410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5607703703703706e-05,
+      "loss": 2.2954,
+      "step": 7415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5604740740740744e-05,
+      "loss": 2.2712,
+      "step": 7420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.560177777777778e-05,
+      "loss": 2.2872,
+      "step": 7425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5598814814814815e-05,
+      "loss": 2.3773,
+      "step": 7430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5595851851851854e-05,
+      "loss": 2.2847,
+      "step": 7435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.559288888888889e-05,
+      "loss": 2.1243,
+      "step": 7440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.558992592592593e-05,
+      "loss": 2.2983,
+      "step": 7445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.5586962962962964e-05,
+      "loss": 2.2257,
+      "step": 7450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.5584e-05,
+      "loss": 2.3544,
+      "step": 7455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.558103703703704e-05,
+      "loss": 2.1814,
+      "step": 7460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.557807407407408e-05,
+      "loss": 2.2129,
+      "step": 7465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.557511111111111e-05,
+      "loss": 2.2303,
+      "step": 7470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.557214814814815e-05,
+      "loss": 2.3359,
+      "step": 7475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.556918518518519e-05,
+      "loss": 2.2404,
+      "step": 7480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.556622222222223e-05,
+      "loss": 2.3884,
+      "step": 7485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.556325925925926e-05,
+      "loss": 2.2906,
+      "step": 7490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.55602962962963e-05,
+      "loss": 2.2593,
+      "step": 7495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.555733333333334e-05,
+      "loss": 2.2034,
+      "step": 7500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.555437037037038e-05,
+      "loss": 2.383,
+      "step": 7505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.555140740740741e-05,
+      "loss": 2.2004,
+      "step": 7510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.554844444444445e-05,
+      "loss": 2.3607,
+      "step": 7515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.554548148148148e-05,
+      "loss": 2.1642,
+      "step": 7520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.5542518518518526e-05,
+      "loss": 2.1457,
+      "step": 7525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.553955555555556e-05,
+      "loss": 2.2545,
+      "step": 7530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.55365925925926e-05,
+      "loss": 2.2655,
+      "step": 7535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.553362962962963e-05,
+      "loss": 2.1745,
+      "step": 7540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.5530666666666674e-05,
+      "loss": 2.3619,
+      "step": 7545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.5527703703703706e-05,
+      "loss": 2.3244,
+      "step": 7550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.5524740740740745e-05,
+      "loss": 2.421,
+      "step": 7555
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.552177777777778e-05,
+      "loss": 2.1832,
+      "step": 7560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.551881481481482e-05,
+      "loss": 2.2433,
+      "step": 7565
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5515851851851855e-05,
+      "loss": 2.3146,
+      "step": 7570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5512888888888894e-05,
+      "loss": 2.2164,
+      "step": 7575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5509925925925926e-05,
+      "loss": 2.13,
+      "step": 7580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5506962962962965e-05,
+      "loss": 2.2406,
+      "step": 7585
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5504000000000003e-05,
+      "loss": 2.1664,
+      "step": 7590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5501037037037035e-05,
+      "loss": 2.2063,
+      "step": 7595
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5498074074074074e-05,
+      "loss": 2.1979,
+      "step": 7600
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.549511111111111e-05,
+      "loss": 2.2476,
+      "step": 7605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.549214814814815e-05,
+      "loss": 2.3224,
+      "step": 7610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.5489185185185184e-05,
+      "loss": 2.1408,
+      "step": 7615
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.548622222222222e-05,
+      "loss": 2.2592,
+      "step": 7620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.548325925925926e-05,
+      "loss": 2.1739,
+      "step": 7625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.54802962962963e-05,
+      "loss": 2.2779,
+      "step": 7630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.547733333333333e-05,
+      "loss": 2.1724,
+      "step": 7635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.547437037037037e-05,
+      "loss": 2.1572,
+      "step": 7640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.547140740740741e-05,
+      "loss": 2.2909,
+      "step": 7645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.546844444444445e-05,
+      "loss": 2.3665,
+      "step": 7650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.546548148148148e-05,
+      "loss": 2.2536,
+      "step": 7655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.546251851851852e-05,
+      "loss": 2.2624,
+      "step": 7660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.545955555555555e-05,
+      "loss": 2.2469,
+      "step": 7665
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.54565925925926e-05,
+      "loss": 2.2626,
+      "step": 7670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.545362962962963e-05,
+      "loss": 2.2693,
+      "step": 7675
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.545066666666667e-05,
+      "loss": 2.0831,
+      "step": 7680
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.54477037037037e-05,
+      "loss": 2.3036,
+      "step": 7685
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5444740740740746e-05,
+      "loss": 2.2692,
+      "step": 7690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.544177777777778e-05,
+      "loss": 2.212,
+      "step": 7695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.543881481481482e-05,
+      "loss": 2.2737,
+      "step": 7700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.543585185185185e-05,
+      "loss": 2.294,
+      "step": 7705
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5432888888888895e-05,
+      "loss": 2.2214,
+      "step": 7710
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5429925925925927e-05,
+      "loss": 2.2875,
+      "step": 7715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5426962962962965e-05,
+      "loss": 2.2726,
+      "step": 7720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.5424e-05,
+      "loss": 2.1564,
+      "step": 7725
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.542103703703704e-05,
+      "loss": 2.3015,
+      "step": 7730
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5418074074074075e-05,
+      "loss": 2.2133,
+      "step": 7735
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5415111111111114e-05,
+      "loss": 2.1914,
+      "step": 7740
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5412148148148146e-05,
+      "loss": 2.1841,
+      "step": 7745
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5409185185185185e-05,
+      "loss": 2.2633,
+      "step": 7750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5406222222222224e-05,
+      "loss": 2.2805,
+      "step": 7755
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.540325925925926e-05,
+      "loss": 2.1288,
+      "step": 7760
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.5400296296296294e-05,
+      "loss": 2.1955,
+      "step": 7765
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.539733333333333e-05,
+      "loss": 2.2045,
+      "step": 7770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.539437037037037e-05,
+      "loss": 2.1914,
+      "step": 7775
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.539140740740741e-05,
+      "loss": 2.2926,
+      "step": 7780
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.538844444444444e-05,
+      "loss": 2.2439,
+      "step": 7785
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.538548148148148e-05,
+      "loss": 2.338,
+      "step": 7790
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.538251851851852e-05,
+      "loss": 2.3185,
+      "step": 7795
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537955555555556e-05,
+      "loss": 2.2419,
+      "step": 7800
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537659259259259e-05,
+      "loss": 2.1946,
+      "step": 7805
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537362962962963e-05,
+      "loss": 2.1165,
+      "step": 7810
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.537066666666667e-05,
+      "loss": 2.352,
+      "step": 7815
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536770370370371e-05,
+      "loss": 2.1178,
+      "step": 7820
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536474074074074e-05,
+      "loss": 2.2348,
+      "step": 7825
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.536177777777778e-05,
+      "loss": 2.2033,
+      "step": 7830
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.535881481481482e-05,
+      "loss": 2.3011,
+      "step": 7835
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.5355851851851856e-05,
+      "loss": 2.2968,
+      "step": 7840
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.535288888888889e-05,
+      "loss": 2.2689,
+      "step": 7845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.534992592592593e-05,
+      "loss": 2.24,
+      "step": 7850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5346962962962966e-05,
+      "loss": 2.2188,
+      "step": 7855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5344000000000005e-05,
+      "loss": 2.1723,
+      "step": 7860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.534103703703704e-05,
+      "loss": 2.2588,
+      "step": 7865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5338074074074076e-05,
+      "loss": 2.2011,
+      "step": 7870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5335111111111115e-05,
+      "loss": 2.343,
+      "step": 7875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5332148148148153e-05,
+      "loss": 2.2202,
+      "step": 7880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5329185185185186e-05,
+      "loss": 2.3572,
+      "step": 7885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5326222222222224e-05,
+      "loss": 2.1689,
+      "step": 7890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.5323259259259256e-05,
+      "loss": 2.248,
+      "step": 7895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.53202962962963e-05,
+      "loss": 2.2938,
+      "step": 7900
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.5317333333333334e-05,
+      "loss": 2.3777,
+      "step": 7905
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.531437037037037e-05,
+      "loss": 2.192,
+      "step": 7910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.5311407407407405e-05,
+      "loss": 2.0929,
+      "step": 7915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.530844444444445e-05,
+      "loss": 2.3745,
+      "step": 7920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.530548148148148e-05,
+      "loss": 2.2037,
+      "step": 7925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.530251851851852e-05,
+      "loss": 2.2358,
+      "step": 7930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.5299555555555553e-05,
+      "loss": 2.2324,
+      "step": 7935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.52965925925926e-05,
+      "loss": 2.3352,
+      "step": 7940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.529362962962963e-05,
+      "loss": 2.277,
+      "step": 7945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.529066666666667e-05,
+      "loss": 2.3493,
+      "step": 7950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.52877037037037e-05,
+      "loss": 2.1944,
+      "step": 7955
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.528474074074075e-05,
+      "loss": 2.3066,
+      "step": 7960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.528177777777778e-05,
+      "loss": 2.2026,
+      "step": 7965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.527881481481482e-05,
+      "loss": 2.1915,
+      "step": 7970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.527585185185185e-05,
+      "loss": 2.1089,
+      "step": 7975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.527288888888889e-05,
+      "loss": 2.0901,
+      "step": 7980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.526992592592593e-05,
+      "loss": 2.1283,
+      "step": 7985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.526696296296297e-05,
+      "loss": 2.328,
+      "step": 7990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.5264e-05,
+      "loss": 2.2328,
+      "step": 7995
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.526103703703704e-05,
+      "loss": 2.2545,
+      "step": 8000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.5258074074074077e-05,
+      "loss": 2.1648,
+      "step": 8005
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.5255111111111115e-05,
+      "loss": 2.2129,
+      "step": 8010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.525214814814815e-05,
+      "loss": 2.2575,
+      "step": 8015
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5249185185185186e-05,
+      "loss": 2.2243,
+      "step": 8020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5246222222222225e-05,
+      "loss": 2.2518,
+      "step": 8025
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5243259259259264e-05,
+      "loss": 2.3025,
+      "step": 8030
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5240296296296296e-05,
+      "loss": 2.2745,
+      "step": 8035
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5237333333333335e-05,
+      "loss": 2.1712,
+      "step": 8040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5234370370370374e-05,
+      "loss": 2.2009,
+      "step": 8045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.523140740740741e-05,
+      "loss": 2.2782,
+      "step": 8050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.5228444444444444e-05,
+      "loss": 2.17,
+      "step": 8055
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.522548148148148e-05,
+      "loss": 2.2484,
+      "step": 8060
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.522251851851852e-05,
+      "loss": 2.1069,
+      "step": 8065
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.521955555555556e-05,
+      "loss": 2.029,
+      "step": 8070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.521659259259259e-05,
+      "loss": 2.1971,
+      "step": 8075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.521362962962963e-05,
+      "loss": 2.2024,
+      "step": 8080
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.521066666666667e-05,
+      "loss": 2.2824,
+      "step": 8085
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.520770370370371e-05,
+      "loss": 2.2766,
+      "step": 8090
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.520474074074074e-05,
+      "loss": 2.2885,
+      "step": 8095
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.520177777777778e-05,
+      "loss": 2.2258,
+      "step": 8100
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.519881481481482e-05,
+      "loss": 2.1843,
+      "step": 8105
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.519585185185186e-05,
+      "loss": 2.262,
+      "step": 8110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.519288888888889e-05,
+      "loss": 2.3421,
+      "step": 8115
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.518992592592593e-05,
+      "loss": 2.2514,
+      "step": 8120
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.518696296296296e-05,
+      "loss": 2.2268,
+      "step": 8125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5184000000000006e-05,
+      "loss": 2.2646,
+      "step": 8130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.518103703703704e-05,
+      "loss": 2.3616,
+      "step": 8135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.517807407407408e-05,
+      "loss": 2.234,
+      "step": 8140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.517511111111111e-05,
+      "loss": 2.2045,
+      "step": 8145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5172148148148155e-05,
+      "loss": 2.2334,
+      "step": 8150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.516918518518519e-05,
+      "loss": 2.2639,
+      "step": 8155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5166222222222226e-05,
+      "loss": 2.2197,
+      "step": 8160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.516325925925926e-05,
+      "loss": 2.2616,
+      "step": 8165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5160296296296304e-05,
+      "loss": 2.3805,
+      "step": 8170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5157333333333336e-05,
+      "loss": 2.2324,
+      "step": 8175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.5154370370370374e-05,
+      "loss": 2.3746,
+      "step": 8180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5151407407407406e-05,
+      "loss": 2.2441,
+      "step": 8185
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.514844444444445e-05,
+      "loss": 2.0798,
+      "step": 8190
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5145481481481484e-05,
+      "loss": 2.2471,
+      "step": 8195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.514251851851852e-05,
+      "loss": 2.3012,
+      "step": 8200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5139555555555555e-05,
+      "loss": 2.2261,
+      "step": 8205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5136592592592594e-05,
+      "loss": 2.2868,
+      "step": 8210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.513362962962963e-05,
+      "loss": 2.1971,
+      "step": 8215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.513066666666667e-05,
+      "loss": 2.2844,
+      "step": 8220
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.5127703703703703e-05,
+      "loss": 2.3089,
+      "step": 8225
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.512474074074074e-05,
+      "loss": 2.265,
+      "step": 8230
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.512177777777778e-05,
+      "loss": 2.3346,
+      "step": 8235
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.511881481481482e-05,
+      "loss": 2.1878,
+      "step": 8240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.511585185185185e-05,
+      "loss": 2.2268,
+      "step": 8245
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.511288888888889e-05,
+      "loss": 2.1943,
+      "step": 8250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.510992592592593e-05,
+      "loss": 2.2257,
+      "step": 8255
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.510696296296297e-05,
+      "loss": 2.3799,
+      "step": 8260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.5104e-05,
+      "loss": 2.3179,
+      "step": 8265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.510103703703704e-05,
+      "loss": 2.2771,
+      "step": 8270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.509807407407408e-05,
+      "loss": 2.279,
+      "step": 8275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.509511111111112e-05,
+      "loss": 2.3063,
+      "step": 8280
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.509214814814815e-05,
+      "loss": 2.3726,
+      "step": 8285
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.508918518518519e-05,
+      "loss": 2.2917,
+      "step": 8290
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.508622222222223e-05,
+      "loss": 2.3271,
+      "step": 8295
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5083259259259265e-05,
+      "loss": 2.2034,
+      "step": 8300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.50802962962963e-05,
+      "loss": 2.2116,
+      "step": 8305
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5077333333333336e-05,
+      "loss": 2.2149,
+      "step": 8310
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5074370370370375e-05,
+      "loss": 2.2908,
+      "step": 8315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5071407407407414e-05,
+      "loss": 2.171,
+      "step": 8320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5068444444444446e-05,
+      "loss": 2.2796,
+      "step": 8325
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5065481481481485e-05,
+      "loss": 2.1624,
+      "step": 8330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5062518518518524e-05,
+      "loss": 2.3399,
+      "step": 8335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.505955555555556e-05,
+      "loss": 2.1823,
+      "step": 8340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5056592592592595e-05,
+      "loss": 2.2171,
+      "step": 8345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.5054222222222224e-05,
+      "loss": 2.3208,
+      "step": 8350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.505125925925926e-05,
+      "loss": 2.2498,
+      "step": 8355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.50482962962963e-05,
+      "loss": 2.3605,
+      "step": 8360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5045333333333334e-05,
+      "loss": 2.2842,
+      "step": 8365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.504237037037037e-05,
+      "loss": 2.0507,
+      "step": 8370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.5039407407407405e-05,
+      "loss": 2.4088,
+      "step": 8375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.503644444444445e-05,
+      "loss": 2.1591,
+      "step": 8380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.503348148148148e-05,
+      "loss": 2.4022,
+      "step": 8385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.503051851851852e-05,
+      "loss": 2.3104,
+      "step": 8390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.502755555555555e-05,
+      "loss": 2.1783,
+      "step": 8395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.50245925925926e-05,
+      "loss": 2.2255,
+      "step": 8400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.502162962962963e-05,
+      "loss": 2.3139,
+      "step": 8405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.501866666666667e-05,
+      "loss": 2.3446,
+      "step": 8410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.50157037037037e-05,
+      "loss": 2.2208,
+      "step": 8415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.501274074074075e-05,
+      "loss": 2.2788,
+      "step": 8420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.500977777777778e-05,
+      "loss": 2.2625,
+      "step": 8425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.500681481481482e-05,
+      "loss": 2.2027,
+      "step": 8430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.500385185185185e-05,
+      "loss": 2.2051,
+      "step": 8435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.500088888888889e-05,
+      "loss": 2.251,
+      "step": 8440
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.499792592592593e-05,
+      "loss": 2.2352,
+      "step": 8445
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.499496296296297e-05,
+      "loss": 2.1112,
+      "step": 8450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.4992e-05,
+      "loss": 2.3001,
+      "step": 8455
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.498903703703704e-05,
+      "loss": 2.3309,
+      "step": 8460
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.4986074074074076e-05,
+      "loss": 2.4054,
+      "step": 8465
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4983111111111115e-05,
+      "loss": 2.361,
+      "step": 8470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.498014814814815e-05,
+      "loss": 2.3468,
+      "step": 8475
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4977185185185186e-05,
+      "loss": 2.1889,
+      "step": 8480
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4974222222222225e-05,
+      "loss": 2.2536,
+      "step": 8485
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4971259259259264e-05,
+      "loss": 2.2048,
+      "step": 8490
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4968296296296296e-05,
+      "loss": 2.1627,
+      "step": 8495
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4965333333333335e-05,
+      "loss": 2.1465,
+      "step": 8500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4962370370370374e-05,
+      "loss": 2.2,
+      "step": 8505
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.495940740740741e-05,
+      "loss": 2.2766,
+      "step": 8510
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.4956444444444444e-05,
+      "loss": 2.2126,
+      "step": 8515
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.495348148148148e-05,
+      "loss": 2.2234,
+      "step": 8520
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.495051851851852e-05,
+      "loss": 2.1466,
+      "step": 8525
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.494755555555556e-05,
+      "loss": 2.2597,
+      "step": 8530
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.494459259259259e-05,
+      "loss": 2.1913,
+      "step": 8535
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.494162962962963e-05,
+      "loss": 2.344,
+      "step": 8540
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.493866666666667e-05,
+      "loss": 2.3772,
+      "step": 8545
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.493570370370371e-05,
+      "loss": 2.2795,
+      "step": 8550
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.493274074074074e-05,
+      "loss": 2.285,
+      "step": 8555
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.492977777777778e-05,
+      "loss": 2.1795,
+      "step": 8560
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.492681481481482e-05,
+      "loss": 2.2825,
+      "step": 8565
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.492385185185186e-05,
+      "loss": 2.2854,
+      "step": 8570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.492088888888889e-05,
+      "loss": 2.1471,
+      "step": 8575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.491792592592593e-05,
+      "loss": 2.1668,
+      "step": 8580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.491496296296297e-05,
+      "loss": 2.1977,
+      "step": 8585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4912000000000006e-05,
+      "loss": 2.2191,
+      "step": 8590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.490903703703704e-05,
+      "loss": 2.2372,
+      "step": 8595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.490607407407408e-05,
+      "loss": 2.3301,
+      "step": 8600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.490311111111111e-05,
+      "loss": 2.1774,
+      "step": 8605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4900148148148155e-05,
+      "loss": 2.3402,
+      "step": 8610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.489718518518519e-05,
+      "loss": 2.0965,
+      "step": 8615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4894222222222226e-05,
+      "loss": 2.2759,
+      "step": 8620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.489125925925926e-05,
+      "loss": 2.1717,
+      "step": 8625
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.4888296296296303e-05,
+      "loss": 2.3114,
+      "step": 8630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4885333333333335e-05,
+      "loss": 2.0952,
+      "step": 8635
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4882370370370374e-05,
+      "loss": 2.2115,
+      "step": 8640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4879407407407406e-05,
+      "loss": 2.3288,
+      "step": 8645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.487644444444445e-05,
+      "loss": 2.0221,
+      "step": 8650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4873481481481484e-05,
+      "loss": 2.1448,
+      "step": 8655
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.487051851851852e-05,
+      "loss": 2.29,
+      "step": 8660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4867555555555555e-05,
+      "loss": 2.24,
+      "step": 8665
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.4864592592592594e-05,
+      "loss": 2.1958,
+      "step": 8670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.486162962962963e-05,
+      "loss": 2.2127,
+      "step": 8675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.485866666666667e-05,
+      "loss": 2.2432,
+      "step": 8680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.48557037037037e-05,
+      "loss": 2.2858,
+      "step": 8685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.485274074074074e-05,
+      "loss": 2.2226,
+      "step": 8690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.484977777777778e-05,
+      "loss": 2.1858,
+      "step": 8695
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.484681481481482e-05,
+      "loss": 2.3423,
+      "step": 8700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.484385185185185e-05,
+      "loss": 2.2348,
+      "step": 8705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.484088888888889e-05,
+      "loss": 2.1366,
+      "step": 8710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.483792592592593e-05,
+      "loss": 2.2243,
+      "step": 8715
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.483496296296297e-05,
+      "loss": 2.3124,
+      "step": 8720
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.4832e-05,
+      "loss": 2.1294,
+      "step": 8725
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.482903703703704e-05,
+      "loss": 2.2274,
+      "step": 8730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.482607407407408e-05,
+      "loss": 2.1691,
+      "step": 8735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.482311111111112e-05,
+      "loss": 2.2492,
+      "step": 8740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.482014814814815e-05,
+      "loss": 2.161,
+      "step": 8745
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.481718518518519e-05,
+      "loss": 2.2844,
+      "step": 8750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4814222222222227e-05,
+      "loss": 2.2789,
+      "step": 8755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4811259259259265e-05,
+      "loss": 2.3224,
+      "step": 8760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.48082962962963e-05,
+      "loss": 2.2029,
+      "step": 8765
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4805333333333336e-05,
+      "loss": 2.2366,
+      "step": 8770
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4802370370370375e-05,
+      "loss": 2.1433,
+      "step": 8775
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4799407407407414e-05,
+      "loss": 2.3118,
+      "step": 8780
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4796444444444446e-05,
+      "loss": 2.3444,
+      "step": 8785
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4793481481481485e-05,
+      "loss": 2.1308,
+      "step": 8790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.4790518518518524e-05,
+      "loss": 2.327,
+      "step": 8795
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.478755555555556e-05,
+      "loss": 2.1839,
+      "step": 8800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4784592592592594e-05,
+      "loss": 2.2696,
+      "step": 8805
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.478162962962963e-05,
+      "loss": 2.0645,
+      "step": 8810
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.477866666666667e-05,
+      "loss": 2.1307,
+      "step": 8815
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.477570370370371e-05,
+      "loss": 2.1694,
+      "step": 8820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.477274074074074e-05,
+      "loss": 2.1367,
+      "step": 8825
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4769777777777775e-05,
+      "loss": 2.3855,
+      "step": 8830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4766814814814814e-05,
+      "loss": 2.2773,
+      "step": 8835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.476385185185185e-05,
+      "loss": 2.1735,
+      "step": 8840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.476088888888889e-05,
+      "loss": 2.3049,
+      "step": 8845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.4757925925925923e-05,
+      "loss": 2.2075,
+      "step": 8850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.475496296296296e-05,
+      "loss": 2.0926,
+      "step": 8855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.4752e-05,
+      "loss": 2.241,
+      "step": 8860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.474903703703704e-05,
+      "loss": 2.2096,
+      "step": 8865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.474607407407407e-05,
+      "loss": 2.0937,
+      "step": 8870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.474311111111111e-05,
+      "loss": 2.0851,
+      "step": 8875
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.474014814814815e-05,
+      "loss": 2.2944,
+      "step": 8880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.473718518518519e-05,
+      "loss": 2.312,
+      "step": 8885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.473422222222222e-05,
+      "loss": 2.3338,
+      "step": 8890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.473125925925926e-05,
+      "loss": 2.2477,
+      "step": 8895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.47282962962963e-05,
+      "loss": 2.227,
+      "step": 8900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.472533333333334e-05,
+      "loss": 2.1642,
+      "step": 8905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.472237037037037e-05,
+      "loss": 2.1958,
+      "step": 8910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.471940740740741e-05,
+      "loss": 2.0787,
+      "step": 8915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.471644444444445e-05,
+      "loss": 2.1999,
+      "step": 8920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4713481481481486e-05,
+      "loss": 2.1504,
+      "step": 8925
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.471051851851852e-05,
+      "loss": 2.2471,
+      "step": 8930
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4707555555555556e-05,
+      "loss": 2.2049,
+      "step": 8935
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4704592592592595e-05,
+      "loss": 2.3103,
+      "step": 8940
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4701629629629634e-05,
+      "loss": 2.2926,
+      "step": 8945
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4698666666666666e-05,
+      "loss": 2.2791,
+      "step": 8950
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4695703703703705e-05,
+      "loss": 2.2122,
+      "step": 8955
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4692740740740744e-05,
+      "loss": 2.0151,
+      "step": 8960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.468977777777778e-05,
+      "loss": 2.2277,
+      "step": 8965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.4686814814814815e-05,
+      "loss": 2.1302,
+      "step": 8970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4683851851851853e-05,
+      "loss": 2.2825,
+      "step": 8975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4680888888888885e-05,
+      "loss": 2.243,
+      "step": 8980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.467792592592593e-05,
+      "loss": 2.2525,
+      "step": 8985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.467496296296296e-05,
+      "loss": 2.1146,
+      "step": 8990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4672e-05,
+      "loss": 2.1477,
+      "step": 8995
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.4669037037037034e-05,
+      "loss": 2.1913,
+      "step": 9000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.466607407407408e-05,
+      "loss": 2.3038,
+      "step": 9005
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.466311111111111e-05,
+      "loss": 2.3616,
+      "step": 9010
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.466014814814815e-05,
+      "loss": 2.2636,
+      "step": 9015
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.465718518518518e-05,
+      "loss": 2.1322,
+      "step": 9020
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.465422222222223e-05,
+      "loss": 2.2828,
+      "step": 9025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.465125925925926e-05,
+      "loss": 2.2577,
+      "step": 9030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.46482962962963e-05,
+      "loss": 2.2734,
+      "step": 9035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.464533333333333e-05,
+      "loss": 2.0607,
+      "step": 9040
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4642370370370377e-05,
+      "loss": 2.242,
+      "step": 9045
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.463940740740741e-05,
+      "loss": 2.3204,
+      "step": 9050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.463644444444445e-05,
+      "loss": 2.3941,
+      "step": 9055
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.463348148148148e-05,
+      "loss": 2.2281,
+      "step": 9060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.463051851851852e-05,
+      "loss": 2.2483,
+      "step": 9065
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.462755555555556e-05,
+      "loss": 2.1758,
+      "step": 9070
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.4624592592592596e-05,
+      "loss": 2.1309,
+      "step": 9075
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.462162962962963e-05,
+      "loss": 2.2749,
+      "step": 9080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.461866666666667e-05,
+      "loss": 2.3262,
+      "step": 9085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4615703703703706e-05,
+      "loss": 2.3108,
+      "step": 9090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4612740740740744e-05,
+      "loss": 2.3365,
+      "step": 9095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4609777777777777e-05,
+      "loss": 2.2091,
+      "step": 9100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4606814814814815e-05,
+      "loss": 2.2945,
+      "step": 9105
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4603851851851854e-05,
+      "loss": 2.2508,
+      "step": 9110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.460088888888889e-05,
+      "loss": 2.3548,
+      "step": 9115
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4597925925925925e-05,
+      "loss": 2.1927,
+      "step": 9120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4594962962962964e-05,
+      "loss": 2.0419,
+      "step": 9125
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4592e-05,
+      "loss": 2.2701,
+      "step": 9130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.458903703703704e-05,
+      "loss": 2.3501,
+      "step": 9135
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.4586074074074074e-05,
+      "loss": 2.1802,
+      "step": 9140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.458311111111111e-05,
+      "loss": 2.2476,
+      "step": 9145
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.458014814814815e-05,
+      "loss": 2.2158,
+      "step": 9150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.457718518518519e-05,
+      "loss": 2.3661,
+      "step": 9155
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.457422222222222e-05,
+      "loss": 2.2258,
+      "step": 9160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.457125925925926e-05,
+      "loss": 2.2263,
+      "step": 9165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.45682962962963e-05,
+      "loss": 2.274,
+      "step": 9170
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.456533333333334e-05,
+      "loss": 2.293,
+      "step": 9175
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.456237037037037e-05,
+      "loss": 2.272,
+      "step": 9180
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.455940740740741e-05,
+      "loss": 2.1729,
+      "step": 9185
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.455644444444445e-05,
+      "loss": 2.2191,
+      "step": 9190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.455348148148149e-05,
+      "loss": 2.1812,
+      "step": 9195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.455051851851852e-05,
+      "loss": 2.1767,
+      "step": 9200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454755555555556e-05,
+      "loss": 2.358,
+      "step": 9205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.454459259259259e-05,
+      "loss": 2.2129,
+      "step": 9210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4541629629629636e-05,
+      "loss": 2.2722,
+      "step": 9215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.453866666666667e-05,
+      "loss": 2.305,
+      "step": 9220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4535703703703706e-05,
+      "loss": 2.1439,
+      "step": 9225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.453274074074074e-05,
+      "loss": 2.1352,
+      "step": 9230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4529777777777784e-05,
+      "loss": 2.2203,
+      "step": 9235
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4526814814814816e-05,
+      "loss": 2.3053,
+      "step": 9240
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.4523851851851855e-05,
+      "loss": 2.1684,
+      "step": 9245
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.452088888888889e-05,
+      "loss": 2.1803,
+      "step": 9250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.451792592592593e-05,
+      "loss": 2.1994,
+      "step": 9255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4514962962962965e-05,
+      "loss": 2.1723,
+      "step": 9260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4512000000000003e-05,
+      "loss": 2.291,
+      "step": 9265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4509037037037035e-05,
+      "loss": 2.2842,
+      "step": 9270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450607407407408e-05,
+      "loss": 2.2466,
+      "step": 9275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450311111111111e-05,
+      "loss": 2.1951,
+      "step": 9280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.450014814814815e-05,
+      "loss": 2.2368,
+      "step": 9285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4497185185185184e-05,
+      "loss": 2.2186,
+      "step": 9290
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.449422222222222e-05,
+      "loss": 2.2374,
+      "step": 9295
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.449125925925926e-05,
+      "loss": 2.1349,
+      "step": 9300
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.44882962962963e-05,
+      "loss": 2.1731,
+      "step": 9305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.448533333333333e-05,
+      "loss": 2.1091,
+      "step": 9310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.448237037037037e-05,
+      "loss": 2.2432,
+      "step": 9315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.447940740740741e-05,
+      "loss": 2.1939,
+      "step": 9320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.447644444444445e-05,
+      "loss": 2.2088,
+      "step": 9325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.447348148148148e-05,
+      "loss": 2.2673,
+      "step": 9330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.447051851851852e-05,
+      "loss": 2.1189,
+      "step": 9335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.446755555555556e-05,
+      "loss": 2.2517,
+      "step": 9340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.44645925925926e-05,
+      "loss": 2.3645,
+      "step": 9345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.446162962962963e-05,
+      "loss": 2.2288,
+      "step": 9350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.445866666666667e-05,
+      "loss": 2.3341,
+      "step": 9355
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.445570370370371e-05,
+      "loss": 2.3481,
+      "step": 9360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4452740740740746e-05,
+      "loss": 2.2295,
+      "step": 9365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.444977777777778e-05,
+      "loss": 2.1133,
+      "step": 9370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.444681481481482e-05,
+      "loss": 2.0986,
+      "step": 9375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4443851851851856e-05,
+      "loss": 2.1724,
+      "step": 9380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4440888888888895e-05,
+      "loss": 2.1093,
+      "step": 9385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4437925925925927e-05,
+      "loss": 2.2069,
+      "step": 9390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4434962962962965e-05,
+      "loss": 2.2396,
+      "step": 9395
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4432000000000004e-05,
+      "loss": 2.1892,
+      "step": 9400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.442903703703704e-05,
+      "loss": 2.2442,
+      "step": 9405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4426074074074075e-05,
+      "loss": 2.1985,
+      "step": 9410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4423111111111114e-05,
+      "loss": 2.2573,
+      "step": 9415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.442014814814815e-05,
+      "loss": 2.2509,
+      "step": 9420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.441718518518519e-05,
+      "loss": 2.3678,
+      "step": 9425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4414222222222224e-05,
+      "loss": 2.2194,
+      "step": 9430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.441125925925926e-05,
+      "loss": 2.1909,
+      "step": 9435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4408296296296294e-05,
+      "loss": 2.3339,
+      "step": 9440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.440533333333334e-05,
+      "loss": 2.1494,
+      "step": 9445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.440237037037037e-05,
+      "loss": 2.2311,
+      "step": 9450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.439940740740741e-05,
+      "loss": 2.0915,
+      "step": 9455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.439644444444444e-05,
+      "loss": 2.2459,
+      "step": 9460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.439348148148149e-05,
+      "loss": 2.2793,
+      "step": 9465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.439051851851852e-05,
+      "loss": 2.2388,
+      "step": 9470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.438755555555556e-05,
+      "loss": 2.2024,
+      "step": 9475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.438459259259259e-05,
+      "loss": 2.3477,
+      "step": 9480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.438162962962964e-05,
+      "loss": 2.1572,
+      "step": 9485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.437866666666667e-05,
+      "loss": 2.2055,
+      "step": 9490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.437570370370371e-05,
+      "loss": 2.1077,
+      "step": 9495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.437274074074074e-05,
+      "loss": 2.3007,
+      "step": 9500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4369777777777786e-05,
+      "loss": 2.2256,
+      "step": 9505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.436681481481482e-05,
+      "loss": 2.2885,
+      "step": 9510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4363851851851856e-05,
+      "loss": 2.2536,
+      "step": 9515
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.436088888888889e-05,
+      "loss": 2.3899,
+      "step": 9520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.435792592592593e-05,
+      "loss": 2.22,
+      "step": 9525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.4354962962962966e-05,
+      "loss": 2.246,
+      "step": 9530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4352000000000005e-05,
+      "loss": 2.2212,
+      "step": 9535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.434903703703704e-05,
+      "loss": 2.1832,
+      "step": 9540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4346074074074076e-05,
+      "loss": 2.33,
+      "step": 9545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4343111111111115e-05,
+      "loss": 2.4236,
+      "step": 9550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4340148148148153e-05,
+      "loss": 2.2868,
+      "step": 9555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4337185185185186e-05,
+      "loss": 2.1766,
+      "step": 9560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4334222222222224e-05,
+      "loss": 2.3585,
+      "step": 9565
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.433125925925926e-05,
+      "loss": 2.2546,
+      "step": 9570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.43282962962963e-05,
+      "loss": 2.346,
+      "step": 9575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4325333333333334e-05,
+      "loss": 2.1642,
+      "step": 9580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.432237037037037e-05,
+      "loss": 2.115,
+      "step": 9585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.431940740740741e-05,
+      "loss": 2.2371,
+      "step": 9590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.431644444444445e-05,
+      "loss": 2.2612,
+      "step": 9595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.431348148148148e-05,
+      "loss": 2.2129,
+      "step": 9600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4310518518518515e-05,
+      "loss": 2.1105,
+      "step": 9605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.430755555555556e-05,
+      "loss": 2.1092,
+      "step": 9610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.430459259259259e-05,
+      "loss": 2.309,
+      "step": 9615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.430162962962963e-05,
+      "loss": 2.2111,
+      "step": 9620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.429866666666666e-05,
+      "loss": 2.3027,
+      "step": 9625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.429570370370371e-05,
+      "loss": 2.347,
+      "step": 9630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.429274074074074e-05,
+      "loss": 2.308,
+      "step": 9635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.428977777777778e-05,
+      "loss": 2.3214,
+      "step": 9640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.428681481481481e-05,
+      "loss": 2.3536,
+      "step": 9645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.428385185185186e-05,
+      "loss": 2.3462,
+      "step": 9650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.428088888888889e-05,
+      "loss": 2.2424,
+      "step": 9655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.427792592592593e-05,
+      "loss": 2.2741,
+      "step": 9660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.427496296296296e-05,
+      "loss": 2.2121,
+      "step": 9665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4272000000000006e-05,
+      "loss": 2.2787,
+      "step": 9670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.426903703703704e-05,
+      "loss": 2.2397,
+      "step": 9675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4266074074074077e-05,
+      "loss": 2.3025,
+      "step": 9680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.426311111111111e-05,
+      "loss": 2.2162,
+      "step": 9685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.426014814814815e-05,
+      "loss": 2.1911,
+      "step": 9690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4257185185185186e-05,
+      "loss": 2.2748,
+      "step": 9695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.4254222222222225e-05,
+      "loss": 2.1526,
+      "step": 9700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.425125925925926e-05,
+      "loss": 2.2678,
+      "step": 9705
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4248296296296296e-05,
+      "loss": 2.268,
+      "step": 9710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4245333333333335e-05,
+      "loss": 2.2567,
+      "step": 9715
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4242370370370374e-05,
+      "loss": 2.2341,
+      "step": 9720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4239407407407406e-05,
+      "loss": 2.2499,
+      "step": 9725
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4236444444444445e-05,
+      "loss": 2.2566,
+      "step": 9730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.423348148148148e-05,
+      "loss": 2.281,
+      "step": 9735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.423051851851852e-05,
+      "loss": 2.1753,
+      "step": 9740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.4227555555555554e-05,
+      "loss": 2.292,
+      "step": 9745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.422459259259259e-05,
+      "loss": 2.1486,
+      "step": 9750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.422162962962963e-05,
+      "loss": 2.1281,
+      "step": 9755
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.421866666666667e-05,
+      "loss": 2.2281,
+      "step": 9760
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.42157037037037e-05,
+      "loss": 2.2341,
+      "step": 9765
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.421274074074074e-05,
+      "loss": 2.1116,
+      "step": 9770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.420977777777778e-05,
+      "loss": 2.1859,
+      "step": 9775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.420681481481482e-05,
+      "loss": 2.2123,
+      "step": 9780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.420385185185185e-05,
+      "loss": 2.1039,
+      "step": 9785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.420088888888889e-05,
+      "loss": 2.3223,
+      "step": 9790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.419792592592593e-05,
+      "loss": 2.2018,
+      "step": 9795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.419496296296297e-05,
+      "loss": 2.1534,
+      "step": 9800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.4192e-05,
+      "loss": 2.1754,
+      "step": 9805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.418903703703704e-05,
+      "loss": 2.1769,
+      "step": 9810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.418607407407408e-05,
+      "loss": 2.1825,
+      "step": 9815
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4183111111111116e-05,
+      "loss": 2.065,
+      "step": 9820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.418014814814815e-05,
+      "loss": 2.1738,
+      "step": 9825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.417718518518519e-05,
+      "loss": 2.1448,
+      "step": 9830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.417422222222222e-05,
+      "loss": 2.1492,
+      "step": 9835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4171259259259265e-05,
+      "loss": 2.298,
+      "step": 9840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.41682962962963e-05,
+      "loss": 2.1479,
+      "step": 9845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4165333333333336e-05,
+      "loss": 2.2852,
+      "step": 9850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.416237037037037e-05,
+      "loss": 2.0958,
+      "step": 9855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.415940740740741e-05,
+      "loss": 2.276,
+      "step": 9860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4156444444444445e-05,
+      "loss": 2.1492,
+      "step": 9865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.4153481481481484e-05,
+      "loss": 2.3101,
+      "step": 9870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.4150518518518516e-05,
+      "loss": 2.2756,
+      "step": 9875
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.414755555555556e-05,
+      "loss": 2.3224,
+      "step": 9880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.4144592592592594e-05,
+      "loss": 2.2795,
+      "step": 9885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.414162962962963e-05,
+      "loss": 2.1955,
+      "step": 9890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.4138666666666665e-05,
+      "loss": 2.2822,
+      "step": 9895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.413570370370371e-05,
+      "loss": 2.2868,
+      "step": 9900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.413274074074074e-05,
+      "loss": 2.1663,
+      "step": 9905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.412977777777778e-05,
+      "loss": 2.3776,
+      "step": 9910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.412681481481481e-05,
+      "loss": 2.276,
+      "step": 9915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.412385185185185e-05,
+      "loss": 2.1808,
+      "step": 9920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.412088888888889e-05,
+      "loss": 2.1493,
+      "step": 9925
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.411792592592593e-05,
+      "loss": 2.1838,
+      "step": 9930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.411496296296296e-05,
+      "loss": 2.2366,
+      "step": 9935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.4112e-05,
+      "loss": 2.228,
+      "step": 9940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.410903703703704e-05,
+      "loss": 2.3133,
+      "step": 9945
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.410607407407408e-05,
+      "loss": 2.1735,
+      "step": 9950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.410311111111111e-05,
+      "loss": 2.3266,
+      "step": 9955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.410014814814815e-05,
+      "loss": 2.2632,
+      "step": 9960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.409718518518519e-05,
+      "loss": 2.164,
+      "step": 9965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.409422222222223e-05,
+      "loss": 2.2565,
+      "step": 9970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.409125925925926e-05,
+      "loss": 2.0973,
+      "step": 9975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.40882962962963e-05,
+      "loss": 2.2179,
+      "step": 9980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4085333333333336e-05,
+      "loss": 2.1716,
+      "step": 9985
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4082370370370375e-05,
+      "loss": 2.061,
+      "step": 9990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.407940740740741e-05,
+      "loss": 2.0537,
+      "step": 9995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4076444444444446e-05,
+      "loss": 2.1978,
+      "step": 10000
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 2.0439772605895996,
+      "eval_rouge2_fmeasure": 0.1667,
+      "eval_rouge2_precision": 0.2047,
+      "eval_rouge2_recall": 0.1491,
+      "eval_runtime": 34261.036,
+      "eval_samples_per_second": 0.146,
+      "eval_steps_per_second": 0.073,
+      "step": 10000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4073481481481485e-05,
+      "loss": 2.31,
+      "step": 10005
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4070518518518524e-05,
+      "loss": 2.2195,
+      "step": 10010
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4067555555555556e-05,
+      "loss": 2.3401,
+      "step": 10015
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4064592592592595e-05,
+      "loss": 2.2386,
+      "step": 10020
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.406162962962963e-05,
+      "loss": 1.9536,
+      "step": 10025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.405866666666667e-05,
+      "loss": 2.2169,
+      "step": 10030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4055703703703704e-05,
+      "loss": 2.286,
+      "step": 10035
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.405274074074074e-05,
+      "loss": 2.2227,
+      "step": 10040
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.404977777777778e-05,
+      "loss": 2.1438,
+      "step": 10045
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.404681481481482e-05,
+      "loss": 2.2947,
+      "step": 10050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.404385185185185e-05,
+      "loss": 2.2414,
+      "step": 10055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.404088888888889e-05,
+      "loss": 2.1051,
+      "step": 10060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4037925925925924e-05,
+      "loss": 2.2663,
+      "step": 10065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.403496296296297e-05,
+      "loss": 2.1975,
+      "step": 10070
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.4032e-05,
+      "loss": 2.2077,
+      "step": 10075
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.402903703703704e-05,
+      "loss": 2.2922,
+      "step": 10080
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.402607407407407e-05,
+      "loss": 2.1373,
+      "step": 10085
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.402311111111112e-05,
+      "loss": 2.2895,
+      "step": 10090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.402014814814815e-05,
+      "loss": 2.1627,
+      "step": 10095
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.401718518518519e-05,
+      "loss": 2.1977,
+      "step": 10100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.401422222222222e-05,
+      "loss": 2.2384,
+      "step": 10105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.4011259259259266e-05,
+      "loss": 2.1434,
+      "step": 10110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.40082962962963e-05,
+      "loss": 2.1962,
+      "step": 10115
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.400533333333334e-05,
+      "loss": 2.2206,
+      "step": 10120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.400237037037037e-05,
+      "loss": 2.1588,
+      "step": 10125
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3999407407407415e-05,
+      "loss": 2.1711,
+      "step": 10130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.399644444444445e-05,
+      "loss": 2.202,
+      "step": 10135
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3993481481481486e-05,
+      "loss": 2.2683,
+      "step": 10140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.399051851851852e-05,
+      "loss": 2.3309,
+      "step": 10145
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3987555555555556e-05,
+      "loss": 2.1929,
+      "step": 10150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3984592592592595e-05,
+      "loss": 2.2509,
+      "step": 10155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3981629629629634e-05,
+      "loss": 2.226,
+      "step": 10160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3978666666666666e-05,
+      "loss": 2.1327,
+      "step": 10165
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3975703703703705e-05,
+      "loss": 2.3077,
+      "step": 10170
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3972740740740744e-05,
+      "loss": 2.2088,
+      "step": 10175
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.396977777777778e-05,
+      "loss": 2.2068,
+      "step": 10180
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3966814814814815e-05,
+      "loss": 2.2497,
+      "step": 10185
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.3963851851851854e-05,
+      "loss": 2.268,
+      "step": 10190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.396088888888889e-05,
+      "loss": 2.2018,
+      "step": 10195
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.395792592592593e-05,
+      "loss": 2.2625,
+      "step": 10200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.395496296296296e-05,
+      "loss": 2.2081,
+      "step": 10205
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.3952e-05,
+      "loss": 2.1937,
+      "step": 10210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.394903703703704e-05,
+      "loss": 2.2145,
+      "step": 10215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.394607407407408e-05,
+      "loss": 2.2441,
+      "step": 10220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.394311111111111e-05,
+      "loss": 2.29,
+      "step": 10225
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.394014814814815e-05,
+      "loss": 2.0711,
+      "step": 10230
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.393718518518519e-05,
+      "loss": 2.2448,
+      "step": 10235
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.393422222222223e-05,
+      "loss": 2.0398,
+      "step": 10240
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.393125925925926e-05,
+      "loss": 2.1704,
+      "step": 10245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.39282962962963e-05,
+      "loss": 2.1463,
+      "step": 10250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.392533333333334e-05,
+      "loss": 2.153,
+      "step": 10255
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.392237037037038e-05,
+      "loss": 2.2118,
+      "step": 10260
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.391940740740741e-05,
+      "loss": 2.245,
+      "step": 10265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.391644444444445e-05,
+      "loss": 2.2596,
+      "step": 10270
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3913481481481486e-05,
+      "loss": 2.2641,
+      "step": 10275
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3910518518518525e-05,
+      "loss": 2.0554,
+      "step": 10280
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.390755555555556e-05,
+      "loss": 2.2307,
+      "step": 10285
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3904592592592596e-05,
+      "loss": 2.2338,
+      "step": 10290
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.390162962962963e-05,
+      "loss": 2.2213,
+      "step": 10295
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3898666666666674e-05,
+      "loss": 2.2879,
+      "step": 10300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3895703703703706e-05,
+      "loss": 2.3313,
+      "step": 10305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.3892740740740745e-05,
+      "loss": 2.1257,
+      "step": 10310
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.388977777777778e-05,
+      "loss": 2.3254,
+      "step": 10315
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.388681481481482e-05,
+      "loss": 2.3248,
+      "step": 10320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3883851851851854e-05,
+      "loss": 2.3274,
+      "step": 10325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.388088888888889e-05,
+      "loss": 2.2155,
+      "step": 10330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3877925925925925e-05,
+      "loss": 2.1047,
+      "step": 10335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.387496296296297e-05,
+      "loss": 2.2408,
+      "step": 10340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3872e-05,
+      "loss": 2.2722,
+      "step": 10345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.386903703703704e-05,
+      "loss": 2.191,
+      "step": 10350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.3866074074074074e-05,
+      "loss": 2.2256,
+      "step": 10355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.386311111111112e-05,
+      "loss": 2.3582,
+      "step": 10360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.386014814814815e-05,
+      "loss": 2.2479,
+      "step": 10365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.385718518518519e-05,
+      "loss": 2.0827,
+      "step": 10370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.385422222222222e-05,
+      "loss": 2.1383,
+      "step": 10375
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.385125925925926e-05,
+      "loss": 2.2418,
+      "step": 10380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.38482962962963e-05,
+      "loss": 2.2962,
+      "step": 10385
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.384533333333333e-05,
+      "loss": 2.2133,
+      "step": 10390
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.384237037037037e-05,
+      "loss": 2.2497,
+      "step": 10395
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.383940740740741e-05,
+      "loss": 2.1867,
+      "step": 10400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.383644444444445e-05,
+      "loss": 2.2258,
+      "step": 10405
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.383348148148148e-05,
+      "loss": 2.0529,
+      "step": 10410
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.383051851851852e-05,
+      "loss": 2.2255,
+      "step": 10415
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.382755555555556e-05,
+      "loss": 2.1513,
+      "step": 10420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.38245925925926e-05,
+      "loss": 2.2955,
+      "step": 10425
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.382162962962963e-05,
+      "loss": 2.1093,
+      "step": 10430
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.381866666666667e-05,
+      "loss": 2.2788,
+      "step": 10435
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3815703703703707e-05,
+      "loss": 2.3515,
+      "step": 10440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3812740740740745e-05,
+      "loss": 2.1513,
+      "step": 10445
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.380977777777778e-05,
+      "loss": 2.2426,
+      "step": 10450
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3806814814814816e-05,
+      "loss": 2.1481,
+      "step": 10455
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.380385185185185e-05,
+      "loss": 2.2667,
+      "step": 10460
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3800888888888894e-05,
+      "loss": 2.2408,
+      "step": 10465
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3797925925925926e-05,
+      "loss": 2.2682,
+      "step": 10470
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3794962962962965e-05,
+      "loss": 2.2294,
+      "step": 10475
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3792e-05,
+      "loss": 2.2253,
+      "step": 10480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.378903703703704e-05,
+      "loss": 2.2201,
+      "step": 10485
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.3786074074074074e-05,
+      "loss": 2.203,
+      "step": 10490
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.378311111111111e-05,
+      "loss": 2.2442,
+      "step": 10495
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3780148148148145e-05,
+      "loss": 2.1727,
+      "step": 10500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.377718518518519e-05,
+      "loss": 2.3798,
+      "step": 10505
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.377422222222222e-05,
+      "loss": 2.0875,
+      "step": 10510
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.377125925925926e-05,
+      "loss": 2.0281,
+      "step": 10515
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.3768296296296294e-05,
+      "loss": 2.1237,
+      "step": 10520
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.376533333333333e-05,
+      "loss": 2.1481,
+      "step": 10525
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.376237037037037e-05,
+      "loss": 2.2084,
+      "step": 10530
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.375940740740741e-05,
+      "loss": 2.1602,
+      "step": 10535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.375644444444444e-05,
+      "loss": 2.1144,
+      "step": 10540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.375348148148148e-05,
+      "loss": 2.2005,
+      "step": 10545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.375051851851852e-05,
+      "loss": 2.2358,
+      "step": 10550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.374755555555556e-05,
+      "loss": 2.13,
+      "step": 10555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.374459259259259e-05,
+      "loss": 2.3751,
+      "step": 10560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.374162962962963e-05,
+      "loss": 2.2317,
+      "step": 10565
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.373866666666667e-05,
+      "loss": 2.2669,
+      "step": 10570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.373570370370371e-05,
+      "loss": 2.1224,
+      "step": 10575
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.373274074074074e-05,
+      "loss": 2.2104,
+      "step": 10580
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.372977777777778e-05,
+      "loss": 2.2166,
+      "step": 10585
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.372681481481482e-05,
+      "loss": 2.1757,
+      "step": 10590
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.3723851851851856e-05,
+      "loss": 2.2154,
+      "step": 10595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.372088888888889e-05,
+      "loss": 2.3143,
+      "step": 10600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.371792592592593e-05,
+      "loss": 2.3835,
+      "step": 10605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3714962962962966e-05,
+      "loss": 2.1537,
+      "step": 10610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3712000000000004e-05,
+      "loss": 2.2768,
+      "step": 10615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3709037037037036e-05,
+      "loss": 2.1266,
+      "step": 10620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3706074074074075e-05,
+      "loss": 2.1756,
+      "step": 10625
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3703111111111114e-05,
+      "loss": 2.2417,
+      "step": 10630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.370014814814815e-05,
+      "loss": 2.17,
+      "step": 10635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3697185185185185e-05,
+      "loss": 2.2912,
+      "step": 10640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.3694222222222224e-05,
+      "loss": 2.2549,
+      "step": 10645
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.369125925925926e-05,
+      "loss": 2.0107,
+      "step": 10650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.36882962962963e-05,
+      "loss": 2.3948,
+      "step": 10655
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.3685333333333333e-05,
+      "loss": 2.2989,
+      "step": 10660
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.368237037037037e-05,
+      "loss": 2.2562,
+      "step": 10665
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.367940740740741e-05,
+      "loss": 2.1361,
+      "step": 10670
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.367644444444445e-05,
+      "loss": 2.2095,
+      "step": 10675
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.367348148148148e-05,
+      "loss": 2.2919,
+      "step": 10680
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.367051851851852e-05,
+      "loss": 2.1821,
+      "step": 10685
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.366755555555555e-05,
+      "loss": 2.2549,
+      "step": 10690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.36645925925926e-05,
+      "loss": 2.3804,
+      "step": 10695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.366162962962963e-05,
+      "loss": 2.3303,
+      "step": 10700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.365866666666667e-05,
+      "loss": 2.111,
+      "step": 10705
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.36557037037037e-05,
+      "loss": 2.1768,
+      "step": 10710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.365274074074075e-05,
+      "loss": 2.0743,
+      "step": 10715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.364977777777778e-05,
+      "loss": 1.9922,
+      "step": 10720
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.364681481481482e-05,
+      "loss": 2.1965,
+      "step": 10725
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.364385185185185e-05,
+      "loss": 2.314,
+      "step": 10730
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3640888888888895e-05,
+      "loss": 2.2277,
+      "step": 10735
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.363792592592593e-05,
+      "loss": 2.2343,
+      "step": 10740
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3634962962962966e-05,
+      "loss": 2.0696,
+      "step": 10745
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3632e-05,
+      "loss": 2.1953,
+      "step": 10750
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.362903703703704e-05,
+      "loss": 2.2502,
+      "step": 10755
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3626074074074076e-05,
+      "loss": 2.1502,
+      "step": 10760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.3623111111111115e-05,
+      "loss": 2.1763,
+      "step": 10765
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.362014814814815e-05,
+      "loss": 2.1015,
+      "step": 10770
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3617185185185186e-05,
+      "loss": 2.2108,
+      "step": 10775
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3614222222222224e-05,
+      "loss": 2.266,
+      "step": 10780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.361125925925926e-05,
+      "loss": 2.3031,
+      "step": 10785
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3608296296296295e-05,
+      "loss": 2.2477,
+      "step": 10790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3605333333333334e-05,
+      "loss": 2.2424,
+      "step": 10795
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.360237037037037e-05,
+      "loss": 2.1454,
+      "step": 10800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.359940740740741e-05,
+      "loss": 2.1509,
+      "step": 10805
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.3596444444444444e-05,
+      "loss": 2.1308,
+      "step": 10810
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.359348148148148e-05,
+      "loss": 2.2903,
+      "step": 10815
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.359051851851852e-05,
+      "loss": 2.1762,
+      "step": 10820
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.358755555555556e-05,
+      "loss": 2.1965,
+      "step": 10825
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.358459259259259e-05,
+      "loss": 2.2008,
+      "step": 10830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.358162962962963e-05,
+      "loss": 2.3109,
+      "step": 10835
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.357866666666667e-05,
+      "loss": 2.0748,
+      "step": 10840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.357570370370371e-05,
+      "loss": 2.085,
+      "step": 10845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.357274074074074e-05,
+      "loss": 2.206,
+      "step": 10850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.356977777777778e-05,
+      "loss": 2.0983,
+      "step": 10855
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.356681481481482e-05,
+      "loss": 2.1558,
+      "step": 10860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.356385185185186e-05,
+      "loss": 2.1374,
+      "step": 10865
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.356088888888889e-05,
+      "loss": 2.2158,
+      "step": 10870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.355792592592593e-05,
+      "loss": 2.1164,
+      "step": 10875
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.355496296296297e-05,
+      "loss": 2.1895,
+      "step": 10880
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3552000000000006e-05,
+      "loss": 2.2881,
+      "step": 10885
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.354903703703704e-05,
+      "loss": 2.272,
+      "step": 10890
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.354607407407408e-05,
+      "loss": 2.0539,
+      "step": 10895
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3543111111111116e-05,
+      "loss": 2.304,
+      "step": 10900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3540148148148154e-05,
+      "loss": 2.1847,
+      "step": 10905
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3537185185185186e-05,
+      "loss": 2.1715,
+      "step": 10910
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3534222222222225e-05,
+      "loss": 2.1479,
+      "step": 10915
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.353125925925926e-05,
+      "loss": 2.3192,
+      "step": 10920
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.35282962962963e-05,
+      "loss": 2.2156,
+      "step": 10925
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3525333333333335e-05,
+      "loss": 2.2539,
+      "step": 10930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3522370370370374e-05,
+      "loss": 2.1328,
+      "step": 10935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.3519407407407406e-05,
+      "loss": 2.029,
+      "step": 10940
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.351644444444445e-05,
+      "loss": 2.0718,
+      "step": 10945
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3513481481481483e-05,
+      "loss": 2.2239,
+      "step": 10950
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.351051851851852e-05,
+      "loss": 2.2558,
+      "step": 10955
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.3507555555555554e-05,
+      "loss": 2.3063,
+      "step": 10960
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.35045925925926e-05,
+      "loss": 2.2471,
+      "step": 10965
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.350162962962963e-05,
+      "loss": 2.231,
+      "step": 10970
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.349866666666667e-05,
+      "loss": 2.3588,
+      "step": 10975
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.34957037037037e-05,
+      "loss": 2.0912,
+      "step": 10980
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.349274074074074e-05,
+      "loss": 2.2122,
+      "step": 10985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.348977777777778e-05,
+      "loss": 2.3138,
+      "step": 10990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.348681481481482e-05,
+      "loss": 2.2255,
+      "step": 10995
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.348385185185185e-05,
+      "loss": 2.3409,
+      "step": 11000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.348088888888889e-05,
+      "loss": 2.3846,
+      "step": 11005
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.347792592592593e-05,
+      "loss": 2.1558,
+      "step": 11010
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.347496296296297e-05,
+      "loss": 2.296,
+      "step": 11015
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3472e-05,
+      "loss": 2.2548,
+      "step": 11020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.346903703703704e-05,
+      "loss": 2.0732,
+      "step": 11025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.346607407407408e-05,
+      "loss": 2.1148,
+      "step": 11030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3463111111111116e-05,
+      "loss": 2.2872,
+      "step": 11035
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.346014814814815e-05,
+      "loss": 2.1739,
+      "step": 11040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.345718518518519e-05,
+      "loss": 2.2737,
+      "step": 11045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.3454222222222226e-05,
+      "loss": 2.2568,
+      "step": 11050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3451259259259265e-05,
+      "loss": 2.2822,
+      "step": 11055
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.34482962962963e-05,
+      "loss": 2.173,
+      "step": 11060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3445333333333336e-05,
+      "loss": 2.2033,
+      "step": 11065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3442370370370375e-05,
+      "loss": 2.316,
+      "step": 11070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.343940740740741e-05,
+      "loss": 2.2675,
+      "step": 11075
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3436444444444445e-05,
+      "loss": 2.2084,
+      "step": 11080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3433481481481484e-05,
+      "loss": 2.1887,
+      "step": 11085
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.343051851851852e-05,
+      "loss": 2.2739,
+      "step": 11090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.342755555555556e-05,
+      "loss": 2.2122,
+      "step": 11095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.3424592592592594e-05,
+      "loss": 2.2178,
+      "step": 11100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.342162962962963e-05,
+      "loss": 2.1574,
+      "step": 11105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.341866666666667e-05,
+      "loss": 2.2147,
+      "step": 11110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.341570370370371e-05,
+      "loss": 2.1124,
+      "step": 11115
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.341274074074074e-05,
+      "loss": 2.1672,
+      "step": 11120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.340977777777778e-05,
+      "loss": 2.2424,
+      "step": 11125
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.340681481481482e-05,
+      "loss": 2.2263,
+      "step": 11130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.340385185185186e-05,
+      "loss": 2.1193,
+      "step": 11135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.340088888888889e-05,
+      "loss": 2.2944,
+      "step": 11140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.339792592592593e-05,
+      "loss": 2.2848,
+      "step": 11145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.339496296296296e-05,
+      "loss": 2.3274,
+      "step": 11150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.3392e-05,
+      "loss": 2.2382,
+      "step": 11155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.338903703703704e-05,
+      "loss": 2.1853,
+      "step": 11160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.338607407407407e-05,
+      "loss": 2.1775,
+      "step": 11165
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.338311111111111e-05,
+      "loss": 2.2032,
+      "step": 11170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.338014814814815e-05,
+      "loss": 2.1721,
+      "step": 11175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.337718518518519e-05,
+      "loss": 2.239,
+      "step": 11180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.337422222222222e-05,
+      "loss": 2.335,
+      "step": 11185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.337125925925926e-05,
+      "loss": 2.2095,
+      "step": 11190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.33682962962963e-05,
+      "loss": 2.3582,
+      "step": 11195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3365333333333336e-05,
+      "loss": 2.2294,
+      "step": 11200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.336237037037037e-05,
+      "loss": 2.1753,
+      "step": 11205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.335940740740741e-05,
+      "loss": 2.3546,
+      "step": 11210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3356444444444446e-05,
+      "loss": 2.2901,
+      "step": 11215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.3353481481481485e-05,
+      "loss": 2.1788,
+      "step": 11220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.335051851851852e-05,
+      "loss": 2.2092,
+      "step": 11225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3347555555555556e-05,
+      "loss": 2.0376,
+      "step": 11230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3344592592592595e-05,
+      "loss": 2.1752,
+      "step": 11235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3341629629629633e-05,
+      "loss": 2.1688,
+      "step": 11240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3338666666666666e-05,
+      "loss": 2.2212,
+      "step": 11245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3335703703703704e-05,
+      "loss": 2.2072,
+      "step": 11250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.333274074074074e-05,
+      "loss": 2.0503,
+      "step": 11255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.332977777777778e-05,
+      "loss": 1.8864,
+      "step": 11260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.3326814814814814e-05,
+      "loss": 1.9188,
+      "step": 11265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.332385185185185e-05,
+      "loss": 2.1106,
+      "step": 11270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.332088888888889e-05,
+      "loss": 2.105,
+      "step": 11275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.331792592592593e-05,
+      "loss": 2.08,
+      "step": 11280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.331496296296296e-05,
+      "loss": 2.0219,
+      "step": 11285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.3312e-05,
+      "loss": 2.1002,
+      "step": 11290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.330903703703704e-05,
+      "loss": 2.1142,
+      "step": 11295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.330607407407408e-05,
+      "loss": 2.1429,
+      "step": 11300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.330311111111111e-05,
+      "loss": 1.9445,
+      "step": 11305
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.330014814814815e-05,
+      "loss": 1.9515,
+      "step": 11310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.329718518518518e-05,
+      "loss": 2.2021,
+      "step": 11315
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.329422222222223e-05,
+      "loss": 2.0118,
+      "step": 11320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.329125925925926e-05,
+      "loss": 2.0163,
+      "step": 11325
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.32882962962963e-05,
+      "loss": 2.0078,
+      "step": 11330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.328533333333333e-05,
+      "loss": 2.0841,
+      "step": 11335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3282370370370376e-05,
+      "loss": 2.0808,
+      "step": 11340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.327940740740741e-05,
+      "loss": 2.0091,
+      "step": 11345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.327644444444445e-05,
+      "loss": 2.1778,
+      "step": 11350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.327348148148148e-05,
+      "loss": 2.0817,
+      "step": 11355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3270518518518525e-05,
+      "loss": 2.1222,
+      "step": 11360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3267555555555557e-05,
+      "loss": 2.156,
+      "step": 11365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3264592592592595e-05,
+      "loss": 2.0951,
+      "step": 11370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.326162962962963e-05,
+      "loss": 2.0884,
+      "step": 11375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3258666666666666e-05,
+      "loss": 2.1757,
+      "step": 11380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3255703703703705e-05,
+      "loss": 1.9914,
+      "step": 11385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.3252740740740744e-05,
+      "loss": 2.1065,
+      "step": 11390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3249777777777776e-05,
+      "loss": 2.1114,
+      "step": 11395
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3246814814814815e-05,
+      "loss": 2.0487,
+      "step": 11400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3243851851851854e-05,
+      "loss": 2.0324,
+      "step": 11405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.324088888888889e-05,
+      "loss": 2.0874,
+      "step": 11410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3237925925925924e-05,
+      "loss": 1.9281,
+      "step": 11415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.323496296296296e-05,
+      "loss": 2.002,
+      "step": 11420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.3232e-05,
+      "loss": 1.9988,
+      "step": 11425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.322903703703704e-05,
+      "loss": 2.0591,
+      "step": 11430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.322607407407407e-05,
+      "loss": 2.2524,
+      "step": 11435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.322311111111111e-05,
+      "loss": 2.1542,
+      "step": 11440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.322014814814815e-05,
+      "loss": 2.1468,
+      "step": 11445
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.321718518518519e-05,
+      "loss": 2.1932,
+      "step": 11450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.321422222222222e-05,
+      "loss": 1.9458,
+      "step": 11455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.321125925925926e-05,
+      "loss": 2.178,
+      "step": 11460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.32082962962963e-05,
+      "loss": 2.0837,
+      "step": 11465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.320533333333334e-05,
+      "loss": 2.0219,
+      "step": 11470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.320237037037037e-05,
+      "loss": 2.0984,
+      "step": 11475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.319940740740741e-05,
+      "loss": 2.0556,
+      "step": 11480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.319644444444445e-05,
+      "loss": 2.0566,
+      "step": 11485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.3193481481481487e-05,
+      "loss": 2.1313,
+      "step": 11490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.319051851851852e-05,
+      "loss": 2.1925,
+      "step": 11495
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.318755555555556e-05,
+      "loss": 2.0852,
+      "step": 11500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3184592592592596e-05,
+      "loss": 2.1343,
+      "step": 11505
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3181629629629635e-05,
+      "loss": 2.0709,
+      "step": 11510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.317866666666667e-05,
+      "loss": 2.0256,
+      "step": 11515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3175703703703706e-05,
+      "loss": 2.0816,
+      "step": 11520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3172740740740745e-05,
+      "loss": 2.1531,
+      "step": 11525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3169777777777784e-05,
+      "loss": 2.1405,
+      "step": 11530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3166814814814816e-05,
+      "loss": 2.0763,
+      "step": 11535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3163851851851854e-05,
+      "loss": 2.1567,
+      "step": 11540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3160888888888886e-05,
+      "loss": 2.0646,
+      "step": 11545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.315792592592593e-05,
+      "loss": 2.0905,
+      "step": 11550
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.3154962962962964e-05,
+      "loss": 2.1443,
+      "step": 11555
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.3152e-05,
+      "loss": 2.2026,
+      "step": 11560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.3149037037037035e-05,
+      "loss": 2.1737,
+      "step": 11565
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.314607407407408e-05,
+      "loss": 2.0629,
+      "step": 11570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.314311111111111e-05,
+      "loss": 2.1468,
+      "step": 11575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.314014814814815e-05,
+      "loss": 2.0219,
+      "step": 11580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.3137185185185183e-05,
+      "loss": 2.117,
+      "step": 11585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.313422222222223e-05,
+      "loss": 2.0584,
+      "step": 11590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.313125925925926e-05,
+      "loss": 2.0323,
+      "step": 11595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.31282962962963e-05,
+      "loss": 2.0484,
+      "step": 11600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.312533333333333e-05,
+      "loss": 2.1403,
+      "step": 11605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.312237037037037e-05,
+      "loss": 2.0259,
+      "step": 11610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.311940740740741e-05,
+      "loss": 2.0897,
+      "step": 11615
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.311644444444445e-05,
+      "loss": 2.2139,
+      "step": 11620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.311348148148148e-05,
+      "loss": 2.1341,
+      "step": 11625
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.311051851851852e-05,
+      "loss": 2.0687,
+      "step": 11630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.310755555555556e-05,
+      "loss": 2.028,
+      "step": 11635
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.31045925925926e-05,
+      "loss": 2.1242,
+      "step": 11640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.310162962962963e-05,
+      "loss": 2.0842,
+      "step": 11645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.309866666666667e-05,
+      "loss": 2.1029,
+      "step": 11650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.309570370370371e-05,
+      "loss": 2.1054,
+      "step": 11655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.3092740740740745e-05,
+      "loss": 2.0464,
+      "step": 11660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.308977777777778e-05,
+      "loss": 2.1943,
+      "step": 11665
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.3086814814814816e-05,
+      "loss": 2.08,
+      "step": 11670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3083851851851855e-05,
+      "loss": 2.1416,
+      "step": 11675
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3080888888888894e-05,
+      "loss": 2.0655,
+      "step": 11680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3077925925925926e-05,
+      "loss": 2.0228,
+      "step": 11685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3074962962962965e-05,
+      "loss": 2.0516,
+      "step": 11690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3072000000000004e-05,
+      "loss": 2.0225,
+      "step": 11695
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.306903703703704e-05,
+      "loss": 2.0841,
+      "step": 11700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.3066074074074075e-05,
+      "loss": 1.9781,
+      "step": 11705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.306311111111111e-05,
+      "loss": 1.9715,
+      "step": 11710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.306014814814815e-05,
+      "loss": 2.0729,
+      "step": 11715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.305718518518519e-05,
+      "loss": 2.0646,
+      "step": 11720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.305422222222222e-05,
+      "loss": 2.2353,
+      "step": 11725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.305125925925926e-05,
+      "loss": 2.0217,
+      "step": 11730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.30482962962963e-05,
+      "loss": 2.0151,
+      "step": 11735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.304533333333334e-05,
+      "loss": 2.1185,
+      "step": 11740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.304237037037037e-05,
+      "loss": 1.9818,
+      "step": 11745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.303940740740741e-05,
+      "loss": 2.0689,
+      "step": 11750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.303644444444445e-05,
+      "loss": 1.9555,
+      "step": 11755
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.303348148148149e-05,
+      "loss": 2.1376,
+      "step": 11760
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.303051851851852e-05,
+      "loss": 1.9713,
+      "step": 11765
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.302755555555556e-05,
+      "loss": 2.082,
+      "step": 11770
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.302459259259259e-05,
+      "loss": 2.056,
+      "step": 11775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.3021629629629637e-05,
+      "loss": 2.2239,
+      "step": 11780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.301866666666667e-05,
+      "loss": 2.1189,
+      "step": 11785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.301570370370371e-05,
+      "loss": 2.1475,
+      "step": 11790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.301274074074074e-05,
+      "loss": 2.1659,
+      "step": 11795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3009777777777785e-05,
+      "loss": 1.982,
+      "step": 11800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.300681481481482e-05,
+      "loss": 2.2411,
+      "step": 11805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3003851851851856e-05,
+      "loss": 2.1064,
+      "step": 11810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.300088888888889e-05,
+      "loss": 2.0269,
+      "step": 11815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2997925925925934e-05,
+      "loss": 2.292,
+      "step": 11820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2994962962962966e-05,
+      "loss": 1.9721,
+      "step": 11825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2992000000000004e-05,
+      "loss": 2.0659,
+      "step": 11830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2989037037037036e-05,
+      "loss": 1.9607,
+      "step": 11835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.2986074074074075e-05,
+      "loss": 2.1082,
+      "step": 11840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2983111111111114e-05,
+      "loss": 2.0569,
+      "step": 11845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298014814814815e-05,
+      "loss": 2.266,
+      "step": 11850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2977185185185185e-05,
+      "loss": 2.117,
+      "step": 11855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2974222222222224e-05,
+      "loss": 2.0836,
+      "step": 11860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297125925925926e-05,
+      "loss": 2.057,
+      "step": 11865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29682962962963e-05,
+      "loss": 2.0447,
+      "step": 11870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2965333333333334e-05,
+      "loss": 2.163,
+      "step": 11875
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.296237037037037e-05,
+      "loss": 2.0831,
+      "step": 11880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295940740740741e-05,
+      "loss": 2.1052,
+      "step": 11885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295644444444445e-05,
+      "loss": 2.0914,
+      "step": 11890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295348148148148e-05,
+      "loss": 2.0496,
+      "step": 11895
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.295051851851852e-05,
+      "loss": 2.0744,
+      "step": 11900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.294755555555556e-05,
+      "loss": 2.2027,
+      "step": 11905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.29445925925926e-05,
+      "loss": 2.0048,
+      "step": 11910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.294162962962963e-05,
+      "loss": 2.0773,
+      "step": 11915
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.293866666666667e-05,
+      "loss": 2.0863,
+      "step": 11920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.293570370370371e-05,
+      "loss": 1.9061,
+      "step": 11925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.293274074074074e-05,
+      "loss": 2.0572,
+      "step": 11930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292977777777778e-05,
+      "loss": 2.1648,
+      "step": 11935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292681481481481e-05,
+      "loss": 2.0614,
+      "step": 11940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292385185185186e-05,
+      "loss": 1.9902,
+      "step": 11945
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.292088888888889e-05,
+      "loss": 2.1492,
+      "step": 11950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.291792592592593e-05,
+      "loss": 1.9929,
+      "step": 11955
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.291496296296296e-05,
+      "loss": 2.0434,
+      "step": 11960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2912000000000005e-05,
+      "loss": 2.0688,
+      "step": 11965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.290903703703704e-05,
+      "loss": 2.0072,
+      "step": 11970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2906074074074076e-05,
+      "loss": 2.0436,
+      "step": 11975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.290311111111111e-05,
+      "loss": 2.2308,
+      "step": 11980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2900148148148154e-05,
+      "loss": 2.0936,
+      "step": 11985
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2897185185185186e-05,
+      "loss": 2.1122,
+      "step": 11990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2894222222222225e-05,
+      "loss": 2.13,
+      "step": 11995
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.289125925925926e-05,
+      "loss": 2.1069,
+      "step": 12000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.2888296296296295e-05,
+      "loss": 2.1227,
+      "step": 12005
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2885333333333334e-05,
+      "loss": 2.134,
+      "step": 12010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.288237037037037e-05,
+      "loss": 2.0549,
+      "step": 12015
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2879407407407405e-05,
+      "loss": 2.1679,
+      "step": 12020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2876444444444444e-05,
+      "loss": 2.1453,
+      "step": 12025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.287348148148148e-05,
+      "loss": 1.9335,
+      "step": 12030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.287051851851852e-05,
+      "loss": 1.9928,
+      "step": 12035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.2867555555555554e-05,
+      "loss": 2.0942,
+      "step": 12040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.286459259259259e-05,
+      "loss": 2.0688,
+      "step": 12045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.286162962962963e-05,
+      "loss": 2.0388,
+      "step": 12050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.285866666666667e-05,
+      "loss": 2.1614,
+      "step": 12055
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.28557037037037e-05,
+      "loss": 2.0595,
+      "step": 12060
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.285274074074074e-05,
+      "loss": 2.0144,
+      "step": 12065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.284977777777778e-05,
+      "loss": 2.2383,
+      "step": 12070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.284681481481482e-05,
+      "loss": 1.941,
+      "step": 12075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.284385185185185e-05,
+      "loss": 2.121,
+      "step": 12080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.284088888888889e-05,
+      "loss": 2.0096,
+      "step": 12085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.283792592592593e-05,
+      "loss": 2.0464,
+      "step": 12090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.283496296296297e-05,
+      "loss": 2.0906,
+      "step": 12095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.2832e-05,
+      "loss": 2.0343,
+      "step": 12100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.282903703703704e-05,
+      "loss": 2.0857,
+      "step": 12105
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.282607407407408e-05,
+      "loss": 2.1547,
+      "step": 12110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.2823111111111116e-05,
+      "loss": 2.161,
+      "step": 12115
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.282014814814815e-05,
+      "loss": 2.1047,
+      "step": 12120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2817185185185187e-05,
+      "loss": 2.1955,
+      "step": 12125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2814222222222225e-05,
+      "loss": 2.069,
+      "step": 12130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2811259259259264e-05,
+      "loss": 2.0278,
+      "step": 12135
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2808296296296296e-05,
+      "loss": 1.9195,
+      "step": 12140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2805333333333335e-05,
+      "loss": 1.9381,
+      "step": 12145
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.280237037037037e-05,
+      "loss": 2.1714,
+      "step": 12150
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.279940740740741e-05,
+      "loss": 2.099,
+      "step": 12155
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2796444444444445e-05,
+      "loss": 2.0179,
+      "step": 12160
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2793481481481484e-05,
+      "loss": 2.0875,
+      "step": 12165
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.2790518518518516e-05,
+      "loss": 2.1294,
+      "step": 12170
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.278755555555556e-05,
+      "loss": 2.0409,
+      "step": 12175
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.278459259259259e-05,
+      "loss": 2.0467,
+      "step": 12180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.278162962962963e-05,
+      "loss": 2.0565,
+      "step": 12185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.2778666666666664e-05,
+      "loss": 2.0773,
+      "step": 12190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.277570370370371e-05,
+      "loss": 2.0711,
+      "step": 12195
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.277274074074074e-05,
+      "loss": 2.0719,
+      "step": 12200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.276977777777778e-05,
+      "loss": 1.9875,
+      "step": 12205
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.276681481481481e-05,
+      "loss": 2.0617,
+      "step": 12210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.276385185185186e-05,
+      "loss": 1.9577,
+      "step": 12215
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.276088888888889e-05,
+      "loss": 2.0032,
+      "step": 12220
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.275792592592593e-05,
+      "loss": 2.0531,
+      "step": 12225
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.275496296296296e-05,
+      "loss": 2.0201,
+      "step": 12230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2752e-05,
+      "loss": 2.2203,
+      "step": 12235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.274903703703704e-05,
+      "loss": 2.1142,
+      "step": 12240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.274607407407408e-05,
+      "loss": 2.145,
+      "step": 12245
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.274311111111111e-05,
+      "loss": 2.0741,
+      "step": 12250
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.274014814814815e-05,
+      "loss": 2.1207,
+      "step": 12255
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.273718518518519e-05,
+      "loss": 2.0753,
+      "step": 12260
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2734222222222226e-05,
+      "loss": 2.0867,
+      "step": 12265
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.273125925925926e-05,
+      "loss": 2.1496,
+      "step": 12270
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.27282962962963e-05,
+      "loss": 2.1154,
+      "step": 12275
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2725333333333336e-05,
+      "loss": 2.1221,
+      "step": 12280
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2722370370370375e-05,
+      "loss": 2.0638,
+      "step": 12285
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.271940740740741e-05,
+      "loss": 2.0265,
+      "step": 12290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2716444444444446e-05,
+      "loss": 2.1495,
+      "step": 12295
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2713481481481484e-05,
+      "loss": 1.947,
+      "step": 12300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.271051851851852e-05,
+      "loss": 2.1065,
+      "step": 12305
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2707555555555555e-05,
+      "loss": 2.0491,
+      "step": 12310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2704592592592594e-05,
+      "loss": 2.1002,
+      "step": 12315
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.270162962962963e-05,
+      "loss": 2.1485,
+      "step": 12320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.269866666666667e-05,
+      "loss": 1.931,
+      "step": 12325
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.2695703703703704e-05,
+      "loss": 2.0992,
+      "step": 12330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.269274074074074e-05,
+      "loss": 2.005,
+      "step": 12335
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.268977777777778e-05,
+      "loss": 2.1412,
+      "step": 12340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.268681481481482e-05,
+      "loss": 2.0787,
+      "step": 12345
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.268385185185185e-05,
+      "loss": 2.1672,
+      "step": 12350
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.268088888888889e-05,
+      "loss": 2.0586,
+      "step": 12355
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.267792592592593e-05,
+      "loss": 2.1221,
+      "step": 12360
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.267496296296297e-05,
+      "loss": 2.0877,
+      "step": 12365
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.2672e-05,
+      "loss": 2.1301,
+      "step": 12370
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.266903703703704e-05,
+      "loss": 1.9854,
+      "step": 12375
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.266607407407407e-05,
+      "loss": 1.9394,
+      "step": 12380
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.266311111111112e-05,
+      "loss": 2.1236,
+      "step": 12385
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.266014814814815e-05,
+      "loss": 2.0078,
+      "step": 12390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.265718518518519e-05,
+      "loss": 2.1959,
+      "step": 12395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.265422222222222e-05,
+      "loss": 1.9486,
+      "step": 12400
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2651259259259266e-05,
+      "loss": 2.1441,
+      "step": 12405
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.26482962962963e-05,
+      "loss": 2.1024,
+      "step": 12410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2645333333333337e-05,
+      "loss": 2.0716,
+      "step": 12415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.264237037037037e-05,
+      "loss": 1.9934,
+      "step": 12420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2639407407407414e-05,
+      "loss": 2.0877,
+      "step": 12425
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2636444444444446e-05,
+      "loss": 2.0199,
+      "step": 12430
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2633481481481485e-05,
+      "loss": 2.1148,
+      "step": 12435
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.263051851851852e-05,
+      "loss": 1.9948,
+      "step": 12440
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.262755555555556e-05,
+      "loss": 2.2159,
+      "step": 12445
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2624592592592595e-05,
+      "loss": 2.1202,
+      "step": 12450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.2621629629629634e-05,
+      "loss": 2.1747,
+      "step": 12455
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.2618666666666666e-05,
+      "loss": 2.0513,
+      "step": 12460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.2615703703703704e-05,
+      "loss": 2.0114,
+      "step": 12465
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.261274074074074e-05,
+      "loss": 2.1077,
+      "step": 12470
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.260977777777778e-05,
+      "loss": 2.0394,
+      "step": 12475
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.2606814814814814e-05,
+      "loss": 1.997,
+      "step": 12480
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.260385185185185e-05,
+      "loss": 2.022,
+      "step": 12485
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.260088888888889e-05,
+      "loss": 2.0506,
+      "step": 12490
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.259792592592593e-05,
+      "loss": 2.0418,
+      "step": 12495
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.259496296296296e-05,
+      "loss": 2.1871,
+      "step": 12500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.2592e-05,
+      "loss": 2.1347,
+      "step": 12505
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.258903703703704e-05,
+      "loss": 1.8785,
+      "step": 12510
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.258607407407408e-05,
+      "loss": 2.0166,
+      "step": 12515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.258311111111111e-05,
+      "loss": 2.1569,
+      "step": 12520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.258014814814815e-05,
+      "loss": 2.0294,
+      "step": 12525
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.257718518518519e-05,
+      "loss": 2.0627,
+      "step": 12530
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.257422222222223e-05,
+      "loss": 2.0042,
+      "step": 12535
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.257125925925926e-05,
+      "loss": 2.0411,
+      "step": 12540
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.25682962962963e-05,
+      "loss": 2.1107,
+      "step": 12545
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.256533333333334e-05,
+      "loss": 2.0758,
+      "step": 12550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.2562370370370376e-05,
+      "loss": 2.2247,
+      "step": 12555
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.255940740740741e-05,
+      "loss": 1.9923,
+      "step": 12560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.255644444444445e-05,
+      "loss": 2.0701,
+      "step": 12565
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.2553481481481486e-05,
+      "loss": 2.0196,
+      "step": 12570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2550518518518525e-05,
+      "loss": 2.1428,
+      "step": 12575
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.254755555555556e-05,
+      "loss": 2.0243,
+      "step": 12580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2544592592592596e-05,
+      "loss": 2.1269,
+      "step": 12585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2541629629629634e-05,
+      "loss": 2.0801,
+      "step": 12590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.253866666666667e-05,
+      "loss": 2.0329,
+      "step": 12595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2535703703703705e-05,
+      "loss": 2.1521,
+      "step": 12600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2532740740740744e-05,
+      "loss": 2.0046,
+      "step": 12605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.252977777777778e-05,
+      "loss": 1.9771,
+      "step": 12610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.252681481481482e-05,
+      "loss": 2.0754,
+      "step": 12615
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.2523851851851854e-05,
+      "loss": 2.111,
+      "step": 12620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.252088888888889e-05,
+      "loss": 1.9362,
+      "step": 12625
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.2517925925925925e-05,
+      "loss": 1.9751,
+      "step": 12630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.251496296296297e-05,
+      "loss": 1.8958,
+      "step": 12635
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.2512e-05,
+      "loss": 2.045,
+      "step": 12640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.250903703703704e-05,
+      "loss": 1.9897,
+      "step": 12645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.250607407407407e-05,
+      "loss": 1.9933,
+      "step": 12650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.250311111111112e-05,
+      "loss": 2.1026,
+      "step": 12655
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.250014814814815e-05,
+      "loss": 2.1265,
+      "step": 12660
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.249718518518519e-05,
+      "loss": 1.9803,
+      "step": 12665
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.249422222222222e-05,
+      "loss": 2.1717,
+      "step": 12670
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.249125925925927e-05,
+      "loss": 2.1805,
+      "step": 12675
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.24882962962963e-05,
+      "loss": 2.0657,
+      "step": 12680
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.248533333333334e-05,
+      "loss": 2.0095,
+      "step": 12685
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.248237037037037e-05,
+      "loss": 2.1996,
+      "step": 12690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.247940740740741e-05,
+      "loss": 2.1165,
+      "step": 12695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.247644444444445e-05,
+      "loss": 2.1073,
+      "step": 12700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.247348148148148e-05,
+      "loss": 1.894,
+      "step": 12705
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.247051851851852e-05,
+      "loss": 2.2013,
+      "step": 12710
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.246755555555556e-05,
+      "loss": 2.0435,
+      "step": 12715
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.2464592592592596e-05,
+      "loss": 2.2429,
+      "step": 12720
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.246162962962963e-05,
+      "loss": 2.0785,
+      "step": 12725
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.245866666666667e-05,
+      "loss": 2.0471,
+      "step": 12730
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.2455703703703706e-05,
+      "loss": 1.8792,
+      "step": 12735
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.2452740740740745e-05,
+      "loss": 2.0019,
+      "step": 12740
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.244977777777778e-05,
+      "loss": 2.1038,
+      "step": 12745
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2446814814814816e-05,
+      "loss": 1.9694,
+      "step": 12750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2443851851851855e-05,
+      "loss": 2.0658,
+      "step": 12755
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.244088888888889e-05,
+      "loss": 2.1507,
+      "step": 12760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2437925925925925e-05,
+      "loss": 2.1381,
+      "step": 12765
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2434962962962964e-05,
+      "loss": 2.1264,
+      "step": 12770
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2431999999999996e-05,
+      "loss": 2.0483,
+      "step": 12775
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.242903703703704e-05,
+      "loss": 2.1985,
+      "step": 12780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2426074074074074e-05,
+      "loss": 2.1627,
+      "step": 12785
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.242311111111111e-05,
+      "loss": 2.0331,
+      "step": 12790
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.2420148148148145e-05,
+      "loss": 2.0645,
+      "step": 12795
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.241718518518519e-05,
+      "loss": 2.0922,
+      "step": 12800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.241422222222222e-05,
+      "loss": 1.9673,
+      "step": 12805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.241125925925926e-05,
+      "loss": 1.9481,
+      "step": 12810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.240829629629629e-05,
+      "loss": 2.0728,
+      "step": 12815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.240533333333334e-05,
+      "loss": 2.001,
+      "step": 12820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.240237037037037e-05,
+      "loss": 2.2028,
+      "step": 12825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.239940740740741e-05,
+      "loss": 2.0098,
+      "step": 12830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.239644444444444e-05,
+      "loss": 1.9691,
+      "step": 12835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.239348148148149e-05,
+      "loss": 1.9647,
+      "step": 12840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.239051851851852e-05,
+      "loss": 2.016,
+      "step": 12845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.238755555555556e-05,
+      "loss": 2.0025,
+      "step": 12850
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.238459259259259e-05,
+      "loss": 2.042,
+      "step": 12855
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.238162962962963e-05,
+      "loss": 2.0047,
+      "step": 12860
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.237866666666667e-05,
+      "loss": 2.1752,
+      "step": 12865
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.237570370370371e-05,
+      "loss": 2.0031,
+      "step": 12870
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.237274074074074e-05,
+      "loss": 2.0408,
+      "step": 12875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.236977777777778e-05,
+      "loss": 2.1455,
+      "step": 12880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.2366814814814816e-05,
+      "loss": 2.1129,
+      "step": 12885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.2363851851851855e-05,
+      "loss": 2.1005,
+      "step": 12890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.236088888888889e-05,
+      "loss": 1.9708,
+      "step": 12895
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.2357925925925926e-05,
+      "loss": 2.1866,
+      "step": 12900
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.2354962962962965e-05,
+      "loss": 1.9787,
+      "step": 12905
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2352000000000004e-05,
+      "loss": 2.133,
+      "step": 12910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2349037037037036e-05,
+      "loss": 2.1486,
+      "step": 12915
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2346074074074075e-05,
+      "loss": 2.0984,
+      "step": 12920
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2343111111111113e-05,
+      "loss": 2.0902,
+      "step": 12925
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.234014814814815e-05,
+      "loss": 2.0028,
+      "step": 12930
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.2337185185185184e-05,
+      "loss": 2.0839,
+      "step": 12935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.233422222222222e-05,
+      "loss": 2.0416,
+      "step": 12940
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.233125925925926e-05,
+      "loss": 2.0289,
+      "step": 12945
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.23282962962963e-05,
+      "loss": 2.061,
+      "step": 12950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.232533333333333e-05,
+      "loss": 2.0404,
+      "step": 12955
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.232237037037037e-05,
+      "loss": 2.0707,
+      "step": 12960
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.231940740740741e-05,
+      "loss": 2.1211,
+      "step": 12965
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.231644444444445e-05,
+      "loss": 2.1527,
+      "step": 12970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.231348148148148e-05,
+      "loss": 2.0934,
+      "step": 12975
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.231051851851852e-05,
+      "loss": 2.0904,
+      "step": 12980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.230755555555556e-05,
+      "loss": 1.9881,
+      "step": 12985
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.23045925925926e-05,
+      "loss": 2.1066,
+      "step": 12990
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.230162962962963e-05,
+      "loss": 2.0826,
+      "step": 12995
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.229866666666667e-05,
+      "loss": 2.1241,
+      "step": 13000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.22957037037037e-05,
+      "loss": 2.2448,
+      "step": 13005
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.2292740740740746e-05,
+      "loss": 2.0667,
+      "step": 13010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.228977777777778e-05,
+      "loss": 1.9399,
+      "step": 13015
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.228681481481482e-05,
+      "loss": 2.2072,
+      "step": 13020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.228385185185185e-05,
+      "loss": 2.1444,
+      "step": 13025
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2280888888888895e-05,
+      "loss": 2.0552,
+      "step": 13030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.227792592592593e-05,
+      "loss": 2.0112,
+      "step": 13035
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2274962962962966e-05,
+      "loss": 2.1551,
+      "step": 13040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2272e-05,
+      "loss": 2.1172,
+      "step": 13045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.226903703703704e-05,
+      "loss": 2.1369,
+      "step": 13050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2266074074074075e-05,
+      "loss": 2.0632,
+      "step": 13055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2263111111111114e-05,
+      "loss": 2.2058,
+      "step": 13060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2260148148148146e-05,
+      "loss": 2.1546,
+      "step": 13065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.225718518518519e-05,
+      "loss": 2.1509,
+      "step": 13070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.2254222222222224e-05,
+      "loss": 2.0703,
+      "step": 13075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.225125925925926e-05,
+      "loss": 2.1884,
+      "step": 13080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.2248296296296295e-05,
+      "loss": 2.1389,
+      "step": 13085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.2245333333333334e-05,
+      "loss": 2.124,
+      "step": 13090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.224237037037037e-05,
+      "loss": 2.1039,
+      "step": 13095
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.223940740740741e-05,
+      "loss": 2.0245,
+      "step": 13100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.223644444444444e-05,
+      "loss": 2.1605,
+      "step": 13105
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.223348148148148e-05,
+      "loss": 1.9646,
+      "step": 13110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.223051851851852e-05,
+      "loss": 2.1198,
+      "step": 13115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.222755555555556e-05,
+      "loss": 2.0129,
+      "step": 13120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.222459259259259e-05,
+      "loss": 2.1546,
+      "step": 13125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.222162962962963e-05,
+      "loss": 2.0764,
+      "step": 13130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.221866666666667e-05,
+      "loss": 2.049,
+      "step": 13135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.221570370370371e-05,
+      "loss": 2.1184,
+      "step": 13140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.221274074074074e-05,
+      "loss": 2.0669,
+      "step": 13145
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.220977777777778e-05,
+      "loss": 2.1905,
+      "step": 13150
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.220681481481482e-05,
+      "loss": 2.2068,
+      "step": 13155
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.220385185185186e-05,
+      "loss": 2.1641,
+      "step": 13160
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.220088888888889e-05,
+      "loss": 1.8973,
+      "step": 13165
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.219792592592593e-05,
+      "loss": 1.9805,
+      "step": 13170
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.2194962962962967e-05,
+      "loss": 2.1717,
+      "step": 13175
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.2192000000000005e-05,
+      "loss": 2.0219,
+      "step": 13180
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.218903703703704e-05,
+      "loss": 2.1896,
+      "step": 13185
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.2186074074074076e-05,
+      "loss": 2.1088,
+      "step": 13190
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2183111111111115e-05,
+      "loss": 2.0254,
+      "step": 13195
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2180148148148154e-05,
+      "loss": 2.1033,
+      "step": 13200
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2177185185185186e-05,
+      "loss": 1.9094,
+      "step": 13205
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2174222222222225e-05,
+      "loss": 2.1021,
+      "step": 13210
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2171259259259264e-05,
+      "loss": 2.1423,
+      "step": 13215
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.21682962962963e-05,
+      "loss": 2.0943,
+      "step": 13220
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2165333333333334e-05,
+      "loss": 2.0726,
+      "step": 13225
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.216237037037037e-05,
+      "loss": 2.0218,
+      "step": 13230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.2159407407407405e-05,
+      "loss": 2.215,
+      "step": 13235
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.215644444444445e-05,
+      "loss": 2.0669,
+      "step": 13240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.215348148148148e-05,
+      "loss": 2.1599,
+      "step": 13245
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.215051851851852e-05,
+      "loss": 2.0902,
+      "step": 13250
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2147555555555554e-05,
+      "loss": 1.9893,
+      "step": 13255
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.21445925925926e-05,
+      "loss": 2.0965,
+      "step": 13260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.214162962962963e-05,
+      "loss": 1.9664,
+      "step": 13265
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.213866666666667e-05,
+      "loss": 2.0104,
+      "step": 13270
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.21357037037037e-05,
+      "loss": 2.0935,
+      "step": 13275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.213274074074075e-05,
+      "loss": 2.1998,
+      "step": 13280
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.212977777777778e-05,
+      "loss": 2.1149,
+      "step": 13285
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.212681481481482e-05,
+      "loss": 2.149,
+      "step": 13290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.212385185185185e-05,
+      "loss": 2.0638,
+      "step": 13295
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2120888888888896e-05,
+      "loss": 1.9766,
+      "step": 13300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.211792592592593e-05,
+      "loss": 2.0908,
+      "step": 13305
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.211496296296297e-05,
+      "loss": 2.1838,
+      "step": 13310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2112e-05,
+      "loss": 2.131,
+      "step": 13315
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210903703703704e-05,
+      "loss": 2.1748,
+      "step": 13320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210607407407408e-05,
+      "loss": 2.072,
+      "step": 13325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2103111111111116e-05,
+      "loss": 2.1037,
+      "step": 13330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.210014814814815e-05,
+      "loss": 2.057,
+      "step": 13335
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.209718518518519e-05,
+      "loss": 2.0885,
+      "step": 13340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2094222222222225e-05,
+      "loss": 2.1631,
+      "step": 13345
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2091259259259264e-05,
+      "loss": 2.1436,
+      "step": 13350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.2088296296296296e-05,
+      "loss": 2.1534,
+      "step": 13355
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2085333333333335e-05,
+      "loss": 2.041,
+      "step": 13360
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2082370370370374e-05,
+      "loss": 2.0053,
+      "step": 13365
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.207940740740741e-05,
+      "loss": 1.9723,
+      "step": 13370
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2076444444444445e-05,
+      "loss": 2.1651,
+      "step": 13375
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2073481481481484e-05,
+      "loss": 2.0194,
+      "step": 13380
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.207051851851852e-05,
+      "loss": 1.9925,
+      "step": 13385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.206755555555556e-05,
+      "loss": 2.2107,
+      "step": 13390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.206459259259259e-05,
+      "loss": 2.0535,
+      "step": 13395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.206162962962963e-05,
+      "loss": 2.163,
+      "step": 13400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.205866666666667e-05,
+      "loss": 2.1978,
+      "step": 13405
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.205570370370371e-05,
+      "loss": 2.0739,
+      "step": 13410
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.205274074074074e-05,
+      "loss": 2.0485,
+      "step": 13415
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204977777777778e-05,
+      "loss": 2.1549,
+      "step": 13420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204681481481482e-05,
+      "loss": 2.1377,
+      "step": 13425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204385185185186e-05,
+      "loss": 2.0472,
+      "step": 13430
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.204088888888889e-05,
+      "loss": 2.1706,
+      "step": 13435
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.203792592592593e-05,
+      "loss": 2.0735,
+      "step": 13440
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.203496296296297e-05,
+      "loss": 2.0858,
+      "step": 13445
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.203200000000001e-05,
+      "loss": 2.0049,
+      "step": 13450
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202903703703704e-05,
+      "loss": 2.0622,
+      "step": 13455
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202607407407408e-05,
+      "loss": 2.0394,
+      "step": 13460
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202311111111111e-05,
+      "loss": 2.091,
+      "step": 13465
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2020148148148155e-05,
+      "loss": 2.0626,
+      "step": 13470
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.201718518518519e-05,
+      "loss": 2.0693,
+      "step": 13475
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.201422222222222e-05,
+      "loss": 2.1128,
+      "step": 13480
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.201125925925926e-05,
+      "loss": 2.1237,
+      "step": 13485
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.20082962962963e-05,
+      "loss": 2.1342,
+      "step": 13490
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.2005333333333336e-05,
+      "loss": 2.0532,
+      "step": 13495
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.200237037037037e-05,
+      "loss": 2.0842,
+      "step": 13500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.199940740740741e-05,
+      "loss": 2.0855,
+      "step": 13505
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1996444444444446e-05,
+      "loss": 2.1203,
+      "step": 13510
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1993481481481484e-05,
+      "loss": 2.0449,
+      "step": 13515
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1990518518518516e-05,
+      "loss": 2.1576,
+      "step": 13520
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1987555555555555e-05,
+      "loss": 2.0758,
+      "step": 13525
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1984592592592594e-05,
+      "loss": 2.1485,
+      "step": 13530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.198162962962963e-05,
+      "loss": 2.0653,
+      "step": 13535
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1978666666666665e-05,
+      "loss": 2.0219,
+      "step": 13540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1975703703703704e-05,
+      "loss": 2.2063,
+      "step": 13545
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.197274074074074e-05,
+      "loss": 2.1276,
+      "step": 13550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.196977777777778e-05,
+      "loss": 2.0958,
+      "step": 13555
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.1966814814814814e-05,
+      "loss": 2.2182,
+      "step": 13560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.196385185185185e-05,
+      "loss": 2.0937,
+      "step": 13565
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.196088888888889e-05,
+      "loss": 1.9599,
+      "step": 13570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.195792592592593e-05,
+      "loss": 2.0422,
+      "step": 13575
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.195496296296296e-05,
+      "loss": 2.0227,
+      "step": 13580
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.1952e-05,
+      "loss": 1.9983,
+      "step": 13585
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.194903703703704e-05,
+      "loss": 2.1914,
+      "step": 13590
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.194607407407408e-05,
+      "loss": 2.0946,
+      "step": 13595
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.194311111111111e-05,
+      "loss": 2.0182,
+      "step": 13600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.194014814814815e-05,
+      "loss": 2.0037,
+      "step": 13605
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.193718518518519e-05,
+      "loss": 2.0642,
+      "step": 13610
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.193422222222223e-05,
+      "loss": 2.1569,
+      "step": 13615
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.193125925925926e-05,
+      "loss": 1.9669,
+      "step": 13620
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.19282962962963e-05,
+      "loss": 2.0406,
+      "step": 13625
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.192533333333333e-05,
+      "loss": 2.3094,
+      "step": 13630
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.1922370370370376e-05,
+      "loss": 2.1298,
+      "step": 13635
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.191940740740741e-05,
+      "loss": 2.1142,
+      "step": 13640
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1916444444444446e-05,
+      "loss": 1.9428,
+      "step": 13645
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.191348148148148e-05,
+      "loss": 2.0835,
+      "step": 13650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1910518518518524e-05,
+      "loss": 2.0534,
+      "step": 13655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1907555555555556e-05,
+      "loss": 2.0673,
+      "step": 13660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1904592592592595e-05,
+      "loss": 2.1224,
+      "step": 13665
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.190162962962963e-05,
+      "loss": 2.1028,
+      "step": 13670
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.189866666666667e-05,
+      "loss": 2.173,
+      "step": 13675
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1895703703703705e-05,
+      "loss": 2.1125,
+      "step": 13680
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1892740740740743e-05,
+      "loss": 2.0076,
+      "step": 13685
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1889777777777775e-05,
+      "loss": 2.2083,
+      "step": 13690
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.1886814814814814e-05,
+      "loss": 1.9489,
+      "step": 13695
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.188385185185185e-05,
+      "loss": 2.1595,
+      "step": 13700
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.188088888888889e-05,
+      "loss": 2.0408,
+      "step": 13705
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1877925925925924e-05,
+      "loss": 2.0923,
+      "step": 13710
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.187496296296296e-05,
+      "loss": 2.2375,
+      "step": 13715
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.1872e-05,
+      "loss": 1.9604,
+      "step": 13720
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.186903703703704e-05,
+      "loss": 1.9845,
+      "step": 13725
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.186607407407407e-05,
+      "loss": 2.0058,
+      "step": 13730
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.186311111111111e-05,
+      "loss": 2.075,
+      "step": 13735
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.186014814814815e-05,
+      "loss": 2.0699,
+      "step": 13740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.185718518518519e-05,
+      "loss": 2.0277,
+      "step": 13745
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 4.185422222222222e-05,
+      "loss": 2.2192,
+      "step": 13750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.185125925925926e-05,
+      "loss": 2.0455,
+      "step": 13755
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.18482962962963e-05,
+      "loss": 2.069,
+      "step": 13760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.184533333333334e-05,
+      "loss": 2.0551,
+      "step": 13765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.184237037037037e-05,
+      "loss": 2.1023,
+      "step": 13770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.183940740740741e-05,
+      "loss": 2.1192,
+      "step": 13775
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.183644444444445e-05,
+      "loss": 2.1519,
+      "step": 13780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.1833481481481486e-05,
+      "loss": 2.1216,
+      "step": 13785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.183051851851852e-05,
+      "loss": 2.0152,
+      "step": 13790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.182755555555556e-05,
+      "loss": 2.2032,
+      "step": 13795
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.1824592592592596e-05,
+      "loss": 2.208,
+      "step": 13800
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.1821629629629634e-05,
+      "loss": 2.0739,
+      "step": 13805
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1818666666666667e-05,
+      "loss": 2.0978,
+      "step": 13810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1815703703703705e-05,
+      "loss": 2.1084,
+      "step": 13815
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1812740740740744e-05,
+      "loss": 1.9446,
+      "step": 13820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.180977777777778e-05,
+      "loss": 2.1051,
+      "step": 13825
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1806814814814815e-05,
+      "loss": 2.0322,
+      "step": 13830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1803851851851854e-05,
+      "loss": 2.0215,
+      "step": 13835
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.180088888888889e-05,
+      "loss": 2.1087,
+      "step": 13840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.179792592592593e-05,
+      "loss": 2.1563,
+      "step": 13845
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1794962962962964e-05,
+      "loss": 2.0637,
+      "step": 13850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1792e-05,
+      "loss": 2.1086,
+      "step": 13855
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.1789037037037034e-05,
+      "loss": 2.0967,
+      "step": 13860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 4.178607407407408e-05,
+      "loss": 2.1491,
+      "step": 13865
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.178311111111111e-05,
+      "loss": 2.2069,
+      "step": 13870
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.178014814814815e-05,
+      "loss": 2.1009,
+      "step": 13875
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.177718518518518e-05,
+      "loss": 2.0797,
+      "step": 13880
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.177422222222223e-05,
+      "loss": 2.0598,
+      "step": 13885
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.177125925925926e-05,
+      "loss": 2.0526,
+      "step": 13890
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.17682962962963e-05,
+      "loss": 2.1235,
+      "step": 13895
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.176533333333333e-05,
+      "loss": 2.2266,
+      "step": 13900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.176237037037038e-05,
+      "loss": 2.1926,
+      "step": 13905
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.175940740740741e-05,
+      "loss": 2.1444,
+      "step": 13910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.175644444444445e-05,
+      "loss": 2.0799,
+      "step": 13915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.175348148148148e-05,
+      "loss": 2.212,
+      "step": 13920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.175051851851852e-05,
+      "loss": 2.1169,
+      "step": 13925
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.174755555555556e-05,
+      "loss": 2.0192,
+      "step": 13930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1744592592592596e-05,
+      "loss": 2.0212,
+      "step": 13935
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.174162962962963e-05,
+      "loss": 2.0407,
+      "step": 13940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.173866666666667e-05,
+      "loss": 1.9824,
+      "step": 13945
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1735703703703706e-05,
+      "loss": 2.0493,
+      "step": 13950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1732740740740745e-05,
+      "loss": 2.1002,
+      "step": 13955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.172977777777778e-05,
+      "loss": 2.009,
+      "step": 13960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1726814814814816e-05,
+      "loss": 2.0869,
+      "step": 13965
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1723851851851855e-05,
+      "loss": 2.2347,
+      "step": 13970
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 4.1720888888888893e-05,
+      "loss": 1.9343,
+      "step": 13975
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1717925925925925e-05,
+      "loss": 2.0489,
+      "step": 13980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1714962962962964e-05,
+      "loss": 2.2685,
+      "step": 13985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1712e-05,
+      "loss": 2.0465,
+      "step": 13990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.170903703703704e-05,
+      "loss": 2.0893,
+      "step": 13995
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.1706074074074074e-05,
+      "loss": 2.1137,
+      "step": 14000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.170311111111111e-05,
+      "loss": 2.053,
+      "step": 14005
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.170014814814815e-05,
+      "loss": 2.2186,
+      "step": 14010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.169718518518519e-05,
+      "loss": 2.0172,
+      "step": 14015
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.169422222222222e-05,
+      "loss": 2.0425,
+      "step": 14020
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.169125925925926e-05,
+      "loss": 2.0626,
+      "step": 14025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.16882962962963e-05,
+      "loss": 1.9898,
+      "step": 14030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.168533333333334e-05,
+      "loss": 2.1142,
+      "step": 14035
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.168237037037037e-05,
+      "loss": 2.0087,
+      "step": 14040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.167940740740741e-05,
+      "loss": 1.8753,
+      "step": 14045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.167644444444445e-05,
+      "loss": 2.0239,
+      "step": 14050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.167348148148149e-05,
+      "loss": 2.2092,
+      "step": 14055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.167051851851852e-05,
+      "loss": 2.0851,
+      "step": 14060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.166755555555556e-05,
+      "loss": 2.0621,
+      "step": 14065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.16645925925926e-05,
+      "loss": 1.9728,
+      "step": 14070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.1661629629629636e-05,
+      "loss": 2.0479,
+      "step": 14075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.165866666666667e-05,
+      "loss": 2.0899,
+      "step": 14080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.165570370370371e-05,
+      "loss": 2.1002,
+      "step": 14085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.165274074074074e-05,
+      "loss": 2.2045,
+      "step": 14090
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1649777777777785e-05,
+      "loss": 2.0023,
+      "step": 14095
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1646814814814817e-05,
+      "loss": 2.0798,
+      "step": 14100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1643851851851855e-05,
+      "loss": 2.1715,
+      "step": 14105
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.164088888888889e-05,
+      "loss": 2.1451,
+      "step": 14110
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.163792592592593e-05,
+      "loss": 2.1356,
+      "step": 14115
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1634962962962965e-05,
+      "loss": 1.9964,
+      "step": 14120
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1632000000000004e-05,
+      "loss": 2.1082,
+      "step": 14125
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1629037037037036e-05,
+      "loss": 2.1252,
+      "step": 14130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.162607407407408e-05,
+      "loss": 2.1712,
+      "step": 14135
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.1623111111111114e-05,
+      "loss": 2.1854,
+      "step": 14140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.162014814814815e-05,
+      "loss": 2.0532,
+      "step": 14145
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.1617185185185184e-05,
+      "loss": 2.1789,
+      "step": 14150
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.161422222222222e-05,
+      "loss": 2.0444,
+      "step": 14155
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.161125925925926e-05,
+      "loss": 1.981,
+      "step": 14160
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.16082962962963e-05,
+      "loss": 2.1252,
+      "step": 14165
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.160533333333333e-05,
+      "loss": 1.9818,
+      "step": 14170
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.160237037037037e-05,
+      "loss": 2.0711,
+      "step": 14175
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.159940740740741e-05,
+      "loss": 2.0776,
+      "step": 14180
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.159644444444445e-05,
+      "loss": 2.158,
+      "step": 14185
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.159348148148148e-05,
+      "loss": 2.1075,
+      "step": 14190
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.159051851851852e-05,
+      "loss": 2.0279,
+      "step": 14195
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.158755555555556e-05,
+      "loss": 1.9855,
+      "step": 14200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.15845925925926e-05,
+      "loss": 1.9874,
+      "step": 14205
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.158162962962963e-05,
+      "loss": 2.1113,
+      "step": 14210
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.157866666666667e-05,
+      "loss": 2.0573,
+      "step": 14215
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.157570370370371e-05,
+      "loss": 2.0298,
+      "step": 14220
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.1572740740740746e-05,
+      "loss": 2.1508,
+      "step": 14225
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.156977777777778e-05,
+      "loss": 2.182,
+      "step": 14230
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.156681481481482e-05,
+      "loss": 2.085,
+      "step": 14235
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.1563851851851856e-05,
+      "loss": 2.144,
+      "step": 14240
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.1560888888888895e-05,
+      "loss": 2.1203,
+      "step": 14245
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.155792592592593e-05,
+      "loss": 2.0401,
+      "step": 14250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.155496296296296e-05,
+      "loss": 2.0923,
+      "step": 14255
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1552000000000005e-05,
+      "loss": 2.1821,
+      "step": 14260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.154903703703704e-05,
+      "loss": 2.0245,
+      "step": 14265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1546074074074076e-05,
+      "loss": 2.0459,
+      "step": 14270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.154311111111111e-05,
+      "loss": 2.0527,
+      "step": 14275
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.154014814814815e-05,
+      "loss": 2.0416,
+      "step": 14280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1537185185185185e-05,
+      "loss": 1.9352,
+      "step": 14285
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1534222222222224e-05,
+      "loss": 2.0967,
+      "step": 14290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1531259259259256e-05,
+      "loss": 2.1743,
+      "step": 14295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.15282962962963e-05,
+      "loss": 2.0947,
+      "step": 14300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1525333333333334e-05,
+      "loss": 2.2204,
+      "step": 14305
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.152237037037037e-05,
+      "loss": 2.0692,
+      "step": 14310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.1519407407407405e-05,
+      "loss": 2.002,
+      "step": 14315
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.1516444444444443e-05,
+      "loss": 2.1378,
+      "step": 14320
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.151348148148148e-05,
+      "loss": 2.0584,
+      "step": 14325
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.151051851851852e-05,
+      "loss": 2.0266,
+      "step": 14330
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.150755555555555e-05,
+      "loss": 2.2201,
+      "step": 14335
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.150459259259259e-05,
+      "loss": 2.0901,
+      "step": 14340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.150162962962963e-05,
+      "loss": 2.085,
+      "step": 14345
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.149866666666667e-05,
+      "loss": 2.2857,
+      "step": 14350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.14957037037037e-05,
+      "loss": 2.0769,
+      "step": 14355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.149274074074074e-05,
+      "loss": 2.1179,
+      "step": 14360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.148977777777778e-05,
+      "loss": 2.1101,
+      "step": 14365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.148681481481482e-05,
+      "loss": 2.1943,
+      "step": 14370
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.148385185185185e-05,
+      "loss": 2.1204,
+      "step": 14375
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.148088888888889e-05,
+      "loss": 1.9206,
+      "step": 14380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.147792592592593e-05,
+      "loss": 2.1541,
+      "step": 14385
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.147496296296297e-05,
+      "loss": 2.0315,
+      "step": 14390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1472e-05,
+      "loss": 1.9857,
+      "step": 14395
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.146903703703704e-05,
+      "loss": 2.1676,
+      "step": 14400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1466074074074076e-05,
+      "loss": 2.032,
+      "step": 14405
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1463111111111115e-05,
+      "loss": 2.1304,
+      "step": 14410
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.146014814814815e-05,
+      "loss": 2.111,
+      "step": 14415
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1457185185185186e-05,
+      "loss": 1.9829,
+      "step": 14420
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.1454222222222225e-05,
+      "loss": 2.045,
+      "step": 14425
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1451259259259264e-05,
+      "loss": 2.0173,
+      "step": 14430
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1448296296296296e-05,
+      "loss": 2.0057,
+      "step": 14435
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1445333333333335e-05,
+      "loss": 2.1537,
+      "step": 14440
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.144237037037037e-05,
+      "loss": 2.0826,
+      "step": 14445
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.143940740740741e-05,
+      "loss": 2.1167,
+      "step": 14450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.1436444444444444e-05,
+      "loss": 2.1742,
+      "step": 14455
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.143348148148148e-05,
+      "loss": 2.1215,
+      "step": 14460
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.143051851851852e-05,
+      "loss": 2.1289,
+      "step": 14465
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.142755555555556e-05,
+      "loss": 2.126,
+      "step": 14470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.142459259259259e-05,
+      "loss": 2.0364,
+      "step": 14475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.142162962962963e-05,
+      "loss": 1.9591,
+      "step": 14480
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1418666666666664e-05,
+      "loss": 2.1057,
+      "step": 14485
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.141570370370371e-05,
+      "loss": 2.02,
+      "step": 14490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.141274074074074e-05,
+      "loss": 2.0651,
+      "step": 14495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.140977777777778e-05,
+      "loss": 2.0719,
+      "step": 14500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.140681481481481e-05,
+      "loss": 2.1098,
+      "step": 14505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.140385185185186e-05,
+      "loss": 2.1732,
+      "step": 14510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.140088888888889e-05,
+      "loss": 1.8967,
+      "step": 14515
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.139792592592593e-05,
+      "loss": 2.1395,
+      "step": 14520
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.139496296296296e-05,
+      "loss": 2.094,
+      "step": 14525
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.1392000000000006e-05,
+      "loss": 2.0397,
+      "step": 14530
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.138903703703704e-05,
+      "loss": 2.1939,
+      "step": 14535
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.138607407407408e-05,
+      "loss": 2.1498,
+      "step": 14540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.138311111111111e-05,
+      "loss": 2.125,
+      "step": 14545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.138014814814815e-05,
+      "loss": 2.0582,
+      "step": 14550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.137718518518519e-05,
+      "loss": 2.0127,
+      "step": 14555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1374222222222226e-05,
+      "loss": 2.0991,
+      "step": 14560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.137125925925926e-05,
+      "loss": 2.0907,
+      "step": 14565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1368296296296296e-05,
+      "loss": 2.0903,
+      "step": 14570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1365333333333335e-05,
+      "loss": 2.0422,
+      "step": 14575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1362370370370374e-05,
+      "loss": 2.0719,
+      "step": 14580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1359407407407406e-05,
+      "loss": 2.0992,
+      "step": 14585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1356444444444445e-05,
+      "loss": 2.1733,
+      "step": 14590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1353481481481484e-05,
+      "loss": 2.0139,
+      "step": 14595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.135051851851852e-05,
+      "loss": 2.0392,
+      "step": 14600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1347555555555555e-05,
+      "loss": 2.0644,
+      "step": 14605
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.1344592592592593e-05,
+      "loss": 2.0225,
+      "step": 14610
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.134162962962963e-05,
+      "loss": 1.9885,
+      "step": 14615
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.133866666666667e-05,
+      "loss": 2.0296,
+      "step": 14620
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.13357037037037e-05,
+      "loss": 1.9292,
+      "step": 14625
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.133274074074074e-05,
+      "loss": 2.0361,
+      "step": 14630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.132977777777778e-05,
+      "loss": 1.9927,
+      "step": 14635
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.132681481481482e-05,
+      "loss": 2.1377,
+      "step": 14640
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.132385185185185e-05,
+      "loss": 1.9959,
+      "step": 14645
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.132088888888889e-05,
+      "loss": 2.1521,
+      "step": 14650
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.131792592592593e-05,
+      "loss": 2.106,
+      "step": 14655
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.131496296296297e-05,
+      "loss": 2.0922,
+      "step": 14660
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.1312e-05,
+      "loss": 2.1658,
+      "step": 14665
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.130903703703704e-05,
+      "loss": 2.1335,
+      "step": 14670
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.130607407407408e-05,
+      "loss": 2.04,
+      "step": 14675
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.130311111111112e-05,
+      "loss": 2.1213,
+      "step": 14680
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.130014814814815e-05,
+      "loss": 2.078,
+      "step": 14685
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.129718518518519e-05,
+      "loss": 2.1521,
+      "step": 14690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.1294222222222226e-05,
+      "loss": 2.1108,
+      "step": 14695
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.1291259259259265e-05,
+      "loss": 1.9757,
+      "step": 14700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.12882962962963e-05,
+      "loss": 2.1106,
+      "step": 14705
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1285333333333336e-05,
+      "loss": 2.1667,
+      "step": 14710
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.128237037037037e-05,
+      "loss": 2.0904,
+      "step": 14715
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1279407407407414e-05,
+      "loss": 1.9551,
+      "step": 14720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1276444444444446e-05,
+      "loss": 2.0572,
+      "step": 14725
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1273481481481485e-05,
+      "loss": 2.1257,
+      "step": 14730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1270518518518517e-05,
+      "loss": 2.0819,
+      "step": 14735
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.126755555555556e-05,
+      "loss": 2.0816,
+      "step": 14740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1265185185185185e-05,
+      "loss": 2.0429,
+      "step": 14745
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1262222222222224e-05,
+      "loss": 2.0619,
+      "step": 14750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1259259259259256e-05,
+      "loss": 2.0228,
+      "step": 14755
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.12562962962963e-05,
+      "loss": 2.1033,
+      "step": 14760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.1253333333333334e-05,
+      "loss": 2.1752,
+      "step": 14765
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.125037037037037e-05,
+      "loss": 2.1219,
+      "step": 14770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1247407407407405e-05,
+      "loss": 1.9767,
+      "step": 14775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.124444444444444e-05,
+      "loss": 2.1594,
+      "step": 14780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.124148148148148e-05,
+      "loss": 1.9246,
+      "step": 14785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.123851851851852e-05,
+      "loss": 2.097,
+      "step": 14790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.123555555555555e-05,
+      "loss": 1.9906,
+      "step": 14795
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.123259259259259e-05,
+      "loss": 2.0414,
+      "step": 14800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.122962962962963e-05,
+      "loss": 2.1419,
+      "step": 14805
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.122666666666667e-05,
+      "loss": 1.9506,
+      "step": 14810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.12237037037037e-05,
+      "loss": 2.0683,
+      "step": 14815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.122074074074074e-05,
+      "loss": 2.2329,
+      "step": 14820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.121777777777778e-05,
+      "loss": 2.1431,
+      "step": 14825
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.121481481481482e-05,
+      "loss": 2.121,
+      "step": 14830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.121185185185185e-05,
+      "loss": 2.0335,
+      "step": 14835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.120888888888889e-05,
+      "loss": 1.9796,
+      "step": 14840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.120592592592593e-05,
+      "loss": 2.0391,
+      "step": 14845
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1202962962962967e-05,
+      "loss": 2.1091,
+      "step": 14850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.12e-05,
+      "loss": 2.0152,
+      "step": 14855
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.119703703703704e-05,
+      "loss": 1.9969,
+      "step": 14860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1194074074074076e-05,
+      "loss": 2.1258,
+      "step": 14865
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.1191111111111115e-05,
+      "loss": 1.9615,
+      "step": 14870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.118814814814815e-05,
+      "loss": 2.0152,
+      "step": 14875
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1185185185185186e-05,
+      "loss": 2.017,
+      "step": 14880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1182814814814816e-05,
+      "loss": 2.0579,
+      "step": 14885
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1179851851851854e-05,
+      "loss": 2.1387,
+      "step": 14890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.117688888888889e-05,
+      "loss": 2.2143,
+      "step": 14895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1173925925925925e-05,
+      "loss": 2.035,
+      "step": 14900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1170962962962964e-05,
+      "loss": 2.1267,
+      "step": 14905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1168e-05,
+      "loss": 2.2027,
+      "step": 14910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.116503703703704e-05,
+      "loss": 2.1033,
+      "step": 14915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.1162074074074074e-05,
+      "loss": 2.2719,
+      "step": 14920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.115911111111111e-05,
+      "loss": 2.0827,
+      "step": 14925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.115614814814815e-05,
+      "loss": 2.1438,
+      "step": 14930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.115318518518519e-05,
+      "loss": 2.0304,
+      "step": 14935
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.115022222222222e-05,
+      "loss": 2.0755,
+      "step": 14940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.114725925925926e-05,
+      "loss": 2.123,
+      "step": 14945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.11442962962963e-05,
+      "loss": 2.1398,
+      "step": 14950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.114133333333334e-05,
+      "loss": 2.0213,
+      "step": 14955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.113837037037037e-05,
+      "loss": 2.0681,
+      "step": 14960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.113540740740741e-05,
+      "loss": 2.0369,
+      "step": 14965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.113244444444445e-05,
+      "loss": 2.0586,
+      "step": 14970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.112948148148149e-05,
+      "loss": 2.0624,
+      "step": 14975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.112651851851852e-05,
+      "loss": 2.1034,
+      "step": 14980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.112355555555556e-05,
+      "loss": 2.078,
+      "step": 14985
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.11205925925926e-05,
+      "loss": 2.0662,
+      "step": 14990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1117629629629636e-05,
+      "loss": 2.0414,
+      "step": 14995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.111466666666667e-05,
+      "loss": 2.0381,
+      "step": 15000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.111170370370371e-05,
+      "loss": 2.1513,
+      "step": 15005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1108740740740745e-05,
+      "loss": 1.9433,
+      "step": 15010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1105777777777784e-05,
+      "loss": 2.0806,
+      "step": 15015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1102814814814816e-05,
+      "loss": 2.0476,
+      "step": 15020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1099851851851855e-05,
+      "loss": 2.0201,
+      "step": 15025
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.109688888888889e-05,
+      "loss": 2.1419,
+      "step": 15030
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.109392592592593e-05,
+      "loss": 1.9521,
+      "step": 15035
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1090962962962965e-05,
+      "loss": 1.9679,
+      "step": 15040
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.1088000000000004e-05,
+      "loss": 2.1852,
+      "step": 15045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1085037037037036e-05,
+      "loss": 2.0972,
+      "step": 15050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.108207407407408e-05,
+      "loss": 2.0936,
+      "step": 15055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.107911111111111e-05,
+      "loss": 1.9501,
+      "step": 15060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.107614814814815e-05,
+      "loss": 2.1829,
+      "step": 15065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.1073185185185184e-05,
+      "loss": 1.9812,
+      "step": 15070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.107022222222223e-05,
+      "loss": 2.163,
+      "step": 15075
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.106725925925926e-05,
+      "loss": 1.9699,
+      "step": 15080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.10642962962963e-05,
+      "loss": 1.9943,
+      "step": 15085
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.106133333333333e-05,
+      "loss": 2.0932,
+      "step": 15090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.105837037037037e-05,
+      "loss": 2.0264,
+      "step": 15095
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.105540740740741e-05,
+      "loss": 2.0437,
+      "step": 15100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.105244444444445e-05,
+      "loss": 2.0824,
+      "step": 15105
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.104948148148148e-05,
+      "loss": 2.1427,
+      "step": 15110
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.104651851851852e-05,
+      "loss": 1.9665,
+      "step": 15115
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.104355555555556e-05,
+      "loss": 2.1389,
+      "step": 15120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.10405925925926e-05,
+      "loss": 2.2224,
+      "step": 15125
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.103762962962963e-05,
+      "loss": 2.0985,
+      "step": 15130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.103466666666667e-05,
+      "loss": 1.9811,
+      "step": 15135
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.103170370370371e-05,
+      "loss": 2.1304,
+      "step": 15140
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.1028740740740746e-05,
+      "loss": 2.0762,
+      "step": 15145
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.102577777777778e-05,
+      "loss": 2.1825,
+      "step": 15150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.102281481481482e-05,
+      "loss": 2.0223,
+      "step": 15155
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1019851851851856e-05,
+      "loss": 2.0827,
+      "step": 15160
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1016888888888895e-05,
+      "loss": 2.0286,
+      "step": 15165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.101392592592593e-05,
+      "loss": 2.0494,
+      "step": 15170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1010962962962966e-05,
+      "loss": 2.1453,
+      "step": 15175
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1008000000000004e-05,
+      "loss": 2.179,
+      "step": 15180
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1005037037037037e-05,
+      "loss": 1.9324,
+      "step": 15185
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.1002074074074075e-05,
+      "loss": 2.1469,
+      "step": 15190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.099911111111111e-05,
+      "loss": 2.1273,
+      "step": 15195
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.099614814814815e-05,
+      "loss": 2.0263,
+      "step": 15200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.0993185185185185e-05,
+      "loss": 2.0346,
+      "step": 15205
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.0990222222222224e-05,
+      "loss": 2.1658,
+      "step": 15210
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.0987259259259256e-05,
+      "loss": 2.1182,
+      "step": 15215
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.09842962962963e-05,
+      "loss": 2.0952,
+      "step": 15220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.0981333333333334e-05,
+      "loss": 2.0503,
+      "step": 15225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.097837037037037e-05,
+      "loss": 2.2561,
+      "step": 15230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.0975407407407404e-05,
+      "loss": 2.0875,
+      "step": 15235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.097244444444445e-05,
+      "loss": 2.0434,
+      "step": 15240
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.096948148148148e-05,
+      "loss": 2.1228,
+      "step": 15245
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.096651851851852e-05,
+      "loss": 2.0239,
+      "step": 15250
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.096355555555555e-05,
+      "loss": 2.23,
+      "step": 15255
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.096059259259259e-05,
+      "loss": 2.1564,
+      "step": 15260
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.095762962962963e-05,
+      "loss": 1.9969,
+      "step": 15265
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.095466666666667e-05,
+      "loss": 2.1184,
+      "step": 15270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.09517037037037e-05,
+      "loss": 1.9873,
+      "step": 15275
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.094874074074074e-05,
+      "loss": 2.07,
+      "step": 15280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.094577777777778e-05,
+      "loss": 2.1316,
+      "step": 15285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.094281481481482e-05,
+      "loss": 1.9668,
+      "step": 15290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.093985185185185e-05,
+      "loss": 2.0465,
+      "step": 15295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.093688888888889e-05,
+      "loss": 2.1029,
+      "step": 15300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.093392592592593e-05,
+      "loss": 2.1664,
+      "step": 15305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.0930962962962966e-05,
+      "loss": 2.0025,
+      "step": 15310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.0928e-05,
+      "loss": 2.0515,
+      "step": 15315
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.092503703703704e-05,
+      "loss": 2.1744,
+      "step": 15320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.0922074074074076e-05,
+      "loss": 2.0385,
+      "step": 15325
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0919111111111115e-05,
+      "loss": 2.1291,
+      "step": 15330
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.091614814814815e-05,
+      "loss": 2.0378,
+      "step": 15335
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0913185185185186e-05,
+      "loss": 2.2602,
+      "step": 15340
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0910222222222225e-05,
+      "loss": 2.1103,
+      "step": 15345
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0907259259259263e-05,
+      "loss": 2.027,
+      "step": 15350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0904296296296295e-05,
+      "loss": 2.1953,
+      "step": 15355
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0901333333333334e-05,
+      "loss": 2.042,
+      "step": 15360
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.089837037037037e-05,
+      "loss": 2.0208,
+      "step": 15365
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.089540740740741e-05,
+      "loss": 2.1944,
+      "step": 15370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0892444444444444e-05,
+      "loss": 2.14,
+      "step": 15375
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.088948148148148e-05,
+      "loss": 2.1005,
+      "step": 15380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.088651851851852e-05,
+      "loss": 2.1982,
+      "step": 15385
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.088355555555556e-05,
+      "loss": 2.1287,
+      "step": 15390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.088059259259259e-05,
+      "loss": 2.1458,
+      "step": 15395
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.087762962962963e-05,
+      "loss": 2.0599,
+      "step": 15400
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.087466666666666e-05,
+      "loss": 2.0354,
+      "step": 15405
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.087170370370371e-05,
+      "loss": 1.9656,
+      "step": 15410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.086874074074074e-05,
+      "loss": 2.1182,
+      "step": 15415
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.086577777777778e-05,
+      "loss": 2.0102,
+      "step": 15420
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.086281481481481e-05,
+      "loss": 2.2395,
+      "step": 15425
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.085985185185186e-05,
+      "loss": 1.9445,
+      "step": 15430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.085688888888889e-05,
+      "loss": 2.1119,
+      "step": 15435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.085392592592593e-05,
+      "loss": 2.1529,
+      "step": 15440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.085096296296296e-05,
+      "loss": 2.0068,
+      "step": 15445
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0848000000000006e-05,
+      "loss": 2.211,
+      "step": 15450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.084503703703704e-05,
+      "loss": 2.0549,
+      "step": 15455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.084207407407408e-05,
+      "loss": 2.1113,
+      "step": 15460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.083911111111111e-05,
+      "loss": 2.0522,
+      "step": 15465
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0836148148148155e-05,
+      "loss": 2.0833,
+      "step": 15470
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0833185185185187e-05,
+      "loss": 2.185,
+      "step": 15475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0830222222222225e-05,
+      "loss": 2.0761,
+      "step": 15480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.082725925925926e-05,
+      "loss": 2.0456,
+      "step": 15485
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0824296296296296e-05,
+      "loss": 2.0464,
+      "step": 15490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.0821333333333335e-05,
+      "loss": 2.0996,
+      "step": 15495
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0818370370370374e-05,
+      "loss": 1.9248,
+      "step": 15500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0815407407407406e-05,
+      "loss": 1.8979,
+      "step": 15505
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0812444444444445e-05,
+      "loss": 2.0678,
+      "step": 15510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0809481481481484e-05,
+      "loss": 2.2282,
+      "step": 15515
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.080651851851852e-05,
+      "loss": 2.1183,
+      "step": 15520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.0803555555555554e-05,
+      "loss": 1.929,
+      "step": 15525
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.080059259259259e-05,
+      "loss": 2.119,
+      "step": 15530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.079822222222223e-05,
+      "loss": 2.1638,
+      "step": 15535
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.079525925925926e-05,
+      "loss": 2.1155,
+      "step": 15540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.07922962962963e-05,
+      "loss": 2.0963,
+      "step": 15545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.078933333333333e-05,
+      "loss": 2.1174,
+      "step": 15550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.078637037037037e-05,
+      "loss": 2.2391,
+      "step": 15555
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.078340740740741e-05,
+      "loss": 2.0453,
+      "step": 15560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.078044444444445e-05,
+      "loss": 1.9399,
+      "step": 15565
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.077748148148148e-05,
+      "loss": 2.0565,
+      "step": 15570
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.077451851851852e-05,
+      "loss": 2.0234,
+      "step": 15575
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.077155555555556e-05,
+      "loss": 2.1458,
+      "step": 15580
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.07685925925926e-05,
+      "loss": 2.0724,
+      "step": 15585
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.076562962962963e-05,
+      "loss": 2.0227,
+      "step": 15590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.076266666666667e-05,
+      "loss": 2.1013,
+      "step": 15595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.075970370370371e-05,
+      "loss": 2.0211,
+      "step": 15600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.0756740740740746e-05,
+      "loss": 2.1538,
+      "step": 15605
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.075377777777778e-05,
+      "loss": 2.0437,
+      "step": 15610
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.075081481481482e-05,
+      "loss": 2.0719,
+      "step": 15615
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0747851851851856e-05,
+      "loss": 2.1299,
+      "step": 15620
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0744888888888895e-05,
+      "loss": 2.1515,
+      "step": 15625
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.074192592592593e-05,
+      "loss": 2.2022,
+      "step": 15630
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0738962962962966e-05,
+      "loss": 2.2535,
+      "step": 15635
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0736000000000004e-05,
+      "loss": 1.9738,
+      "step": 15640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0733037037037036e-05,
+      "loss": 2.1485,
+      "step": 15645
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0730074074074075e-05,
+      "loss": 2.0765,
+      "step": 15650
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.072711111111111e-05,
+      "loss": 2.272,
+      "step": 15655
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.072414814814815e-05,
+      "loss": 2.1975,
+      "step": 15660
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.0721185185185185e-05,
+      "loss": 2.2159,
+      "step": 15665
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0718222222222224e-05,
+      "loss": 2.0837,
+      "step": 15670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0715259259259256e-05,
+      "loss": 2.0685,
+      "step": 15675
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.07122962962963e-05,
+      "loss": 2.1558,
+      "step": 15680
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0709333333333333e-05,
+      "loss": 2.139,
+      "step": 15685
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.070637037037037e-05,
+      "loss": 2.0986,
+      "step": 15690
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.0703407407407404e-05,
+      "loss": 2.1454,
+      "step": 15695
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.070044444444445e-05,
+      "loss": 1.9436,
+      "step": 15700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.069748148148148e-05,
+      "loss": 1.9881,
+      "step": 15705
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.069451851851852e-05,
+      "loss": 2.0824,
+      "step": 15710
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.069155555555555e-05,
+      "loss": 2.1291,
+      "step": 15715
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.068859259259259e-05,
+      "loss": 2.1314,
+      "step": 15720
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.068562962962963e-05,
+      "loss": 2.0585,
+      "step": 15725
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.068266666666667e-05,
+      "loss": 2.0556,
+      "step": 15730
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.06797037037037e-05,
+      "loss": 2.1239,
+      "step": 15735
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.067674074074074e-05,
+      "loss": 2.0041,
+      "step": 15740
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.067377777777778e-05,
+      "loss": 2.0906,
+      "step": 15745
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.067081481481482e-05,
+      "loss": 1.9624,
+      "step": 15750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.066785185185185e-05,
+      "loss": 2.0015,
+      "step": 15755
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.066488888888889e-05,
+      "loss": 2.0668,
+      "step": 15760
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.066192592592593e-05,
+      "loss": 2.0416,
+      "step": 15765
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.0658962962962966e-05,
+      "loss": 2.101,
+      "step": 15770
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.0656e-05,
+      "loss": 2.0801,
+      "step": 15775
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.065303703703704e-05,
+      "loss": 2.0539,
+      "step": 15780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0650074074074076e-05,
+      "loss": 1.981,
+      "step": 15785
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0647111111111115e-05,
+      "loss": 2.0143,
+      "step": 15790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.064414814814815e-05,
+      "loss": 2.0821,
+      "step": 15795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0641185185185186e-05,
+      "loss": 2.056,
+      "step": 15800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0638222222222225e-05,
+      "loss": 2.1024,
+      "step": 15805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.063525925925926e-05,
+      "loss": 2.0003,
+      "step": 15810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0632296296296295e-05,
+      "loss": 2.0825,
+      "step": 15815
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.0629333333333334e-05,
+      "loss": 1.991,
+      "step": 15820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.062637037037037e-05,
+      "loss": 2.2716,
+      "step": 15825
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.062340740740741e-05,
+      "loss": 2.0591,
+      "step": 15830
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.0620444444444444e-05,
+      "loss": 2.2245,
+      "step": 15835
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.061748148148148e-05,
+      "loss": 1.9089,
+      "step": 15840
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.061451851851852e-05,
+      "loss": 2.0742,
+      "step": 15845
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.061155555555556e-05,
+      "loss": 2.1801,
+      "step": 15850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.060859259259259e-05,
+      "loss": 2.0128,
+      "step": 15855
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.060562962962963e-05,
+      "loss": 2.1646,
+      "step": 15860
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.060266666666666e-05,
+      "loss": 2.0965,
+      "step": 15865
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.059970370370371e-05,
+      "loss": 2.0245,
+      "step": 15870
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.059674074074074e-05,
+      "loss": 2.1484,
+      "step": 15875
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.059377777777778e-05,
+      "loss": 2.0386,
+      "step": 15880
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.059081481481481e-05,
+      "loss": 2.0924,
+      "step": 15885
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.058785185185186e-05,
+      "loss": 1.9741,
+      "step": 15890
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.058488888888889e-05,
+      "loss": 2.049,
+      "step": 15895
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.058192592592593e-05,
+      "loss": 2.1096,
+      "step": 15900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.057896296296296e-05,
+      "loss": 2.0821,
+      "step": 15905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0576000000000006e-05,
+      "loss": 1.9492,
+      "step": 15910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.057303703703704e-05,
+      "loss": 2.0418,
+      "step": 15915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.057007407407408e-05,
+      "loss": 2.0218,
+      "step": 15920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.056711111111111e-05,
+      "loss": 2.0541,
+      "step": 15925
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0564148148148154e-05,
+      "loss": 2.0128,
+      "step": 15930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0561185185185186e-05,
+      "loss": 2.0768,
+      "step": 15935
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.0558222222222225e-05,
+      "loss": 2.0105,
+      "step": 15940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.055525925925926e-05,
+      "loss": 2.0679,
+      "step": 15945
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0552296296296296e-05,
+      "loss": 2.0675,
+      "step": 15950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0549333333333335e-05,
+      "loss": 2.1774,
+      "step": 15955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0546370370370374e-05,
+      "loss": 2.1544,
+      "step": 15960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0543407407407406e-05,
+      "loss": 2.0531,
+      "step": 15965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0540444444444445e-05,
+      "loss": 2.0487,
+      "step": 15970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0537481481481483e-05,
+      "loss": 2.1262,
+      "step": 15975
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.053451851851852e-05,
+      "loss": 2.1444,
+      "step": 15980
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0531555555555554e-05,
+      "loss": 2.1414,
+      "step": 15985
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.052859259259259e-05,
+      "loss": 2.0467,
+      "step": 15990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.052562962962963e-05,
+      "loss": 1.9632,
+      "step": 15995
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.052266666666667e-05,
+      "loss": 1.9612,
+      "step": 16000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.05197037037037e-05,
+      "loss": 1.8637,
+      "step": 16005
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.051674074074074e-05,
+      "loss": 2.0664,
+      "step": 16010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.051377777777778e-05,
+      "loss": 2.1201,
+      "step": 16015
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.051081481481482e-05,
+      "loss": 2.068,
+      "step": 16020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.050785185185185e-05,
+      "loss": 2.1233,
+      "step": 16025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.050488888888889e-05,
+      "loss": 2.1708,
+      "step": 16030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.050192592592593e-05,
+      "loss": 1.8747,
+      "step": 16035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.049896296296297e-05,
+      "loss": 2.1514,
+      "step": 16040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.0496e-05,
+      "loss": 2.0846,
+      "step": 16045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.049303703703704e-05,
+      "loss": 2.0597,
+      "step": 16050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.049007407407408e-05,
+      "loss": 2.0686,
+      "step": 16055
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0487111111111116e-05,
+      "loss": 2.1085,
+      "step": 16060
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.048414814814815e-05,
+      "loss": 2.1299,
+      "step": 16065
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.048118518518519e-05,
+      "loss": 2.092,
+      "step": 16070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0478222222222226e-05,
+      "loss": 2.0897,
+      "step": 16075
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0475259259259265e-05,
+      "loss": 2.0561,
+      "step": 16080
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.04722962962963e-05,
+      "loss": 2.1786,
+      "step": 16085
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0469333333333336e-05,
+      "loss": 2.2351,
+      "step": 16090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.046637037037037e-05,
+      "loss": 2.1283,
+      "step": 16095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.046340740740741e-05,
+      "loss": 2.0544,
+      "step": 16100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0460444444444445e-05,
+      "loss": 2.0658,
+      "step": 16105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0457481481481484e-05,
+      "loss": 2.1455,
+      "step": 16110
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0454518518518516e-05,
+      "loss": 2.0699,
+      "step": 16115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.045155555555556e-05,
+      "loss": 2.083,
+      "step": 16120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0448592592592594e-05,
+      "loss": 2.1269,
+      "step": 16125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.044562962962963e-05,
+      "loss": 2.0783,
+      "step": 16130
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0442666666666665e-05,
+      "loss": 2.1505,
+      "step": 16135
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.043970370370371e-05,
+      "loss": 2.0506,
+      "step": 16140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.043674074074074e-05,
+      "loss": 2.085,
+      "step": 16145
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.043377777777778e-05,
+      "loss": 2.0843,
+      "step": 16150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.043081481481481e-05,
+      "loss": 2.1628,
+      "step": 16155
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.042785185185186e-05,
+      "loss": 2.0469,
+      "step": 16160
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.042488888888889e-05,
+      "loss": 2.0351,
+      "step": 16165
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.042192592592593e-05,
+      "loss": 1.9571,
+      "step": 16170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.041896296296296e-05,
+      "loss": 2.0478,
+      "step": 16175
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0416e-05,
+      "loss": 2.0621,
+      "step": 16180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.041303703703704e-05,
+      "loss": 2.0006,
+      "step": 16185
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.041007407407408e-05,
+      "loss": 2.1921,
+      "step": 16190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.040711111111111e-05,
+      "loss": 2.2011,
+      "step": 16195
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.040414814814815e-05,
+      "loss": 2.1462,
+      "step": 16200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.040118518518519e-05,
+      "loss": 2.0548,
+      "step": 16205
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.039822222222223e-05,
+      "loss": 2.0789,
+      "step": 16210
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.039525925925926e-05,
+      "loss": 2.1233,
+      "step": 16215
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.03922962962963e-05,
+      "loss": 2.1864,
+      "step": 16220
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0389333333333336e-05,
+      "loss": 2.0643,
+      "step": 16225
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0386370370370375e-05,
+      "loss": 2.1472,
+      "step": 16230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.038340740740741e-05,
+      "loss": 1.9254,
+      "step": 16235
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0380444444444446e-05,
+      "loss": 2.0204,
+      "step": 16240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0377481481481485e-05,
+      "loss": 1.9877,
+      "step": 16245
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0374518518518524e-05,
+      "loss": 2.0185,
+      "step": 16250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0371555555555556e-05,
+      "loss": 2.0657,
+      "step": 16255
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0368592592592595e-05,
+      "loss": 1.9766,
+      "step": 16260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0365629629629634e-05,
+      "loss": 1.9813,
+      "step": 16265
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.036266666666667e-05,
+      "loss": 2.1265,
+      "step": 16270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.0359703703703704e-05,
+      "loss": 2.1267,
+      "step": 16275
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.035674074074074e-05,
+      "loss": 2.0729,
+      "step": 16280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.035377777777778e-05,
+      "loss": 2.0036,
+      "step": 16285
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.035081481481482e-05,
+      "loss": 2.1389,
+      "step": 16290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.034785185185185e-05,
+      "loss": 2.0893,
+      "step": 16295
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.034488888888889e-05,
+      "loss": 2.2427,
+      "step": 16300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.034192592592593e-05,
+      "loss": 2.1558,
+      "step": 16305
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.033896296296297e-05,
+      "loss": 2.0431,
+      "step": 16310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.0336e-05,
+      "loss": 2.1553,
+      "step": 16315
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.033303703703704e-05,
+      "loss": 2.0283,
+      "step": 16320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.033007407407408e-05,
+      "loss": 2.1257,
+      "step": 16325
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.032711111111112e-05,
+      "loss": 2.1653,
+      "step": 16330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.032414814814815e-05,
+      "loss": 2.0811,
+      "step": 16335
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.032118518518519e-05,
+      "loss": 2.0757,
+      "step": 16340
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.031822222222222e-05,
+      "loss": 2.0354,
+      "step": 16345
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0315259259259266e-05,
+      "loss": 2.1375,
+      "step": 16350
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.03122962962963e-05,
+      "loss": 2.0141,
+      "step": 16355
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.030933333333334e-05,
+      "loss": 2.0928,
+      "step": 16360
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.030637037037037e-05,
+      "loss": 2.0547,
+      "step": 16365
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0303407407407415e-05,
+      "loss": 2.0986,
+      "step": 16370
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.030044444444445e-05,
+      "loss": 2.09,
+      "step": 16375
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0297481481481486e-05,
+      "loss": 2.1528,
+      "step": 16380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.029451851851852e-05,
+      "loss": 2.1105,
+      "step": 16385
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0291555555555563e-05,
+      "loss": 2.1813,
+      "step": 16390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.0288592592592595e-05,
+      "loss": 2.0998,
+      "step": 16395
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0285629629629634e-05,
+      "loss": 1.9956,
+      "step": 16400
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0282666666666666e-05,
+      "loss": 2.1837,
+      "step": 16405
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0279703703703705e-05,
+      "loss": 2.015,
+      "step": 16410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0276740740740744e-05,
+      "loss": 2.1512,
+      "step": 16415
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0273777777777776e-05,
+      "loss": 1.9866,
+      "step": 16420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0270814814814815e-05,
+      "loss": 2.0019,
+      "step": 16425
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0267851851851854e-05,
+      "loss": 2.0177,
+      "step": 16430
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.026488888888889e-05,
+      "loss": 2.0448,
+      "step": 16435
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0261925925925925e-05,
+      "loss": 2.2202,
+      "step": 16440
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.025896296296296e-05,
+      "loss": 1.9852,
+      "step": 16445
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0256e-05,
+      "loss": 1.8883,
+      "step": 16450
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.025303703703704e-05,
+      "loss": 1.9739,
+      "step": 16455
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.025007407407407e-05,
+      "loss": 2.116,
+      "step": 16460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.024711111111111e-05,
+      "loss": 2.0663,
+      "step": 16465
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.024414814814815e-05,
+      "loss": 1.9692,
+      "step": 16470
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.024118518518519e-05,
+      "loss": 1.9749,
+      "step": 16475
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.023822222222222e-05,
+      "loss": 1.9819,
+      "step": 16480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.023525925925926e-05,
+      "loss": 2.0088,
+      "step": 16485
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.023229629629629e-05,
+      "loss": 2.1145,
+      "step": 16490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.022933333333334e-05,
+      "loss": 2.0532,
+      "step": 16495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.022637037037037e-05,
+      "loss": 2.0437,
+      "step": 16500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.022340740740741e-05,
+      "loss": 2.0047,
+      "step": 16505
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.022044444444444e-05,
+      "loss": 2.1703,
+      "step": 16510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0217481481481487e-05,
+      "loss": 2.1873,
+      "step": 16515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.021451851851852e-05,
+      "loss": 2.0383,
+      "step": 16520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.021155555555556e-05,
+      "loss": 2.0219,
+      "step": 16525
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.020859259259259e-05,
+      "loss": 2.0059,
+      "step": 16530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0205629629629635e-05,
+      "loss": 2.0807,
+      "step": 16535
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.020266666666667e-05,
+      "loss": 2.129,
+      "step": 16540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0199703703703706e-05,
+      "loss": 2.2186,
+      "step": 16545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.019674074074074e-05,
+      "loss": 2.1436,
+      "step": 16550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0193777777777784e-05,
+      "loss": 2.16,
+      "step": 16555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0190814814814816e-05,
+      "loss": 2.1898,
+      "step": 16560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0187851851851854e-05,
+      "loss": 2.1175,
+      "step": 16565
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0184888888888886e-05,
+      "loss": 1.99,
+      "step": 16570
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0181925925925925e-05,
+      "loss": 2.023,
+      "step": 16575
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0178962962962964e-05,
+      "loss": 2.2951,
+      "step": 16580
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0176e-05,
+      "loss": 2.1258,
+      "step": 16585
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0173037037037035e-05,
+      "loss": 1.9182,
+      "step": 16590
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0170074074074074e-05,
+      "loss": 2.1291,
+      "step": 16595
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.016711111111111e-05,
+      "loss": 2.0255,
+      "step": 16600
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.016414814814815e-05,
+      "loss": 2.0234,
+      "step": 16605
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.0161185185185184e-05,
+      "loss": 2.1274,
+      "step": 16610
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.015822222222222e-05,
+      "loss": 2.0428,
+      "step": 16615
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.015525925925926e-05,
+      "loss": 2.164,
+      "step": 16620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.01522962962963e-05,
+      "loss": 1.9993,
+      "step": 16625
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.014933333333333e-05,
+      "loss": 2.0875,
+      "step": 16630
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.014637037037037e-05,
+      "loss": 2.0258,
+      "step": 16635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.014340740740741e-05,
+      "loss": 2.099,
+      "step": 16640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.014044444444445e-05,
+      "loss": 2.0906,
+      "step": 16645
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.013748148148148e-05,
+      "loss": 1.9489,
+      "step": 16650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.013451851851852e-05,
+      "loss": 2.1682,
+      "step": 16655
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.013155555555556e-05,
+      "loss": 2.0158,
+      "step": 16660
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.01285925925926e-05,
+      "loss": 2.0507,
+      "step": 16665
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.012562962962963e-05,
+      "loss": 1.9166,
+      "step": 16670
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.012266666666667e-05,
+      "loss": 2.155,
+      "step": 16675
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.011970370370371e-05,
+      "loss": 2.107,
+      "step": 16680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0116740740740746e-05,
+      "loss": 2.0062,
+      "step": 16685
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.011377777777778e-05,
+      "loss": 2.1848,
+      "step": 16690
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0110814814814816e-05,
+      "loss": 2.0409,
+      "step": 16695
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0107851851851855e-05,
+      "loss": 2.1192,
+      "step": 16700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0104888888888894e-05,
+      "loss": 2.07,
+      "step": 16705
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0101925925925926e-05,
+      "loss": 2.1515,
+      "step": 16710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0098962962962965e-05,
+      "loss": 2.0634,
+      "step": 16715
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0096e-05,
+      "loss": 2.1046,
+      "step": 16720
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.009303703703704e-05,
+      "loss": 2.1359,
+      "step": 16725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0090074074074075e-05,
+      "loss": 2.0086,
+      "step": 16730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0087111111111113e-05,
+      "loss": 2.0031,
+      "step": 16735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0084148148148145e-05,
+      "loss": 2.0127,
+      "step": 16740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.008118518518519e-05,
+      "loss": 2.0139,
+      "step": 16745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.007822222222222e-05,
+      "loss": 2.0884,
+      "step": 16750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.007525925925926e-05,
+      "loss": 1.9543,
+      "step": 16755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.0072296296296294e-05,
+      "loss": 1.9322,
+      "step": 16760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.006933333333334e-05,
+      "loss": 2.1212,
+      "step": 16765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.006637037037037e-05,
+      "loss": 2.0169,
+      "step": 16770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.006340740740741e-05,
+      "loss": 2.0677,
+      "step": 16775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.006044444444444e-05,
+      "loss": 1.9593,
+      "step": 16780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.005748148148149e-05,
+      "loss": 2.1301,
+      "step": 16785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.005451851851852e-05,
+      "loss": 2.0452,
+      "step": 16790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.005155555555556e-05,
+      "loss": 2.159,
+      "step": 16795
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.004859259259259e-05,
+      "loss": 1.9793,
+      "step": 16800
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.004562962962963e-05,
+      "loss": 1.9773,
+      "step": 16805
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.004266666666667e-05,
+      "loss": 2.0183,
+      "step": 16810
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.003970370370371e-05,
+      "loss": 2.0693,
+      "step": 16815
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.003674074074074e-05,
+      "loss": 2.1873,
+      "step": 16820
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.003377777777778e-05,
+      "loss": 2.1756,
+      "step": 16825
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.003081481481482e-05,
+      "loss": 2.1226,
+      "step": 16830
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0027851851851856e-05,
+      "loss": 2.115,
+      "step": 16835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.002488888888889e-05,
+      "loss": 1.9618,
+      "step": 16840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.002192592592593e-05,
+      "loss": 2.2295,
+      "step": 16845
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0018962962962966e-05,
+      "loss": 2.1335,
+      "step": 16850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0016000000000004e-05,
+      "loss": 2.0626,
+      "step": 16855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0013037037037037e-05,
+      "loss": 2.2053,
+      "step": 16860
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0010074074074075e-05,
+      "loss": 2.0684,
+      "step": 16865
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0007111111111114e-05,
+      "loss": 2.1864,
+      "step": 16870
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.000414814814815e-05,
+      "loss": 2.0419,
+      "step": 16875
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.0001185185185185e-05,
+      "loss": 1.804,
+      "step": 16880
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9998222222222224e-05,
+      "loss": 1.885,
+      "step": 16885
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.999525925925926e-05,
+      "loss": 2.0829,
+      "step": 16890
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.99922962962963e-05,
+      "loss": 1.9241,
+      "step": 16895
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.9989333333333334e-05,
+      "loss": 1.9544,
+      "step": 16900
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.998637037037037e-05,
+      "loss": 1.9004,
+      "step": 16905
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.998340740740741e-05,
+      "loss": 1.921,
+      "step": 16910
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.998044444444445e-05,
+      "loss": 1.9571,
+      "step": 16915
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.997748148148148e-05,
+      "loss": 2.0394,
+      "step": 16920
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.997451851851852e-05,
+      "loss": 1.8959,
+      "step": 16925
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.997155555555556e-05,
+      "loss": 1.8576,
+      "step": 16930
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.99685925925926e-05,
+      "loss": 1.9712,
+      "step": 16935
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.996562962962963e-05,
+      "loss": 2.0038,
+      "step": 16940
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.996266666666667e-05,
+      "loss": 1.9133,
+      "step": 16945
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.99597037037037e-05,
+      "loss": 2.0039,
+      "step": 16950
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.995674074074075e-05,
+      "loss": 1.9632,
+      "step": 16955
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.995377777777778e-05,
+      "loss": 1.921,
+      "step": 16960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.995081481481482e-05,
+      "loss": 1.9543,
+      "step": 16965
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.994785185185185e-05,
+      "loss": 2.0106,
+      "step": 16970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9944888888888896e-05,
+      "loss": 1.9748,
+      "step": 16975
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.994192592592593e-05,
+      "loss": 1.8221,
+      "step": 16980
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9938962962962966e-05,
+      "loss": 2.0455,
+      "step": 16985
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9936e-05,
+      "loss": 1.9707,
+      "step": 16990
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9933037037037044e-05,
+      "loss": 2.0332,
+      "step": 16995
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9930074074074076e-05,
+      "loss": 1.9653,
+      "step": 17000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.9927111111111115e-05,
+      "loss": 1.9639,
+      "step": 17005
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.992414814814815e-05,
+      "loss": 1.9335,
+      "step": 17010
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.992118518518519e-05,
+      "loss": 1.8478,
+      "step": 17015
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9918222222222225e-05,
+      "loss": 1.9567,
+      "step": 17020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9915259259259263e-05,
+      "loss": 1.9285,
+      "step": 17025
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9912296296296295e-05,
+      "loss": 1.9318,
+      "step": 17030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9909333333333334e-05,
+      "loss": 1.9002,
+      "step": 17035
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.990637037037037e-05,
+      "loss": 1.9526,
+      "step": 17040
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.990340740740741e-05,
+      "loss": 2.0249,
+      "step": 17045
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.9900444444444444e-05,
+      "loss": 1.9359,
+      "step": 17050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.989748148148148e-05,
+      "loss": 1.9477,
+      "step": 17055
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.989451851851852e-05,
+      "loss": 2.0777,
+      "step": 17060
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.989155555555556e-05,
+      "loss": 1.9016,
+      "step": 17065
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.988859259259259e-05,
+      "loss": 2.0511,
+      "step": 17070
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.988562962962963e-05,
+      "loss": 2.016,
+      "step": 17075
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.988266666666667e-05,
+      "loss": 1.9095,
+      "step": 17080
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.987970370370371e-05,
+      "loss": 1.8658,
+      "step": 17085
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.987674074074074e-05,
+      "loss": 2.0141,
+      "step": 17090
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.987377777777778e-05,
+      "loss": 2.0966,
+      "step": 17095
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.987081481481482e-05,
+      "loss": 1.9256,
+      "step": 17100
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.986785185185186e-05,
+      "loss": 1.937,
+      "step": 17105
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.986488888888889e-05,
+      "loss": 1.9478,
+      "step": 17110
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.986192592592593e-05,
+      "loss": 2.0359,
+      "step": 17115
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.985896296296297e-05,
+      "loss": 1.9003,
+      "step": 17120
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.9856000000000006e-05,
+      "loss": 1.853,
+      "step": 17125
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.985303703703704e-05,
+      "loss": 1.975,
+      "step": 17130
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.985007407407408e-05,
+      "loss": 2.0627,
+      "step": 17135
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9847111111111116e-05,
+      "loss": 1.867,
+      "step": 17140
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9844148148148155e-05,
+      "loss": 1.8958,
+      "step": 17145
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9841185185185187e-05,
+      "loss": 1.8877,
+      "step": 17150
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9838222222222225e-05,
+      "loss": 1.9654,
+      "step": 17155
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9835259259259264e-05,
+      "loss": 1.9467,
+      "step": 17160
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.98322962962963e-05,
+      "loss": 1.9552,
+      "step": 17165
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9829333333333335e-05,
+      "loss": 1.9617,
+      "step": 17170
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9826370370370374e-05,
+      "loss": 1.8018,
+      "step": 17175
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.9823407407407406e-05,
+      "loss": 1.9213,
+      "step": 17180
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.982044444444445e-05,
+      "loss": 1.9823,
+      "step": 17185
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9817481481481484e-05,
+      "loss": 1.9261,
+      "step": 17190
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9814518518518516e-05,
+      "loss": 1.9203,
+      "step": 17195
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9811555555555554e-05,
+      "loss": 1.884,
+      "step": 17200
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.980859259259259e-05,
+      "loss": 1.7776,
+      "step": 17205
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.980562962962963e-05,
+      "loss": 1.9108,
+      "step": 17210
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.9802666666666664e-05,
+      "loss": 2.0373,
+      "step": 17215
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.97997037037037e-05,
+      "loss": 1.9052,
+      "step": 17220
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.979674074074074e-05,
+      "loss": 1.871,
+      "step": 17225
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.979377777777778e-05,
+      "loss": 2.0622,
+      "step": 17230
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.979081481481481e-05,
+      "loss": 2.038,
+      "step": 17235
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.978785185185185e-05,
+      "loss": 1.9585,
+      "step": 17240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.978488888888889e-05,
+      "loss": 1.8964,
+      "step": 17245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.978192592592593e-05,
+      "loss": 2.0654,
+      "step": 17250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.977896296296296e-05,
+      "loss": 1.9225,
+      "step": 17255
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.9776e-05,
+      "loss": 2.111,
+      "step": 17260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.977303703703704e-05,
+      "loss": 1.9737,
+      "step": 17265
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.977007407407408e-05,
+      "loss": 1.9816,
+      "step": 17270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.976711111111111e-05,
+      "loss": 2.0066,
+      "step": 17275
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.976414814814815e-05,
+      "loss": 1.844,
+      "step": 17280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.976118518518519e-05,
+      "loss": 1.9811,
+      "step": 17285
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.9758222222222226e-05,
+      "loss": 1.9123,
+      "step": 17290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.975525925925926e-05,
+      "loss": 1.9633,
+      "step": 17295
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.97522962962963e-05,
+      "loss": 1.9782,
+      "step": 17300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9749333333333336e-05,
+      "loss": 1.9844,
+      "step": 17305
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9746370370370375e-05,
+      "loss": 1.8245,
+      "step": 17310
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.974340740740741e-05,
+      "loss": 1.9811,
+      "step": 17315
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9740444444444446e-05,
+      "loss": 2.0318,
+      "step": 17320
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9737481481481484e-05,
+      "loss": 1.7933,
+      "step": 17325
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.973451851851852e-05,
+      "loss": 1.9597,
+      "step": 17330
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9731555555555555e-05,
+      "loss": 1.9063,
+      "step": 17335
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9728592592592594e-05,
+      "loss": 1.912,
+      "step": 17340
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.9725629629629626e-05,
+      "loss": 1.9685,
+      "step": 17345
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.972266666666667e-05,
+      "loss": 1.9292,
+      "step": 17350
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.9719703703703704e-05,
+      "loss": 1.9578,
+      "step": 17355
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.971674074074074e-05,
+      "loss": 1.9085,
+      "step": 17360
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.9713777777777775e-05,
+      "loss": 1.8732,
+      "step": 17365
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.971081481481482e-05,
+      "loss": 1.9264,
+      "step": 17370
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.970785185185185e-05,
+      "loss": 2.0191,
+      "step": 17375
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.970488888888889e-05,
+      "loss": 2.0001,
+      "step": 17380
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.970192592592592e-05,
+      "loss": 1.9532,
+      "step": 17385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.969896296296297e-05,
+      "loss": 1.891,
+      "step": 17390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.9696e-05,
+      "loss": 1.9123,
+      "step": 17395
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.969303703703704e-05,
+      "loss": 2.018,
+      "step": 17400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.969007407407407e-05,
+      "loss": 1.9841,
+      "step": 17405
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.968711111111111e-05,
+      "loss": 2.0667,
+      "step": 17410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.968414814814815e-05,
+      "loss": 2.1308,
+      "step": 17415
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.968118518518519e-05,
+      "loss": 2.0163,
+      "step": 17420
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.967822222222222e-05,
+      "loss": 1.8905,
+      "step": 17425
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.967525925925926e-05,
+      "loss": 1.9742,
+      "step": 17430
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.96722962962963e-05,
+      "loss": 1.7909,
+      "step": 17435
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.9669333333333337e-05,
+      "loss": 1.9992,
+      "step": 17440
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.966637037037037e-05,
+      "loss": 2.0302,
+      "step": 17445
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.966340740740741e-05,
+      "loss": 1.8409,
+      "step": 17450
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.9660444444444446e-05,
+      "loss": 2.0036,
+      "step": 17455
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.9657481481481485e-05,
+      "loss": 1.851,
+      "step": 17460
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.965451851851852e-05,
+      "loss": 1.9512,
+      "step": 17465
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9651555555555556e-05,
+      "loss": 1.9695,
+      "step": 17470
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9648592592592595e-05,
+      "loss": 1.9616,
+      "step": 17475
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9645629629629634e-05,
+      "loss": 1.9008,
+      "step": 17480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9642666666666666e-05,
+      "loss": 2.038,
+      "step": 17485
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9639703703703705e-05,
+      "loss": 1.9402,
+      "step": 17490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.963674074074074e-05,
+      "loss": 1.9161,
+      "step": 17495
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.963377777777778e-05,
+      "loss": 1.9672,
+      "step": 17500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.9630814814814814e-05,
+      "loss": 1.9482,
+      "step": 17505
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.962785185185185e-05,
+      "loss": 1.9604,
+      "step": 17510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.962488888888889e-05,
+      "loss": 2.0035,
+      "step": 17515
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.962192592592593e-05,
+      "loss": 1.9753,
+      "step": 17520
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.961896296296296e-05,
+      "loss": 1.9415,
+      "step": 17525
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.9616e-05,
+      "loss": 1.7871,
+      "step": 17530
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.961303703703704e-05,
+      "loss": 1.9716,
+      "step": 17535
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.961007407407408e-05,
+      "loss": 1.965,
+      "step": 17540
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.960711111111111e-05,
+      "loss": 2.0166,
+      "step": 17545
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.960414814814815e-05,
+      "loss": 1.9227,
+      "step": 17550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.960118518518519e-05,
+      "loss": 2.0942,
+      "step": 17555
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.959822222222223e-05,
+      "loss": 2.0826,
+      "step": 17560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.959525925925926e-05,
+      "loss": 1.9108,
+      "step": 17565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.95922962962963e-05,
+      "loss": 1.9744,
+      "step": 17570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.958933333333333e-05,
+      "loss": 1.9758,
+      "step": 17575
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9586370370370376e-05,
+      "loss": 1.977,
+      "step": 17580
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.958340740740741e-05,
+      "loss": 1.9413,
+      "step": 17585
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.958044444444445e-05,
+      "loss": 1.9017,
+      "step": 17590
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.957748148148148e-05,
+      "loss": 1.9962,
+      "step": 17595
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9574518518518525e-05,
+      "loss": 1.8783,
+      "step": 17600
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.957155555555556e-05,
+      "loss": 1.9721,
+      "step": 17605
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9568592592592596e-05,
+      "loss": 1.9635,
+      "step": 17610
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.956562962962963e-05,
+      "loss": 1.918,
+      "step": 17615
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.956266666666667e-05,
+      "loss": 1.8462,
+      "step": 17620
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9559703703703705e-05,
+      "loss": 1.8161,
+      "step": 17625
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.9556740740740744e-05,
+      "loss": 2.0073,
+      "step": 17630
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9553777777777776e-05,
+      "loss": 1.8845,
+      "step": 17635
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9550814814814815e-05,
+      "loss": 1.9536,
+      "step": 17640
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9547851851851854e-05,
+      "loss": 1.9986,
+      "step": 17645
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9545481481481483e-05,
+      "loss": 1.8102,
+      "step": 17650
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.954251851851852e-05,
+      "loss": 2.0962,
+      "step": 17655
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9539555555555554e-05,
+      "loss": 1.98,
+      "step": 17660
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.953659259259259e-05,
+      "loss": 1.877,
+      "step": 17665
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.953362962962963e-05,
+      "loss": 1.8747,
+      "step": 17670
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.9530666666666664e-05,
+      "loss": 1.9352,
+      "step": 17675
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.95277037037037e-05,
+      "loss": 1.9904,
+      "step": 17680
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.952474074074074e-05,
+      "loss": 1.9266,
+      "step": 17685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.952177777777778e-05,
+      "loss": 1.9778,
+      "step": 17690
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.951881481481481e-05,
+      "loss": 1.931,
+      "step": 17695
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.951585185185185e-05,
+      "loss": 2.0493,
+      "step": 17700
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.951288888888889e-05,
+      "loss": 1.9475,
+      "step": 17705
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.950992592592593e-05,
+      "loss": 1.9939,
+      "step": 17710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.950696296296296e-05,
+      "loss": 1.9004,
+      "step": 17715
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.9504e-05,
+      "loss": 1.9634,
+      "step": 17720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.950103703703704e-05,
+      "loss": 1.9425,
+      "step": 17725
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.949807407407408e-05,
+      "loss": 2.0999,
+      "step": 17730
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.949511111111111e-05,
+      "loss": 1.8698,
+      "step": 17735
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.949214814814815e-05,
+      "loss": 2.0053,
+      "step": 17740
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.948918518518519e-05,
+      "loss": 1.8978,
+      "step": 17745
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9486222222222226e-05,
+      "loss": 1.9193,
+      "step": 17750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.948325925925926e-05,
+      "loss": 1.9145,
+      "step": 17755
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.94802962962963e-05,
+      "loss": 1.883,
+      "step": 17760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9477333333333336e-05,
+      "loss": 1.9842,
+      "step": 17765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9474370370370375e-05,
+      "loss": 1.8734,
+      "step": 17770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9471407407407407e-05,
+      "loss": 1.9567,
+      "step": 17775
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9468444444444445e-05,
+      "loss": 1.9424,
+      "step": 17780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9465481481481484e-05,
+      "loss": 1.8471,
+      "step": 17785
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.946251851851852e-05,
+      "loss": 1.9656,
+      "step": 17790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9459555555555555e-05,
+      "loss": 2.0159,
+      "step": 17795
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.9456592592592594e-05,
+      "loss": 1.8562,
+      "step": 17800
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.9453629629629626e-05,
+      "loss": 1.9071,
+      "step": 17805
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.945066666666667e-05,
+      "loss": 1.9102,
+      "step": 17810
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.9447703703703704e-05,
+      "loss": 1.972,
+      "step": 17815
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.944474074074074e-05,
+      "loss": 2.011,
+      "step": 17820
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.9441777777777774e-05,
+      "loss": 1.9088,
+      "step": 17825
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.943881481481482e-05,
+      "loss": 1.9605,
+      "step": 17830
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.943585185185185e-05,
+      "loss": 1.9324,
+      "step": 17835
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.943288888888889e-05,
+      "loss": 1.9872,
+      "step": 17840
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.942992592592592e-05,
+      "loss": 1.8406,
+      "step": 17845
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.942696296296297e-05,
+      "loss": 1.8601,
+      "step": 17850
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.9424e-05,
+      "loss": 1.9761,
+      "step": 17855
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.942103703703704e-05,
+      "loss": 1.9923,
+      "step": 17860
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.941807407407407e-05,
+      "loss": 1.8996,
+      "step": 17865
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.941511111111112e-05,
+      "loss": 1.9434,
+      "step": 17870
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.941214814814815e-05,
+      "loss": 1.9303,
+      "step": 17875
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.940918518518519e-05,
+      "loss": 1.9858,
+      "step": 17880
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.940622222222222e-05,
+      "loss": 1.8389,
+      "step": 17885
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.940325925925926e-05,
+      "loss": 1.9791,
+      "step": 17890
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.94002962962963e-05,
+      "loss": 1.9623,
+      "step": 17895
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.9397333333333337e-05,
+      "loss": 1.992,
+      "step": 17900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.939437037037037e-05,
+      "loss": 1.8295,
+      "step": 17905
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.939140740740741e-05,
+      "loss": 2.0538,
+      "step": 17910
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.9388444444444446e-05,
+      "loss": 1.8538,
+      "step": 17915
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9385481481481485e-05,
+      "loss": 1.9485,
+      "step": 17920
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.938251851851852e-05,
+      "loss": 1.8919,
+      "step": 17925
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9379555555555556e-05,
+      "loss": 1.9121,
+      "step": 17930
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9376592592592595e-05,
+      "loss": 1.8518,
+      "step": 17935
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9373629629629634e-05,
+      "loss": 1.894,
+      "step": 17940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9370666666666666e-05,
+      "loss": 1.8834,
+      "step": 17945
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9367703703703704e-05,
+      "loss": 2.1225,
+      "step": 17950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.936474074074074e-05,
+      "loss": 2.0327,
+      "step": 17955
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.936177777777778e-05,
+      "loss": 1.9276,
+      "step": 17960
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.9358814814814814e-05,
+      "loss": 1.9549,
+      "step": 17965
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.935585185185185e-05,
+      "loss": 1.9528,
+      "step": 17970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.935288888888889e-05,
+      "loss": 1.8547,
+      "step": 17975
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.934992592592593e-05,
+      "loss": 1.8724,
+      "step": 17980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.934696296296296e-05,
+      "loss": 1.9825,
+      "step": 17985
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.9344e-05,
+      "loss": 1.9112,
+      "step": 17990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.934103703703704e-05,
+      "loss": 2.068,
+      "step": 17995
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.933807407407408e-05,
+      "loss": 1.9957,
+      "step": 18000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.933511111111111e-05,
+      "loss": 1.8978,
+      "step": 18005
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.933214814814815e-05,
+      "loss": 1.9631,
+      "step": 18010
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.932918518518519e-05,
+      "loss": 1.9016,
+      "step": 18015
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.932622222222223e-05,
+      "loss": 2.1178,
+      "step": 18020
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.932325925925926e-05,
+      "loss": 1.992,
+      "step": 18025
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.93202962962963e-05,
+      "loss": 2.0326,
+      "step": 18030
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.931733333333333e-05,
+      "loss": 1.8142,
+      "step": 18035
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.9314370370370376e-05,
+      "loss": 1.9359,
+      "step": 18040
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.931140740740741e-05,
+      "loss": 2.0913,
+      "step": 18045
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.930844444444445e-05,
+      "loss": 1.997,
+      "step": 18050
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.930548148148148e-05,
+      "loss": 1.9322,
+      "step": 18055
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.9302518518518525e-05,
+      "loss": 1.9886,
+      "step": 18060
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.929955555555556e-05,
+      "loss": 2.0672,
+      "step": 18065
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.9296592592592595e-05,
+      "loss": 1.9143,
+      "step": 18070
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.929362962962963e-05,
+      "loss": 2.1223,
+      "step": 18075
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.929066666666667e-05,
+      "loss": 1.9278,
+      "step": 18080
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9287703703703705e-05,
+      "loss": 2.0303,
+      "step": 18085
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9284740740740744e-05,
+      "loss": 2.0346,
+      "step": 18090
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9281777777777776e-05,
+      "loss": 1.9867,
+      "step": 18095
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.927881481481482e-05,
+      "loss": 2.0778,
+      "step": 18100
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9275851851851854e-05,
+      "loss": 1.8423,
+      "step": 18105
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.927288888888889e-05,
+      "loss": 1.9294,
+      "step": 18110
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9269925925925925e-05,
+      "loss": 2.0413,
+      "step": 18115
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.926696296296296e-05,
+      "loss": 1.9332,
+      "step": 18120
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.9264e-05,
+      "loss": 1.9159,
+      "step": 18125
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.926103703703704e-05,
+      "loss": 1.9301,
+      "step": 18130
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.925807407407407e-05,
+      "loss": 2.0759,
+      "step": 18135
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.925511111111111e-05,
+      "loss": 2.0338,
+      "step": 18140
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.925214814814815e-05,
+      "loss": 2.0315,
+      "step": 18145
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.924918518518519e-05,
+      "loss": 1.937,
+      "step": 18150
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.924622222222222e-05,
+      "loss": 1.9797,
+      "step": 18155
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.924325925925926e-05,
+      "loss": 1.8933,
+      "step": 18160
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.92402962962963e-05,
+      "loss": 1.9699,
+      "step": 18165
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.923733333333334e-05,
+      "loss": 1.9557,
+      "step": 18170
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.923437037037037e-05,
+      "loss": 2.0219,
+      "step": 18175
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.923140740740741e-05,
+      "loss": 2.0288,
+      "step": 18180
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.922844444444445e-05,
+      "loss": 1.908,
+      "step": 18185
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.9225481481481487e-05,
+      "loss": 1.9916,
+      "step": 18190
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.922251851851852e-05,
+      "loss": 1.9366,
+      "step": 18195
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.921955555555556e-05,
+      "loss": 1.9199,
+      "step": 18200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9216592592592596e-05,
+      "loss": 1.9575,
+      "step": 18205
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9213629629629635e-05,
+      "loss": 1.944,
+      "step": 18210
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.921066666666667e-05,
+      "loss": 1.8943,
+      "step": 18215
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9207703703703706e-05,
+      "loss": 1.9914,
+      "step": 18220
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9204740740740745e-05,
+      "loss": 2.0343,
+      "step": 18225
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9201777777777784e-05,
+      "loss": 1.9615,
+      "step": 18230
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9198814814814816e-05,
+      "loss": 2.0135,
+      "step": 18235
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.9195851851851854e-05,
+      "loss": 1.9468,
+      "step": 18240
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.919288888888889e-05,
+      "loss": 2.0036,
+      "step": 18245
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.918992592592593e-05,
+      "loss": 1.9763,
+      "step": 18250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9186962962962964e-05,
+      "loss": 2.0044,
+      "step": 18255
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9184e-05,
+      "loss": 2.0888,
+      "step": 18260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9181037037037035e-05,
+      "loss": 1.9449,
+      "step": 18265
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.917807407407408e-05,
+      "loss": 1.8848,
+      "step": 18270
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.917511111111111e-05,
+      "loss": 2.011,
+      "step": 18275
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.917214814814815e-05,
+      "loss": 1.9649,
+      "step": 18280
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.9169185185185184e-05,
+      "loss": 1.9269,
+      "step": 18285
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.916622222222223e-05,
+      "loss": 1.9048,
+      "step": 18290
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.916325925925926e-05,
+      "loss": 2.0052,
+      "step": 18295
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.91602962962963e-05,
+      "loss": 2.0078,
+      "step": 18300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.915733333333333e-05,
+      "loss": 2.0223,
+      "step": 18305
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.915437037037038e-05,
+      "loss": 1.9351,
+      "step": 18310
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.915140740740741e-05,
+      "loss": 1.9307,
+      "step": 18315
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.914844444444445e-05,
+      "loss": 1.9483,
+      "step": 18320
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.914548148148148e-05,
+      "loss": 1.9641,
+      "step": 18325
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.9142518518518526e-05,
+      "loss": 2.0143,
+      "step": 18330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.913955555555556e-05,
+      "loss": 2.0224,
+      "step": 18335
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.91365925925926e-05,
+      "loss": 1.9595,
+      "step": 18340
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.913362962962963e-05,
+      "loss": 1.9511,
+      "step": 18345
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.913066666666667e-05,
+      "loss": 1.836,
+      "step": 18350
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.912770370370371e-05,
+      "loss": 1.9559,
+      "step": 18355
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.9124740740740746e-05,
+      "loss": 1.8311,
+      "step": 18360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.912177777777778e-05,
+      "loss": 1.9132,
+      "step": 18365
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9118814814814816e-05,
+      "loss": 1.9175,
+      "step": 18370
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9115851851851855e-05,
+      "loss": 1.9778,
+      "step": 18375
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9112888888888894e-05,
+      "loss": 1.9395,
+      "step": 18380
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9109925925925926e-05,
+      "loss": 1.9238,
+      "step": 18385
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9106962962962965e-05,
+      "loss": 1.9268,
+      "step": 18390
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9104000000000004e-05,
+      "loss": 1.9828,
+      "step": 18395
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.910103703703704e-05,
+      "loss": 1.8026,
+      "step": 18400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9098074074074075e-05,
+      "loss": 1.9459,
+      "step": 18405
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.9095111111111113e-05,
+      "loss": 2.0032,
+      "step": 18410
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.909214814814815e-05,
+      "loss": 1.8915,
+      "step": 18415
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.908918518518519e-05,
+      "loss": 1.9367,
+      "step": 18420
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.908622222222222e-05,
+      "loss": 1.928,
+      "step": 18425
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.908325925925926e-05,
+      "loss": 1.9992,
+      "step": 18430
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.90802962962963e-05,
+      "loss": 2.0017,
+      "step": 18435
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.907733333333333e-05,
+      "loss": 1.9322,
+      "step": 18440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.907437037037037e-05,
+      "loss": 1.9007,
+      "step": 18445
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.9071407407407404e-05,
+      "loss": 1.8458,
+      "step": 18450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.906844444444445e-05,
+      "loss": 1.9432,
+      "step": 18455
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.906548148148148e-05,
+      "loss": 1.9397,
+      "step": 18460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.906251851851852e-05,
+      "loss": 1.8468,
+      "step": 18465
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.905955555555555e-05,
+      "loss": 1.9747,
+      "step": 18470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.90565925925926e-05,
+      "loss": 1.9483,
+      "step": 18475
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.905362962962963e-05,
+      "loss": 2.1552,
+      "step": 18480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.905066666666667e-05,
+      "loss": 1.9504,
+      "step": 18485
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.90477037037037e-05,
+      "loss": 2.0795,
+      "step": 18490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.904474074074074e-05,
+      "loss": 1.8503,
+      "step": 18495
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.904177777777778e-05,
+      "loss": 1.9521,
+      "step": 18500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.903881481481482e-05,
+      "loss": 2.004,
+      "step": 18505
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.903585185185185e-05,
+      "loss": 2.0594,
+      "step": 18510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.903288888888889e-05,
+      "loss": 1.8791,
+      "step": 18515
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.902992592592593e-05,
+      "loss": 1.9404,
+      "step": 18520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.9026962962962966e-05,
+      "loss": 1.8909,
+      "step": 18525
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.9024e-05,
+      "loss": 2.0197,
+      "step": 18530
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9021037037037037e-05,
+      "loss": 1.8873,
+      "step": 18535
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9018074074074075e-05,
+      "loss": 1.97,
+      "step": 18540
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9015111111111114e-05,
+      "loss": 1.9168,
+      "step": 18545
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9012148148148146e-05,
+      "loss": 1.945,
+      "step": 18550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9009185185185185e-05,
+      "loss": 1.9144,
+      "step": 18555
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9006222222222224e-05,
+      "loss": 1.9031,
+      "step": 18560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.900325925925926e-05,
+      "loss": 1.768,
+      "step": 18565
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.9000296296296295e-05,
+      "loss": 1.95,
+      "step": 18570
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.8997333333333334e-05,
+      "loss": 2.0491,
+      "step": 18575
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.899437037037037e-05,
+      "loss": 1.919,
+      "step": 18580
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.899140740740741e-05,
+      "loss": 1.8325,
+      "step": 18585
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.898844444444444e-05,
+      "loss": 2.0896,
+      "step": 18590
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.898548148148148e-05,
+      "loss": 1.9612,
+      "step": 18595
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.898251851851852e-05,
+      "loss": 1.9357,
+      "step": 18600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.897955555555556e-05,
+      "loss": 1.8241,
+      "step": 18605
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.897659259259259e-05,
+      "loss": 1.9956,
+      "step": 18610
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.897362962962963e-05,
+      "loss": 2.0243,
+      "step": 18615
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.897066666666667e-05,
+      "loss": 1.908,
+      "step": 18620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.896770370370371e-05,
+      "loss": 1.8279,
+      "step": 18625
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.896474074074074e-05,
+      "loss": 1.9317,
+      "step": 18630
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.896177777777778e-05,
+      "loss": 2.0109,
+      "step": 18635
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.895881481481482e-05,
+      "loss": 1.9514,
+      "step": 18640
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.895585185185186e-05,
+      "loss": 1.8832,
+      "step": 18645
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.895288888888889e-05,
+      "loss": 1.9389,
+      "step": 18650
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.894992592592593e-05,
+      "loss": 1.9391,
+      "step": 18655
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.894696296296296e-05,
+      "loss": 2.0754,
+      "step": 18660
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.8944000000000005e-05,
+      "loss": 1.9924,
+      "step": 18665
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.894103703703704e-05,
+      "loss": 1.9044,
+      "step": 18670
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.8938074074074076e-05,
+      "loss": 2.0272,
+      "step": 18675
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.893511111111111e-05,
+      "loss": 1.8972,
+      "step": 18680
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.8932148148148154e-05,
+      "loss": 1.8988,
+      "step": 18685
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.8929185185185186e-05,
+      "loss": 1.9523,
+      "step": 18690
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.8926222222222225e-05,
+      "loss": 1.9462,
+      "step": 18695
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 3.892325925925926e-05,
+      "loss": 1.9631,
+      "step": 18700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.89202962962963e-05,
+      "loss": 2.0508,
+      "step": 18705
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.8917333333333334e-05,
+      "loss": 1.9613,
+      "step": 18710
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.891437037037037e-05,
+      "loss": 2.0406,
+      "step": 18715
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.8911407407407405e-05,
+      "loss": 1.933,
+      "step": 18720
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.8908444444444444e-05,
+      "loss": 1.9617,
+      "step": 18725
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.890548148148148e-05,
+      "loss": 1.9192,
+      "step": 18730
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.890251851851852e-05,
+      "loss": 2.0134,
+      "step": 18735
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.8899555555555554e-05,
+      "loss": 1.9253,
+      "step": 18740
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.889659259259259e-05,
+      "loss": 1.9259,
+      "step": 18745
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.889362962962963e-05,
+      "loss": 1.8575,
+      "step": 18750
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.889066666666667e-05,
+      "loss": 1.8478,
+      "step": 18755
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.88877037037037e-05,
+      "loss": 1.9355,
+      "step": 18760
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.888474074074074e-05,
+      "loss": 1.8894,
+      "step": 18765
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.888177777777778e-05,
+      "loss": 2.1438,
+      "step": 18770
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.887881481481482e-05,
+      "loss": 1.9795,
+      "step": 18775
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.887585185185185e-05,
+      "loss": 1.93,
+      "step": 18780
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.887288888888889e-05,
+      "loss": 1.9843,
+      "step": 18785
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.886992592592593e-05,
+      "loss": 2.0063,
+      "step": 18790
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.886696296296297e-05,
+      "loss": 1.9405,
+      "step": 18795
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.8864e-05,
+      "loss": 2.1478,
+      "step": 18800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.886103703703704e-05,
+      "loss": 1.8635,
+      "step": 18805
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.885807407407408e-05,
+      "loss": 1.9148,
+      "step": 18810
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.8855111111111116e-05,
+      "loss": 2.0098,
+      "step": 18815
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.885214814814815e-05,
+      "loss": 1.9239,
+      "step": 18820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8849185185185187e-05,
+      "loss": 1.8639,
+      "step": 18825
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8846222222222225e-05,
+      "loss": 1.8961,
+      "step": 18830
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8843259259259264e-05,
+      "loss": 2.0071,
+      "step": 18835
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8840296296296296e-05,
+      "loss": 1.9794,
+      "step": 18840
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8837333333333335e-05,
+      "loss": 1.8483,
+      "step": 18845
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8834370370370374e-05,
+      "loss": 2.0045,
+      "step": 18850
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.883140740740741e-05,
+      "loss": 1.9456,
+      "step": 18855
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8828444444444445e-05,
+      "loss": 1.826,
+      "step": 18860
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.8825481481481484e-05,
+      "loss": 2.0796,
+      "step": 18865
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.882251851851852e-05,
+      "loss": 2.0733,
+      "step": 18870
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.881955555555556e-05,
+      "loss": 1.9467,
+      "step": 18875
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.881659259259259e-05,
+      "loss": 1.9626,
+      "step": 18880
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.881362962962963e-05,
+      "loss": 1.8671,
+      "step": 18885
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.8810666666666664e-05,
+      "loss": 1.9425,
+      "step": 18890
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.880770370370371e-05,
+      "loss": 1.9938,
+      "step": 18895
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.880474074074074e-05,
+      "loss": 2.0454,
+      "step": 18900
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.880177777777778e-05,
+      "loss": 1.8612,
+      "step": 18905
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.879881481481481e-05,
+      "loss": 1.9771,
+      "step": 18910
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.879585185185186e-05,
+      "loss": 1.915,
+      "step": 18915
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.879288888888889e-05,
+      "loss": 1.9095,
+      "step": 18920
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.878992592592593e-05,
+      "loss": 2.052,
+      "step": 18925
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.878696296296296e-05,
+      "loss": 1.9379,
+      "step": 18930
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.878400000000001e-05,
+      "loss": 2.0275,
+      "step": 18935
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.878103703703704e-05,
+      "loss": 1.9561,
+      "step": 18940
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.877807407407408e-05,
+      "loss": 1.9538,
+      "step": 18945
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.877511111111111e-05,
+      "loss": 2.0729,
+      "step": 18950
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.877214814814815e-05,
+      "loss": 1.8977,
+      "step": 18955
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.876918518518519e-05,
+      "loss": 2.0689,
+      "step": 18960
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.8766222222222226e-05,
+      "loss": 1.9456,
+      "step": 18965
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.876325925925926e-05,
+      "loss": 1.9149,
+      "step": 18970
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.87602962962963e-05,
+      "loss": 2.0262,
+      "step": 18975
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.8757333333333336e-05,
+      "loss": 1.9573,
+      "step": 18980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8754370370370375e-05,
+      "loss": 1.9729,
+      "step": 18985
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.875140740740741e-05,
+      "loss": 1.9602,
+      "step": 18990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8748444444444446e-05,
+      "loss": 2.0735,
+      "step": 18995
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8745481481481484e-05,
+      "loss": 2.0241,
+      "step": 19000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.874251851851852e-05,
+      "loss": 2.0277,
+      "step": 19005
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8739555555555555e-05,
+      "loss": 1.9859,
+      "step": 19010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8736592592592594e-05,
+      "loss": 2.0162,
+      "step": 19015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.873362962962963e-05,
+      "loss": 2.0295,
+      "step": 19020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.873066666666667e-05,
+      "loss": 1.8969,
+      "step": 19025
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.8727703703703704e-05,
+      "loss": 1.9722,
+      "step": 19030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.872474074074074e-05,
+      "loss": 1.9406,
+      "step": 19035
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.872177777777778e-05,
+      "loss": 1.8335,
+      "step": 19040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.871881481481482e-05,
+      "loss": 1.9987,
+      "step": 19045
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.871585185185185e-05,
+      "loss": 1.959,
+      "step": 19050
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.871288888888889e-05,
+      "loss": 1.8879,
+      "step": 19055
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.870992592592593e-05,
+      "loss": 1.9152,
+      "step": 19060
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.870696296296297e-05,
+      "loss": 2.0334,
+      "step": 19065
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.8704e-05,
+      "loss": 1.9067,
+      "step": 19070
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.870103703703704e-05,
+      "loss": 1.9722,
+      "step": 19075
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.869807407407408e-05,
+      "loss": 2.0338,
+      "step": 19080
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.869511111111112e-05,
+      "loss": 1.8726,
+      "step": 19085
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.869214814814815e-05,
+      "loss": 2.0983,
+      "step": 19090
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.868918518518519e-05,
+      "loss": 1.9237,
+      "step": 19095
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.868622222222223e-05,
+      "loss": 1.93,
+      "step": 19100
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8683259259259266e-05,
+      "loss": 1.9989,
+      "step": 19105
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.86802962962963e-05,
+      "loss": 2.0115,
+      "step": 19110
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.867733333333334e-05,
+      "loss": 2.0316,
+      "step": 19115
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.867437037037037e-05,
+      "loss": 1.9183,
+      "step": 19120
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8671407407407414e-05,
+      "loss": 2.0088,
+      "step": 19125
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8668444444444446e-05,
+      "loss": 2.0003,
+      "step": 19130
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8665481481481485e-05,
+      "loss": 1.8878,
+      "step": 19135
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.866251851851852e-05,
+      "loss": 2.0486,
+      "step": 19140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.865955555555556e-05,
+      "loss": 2.0623,
+      "step": 19145
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.8656592592592595e-05,
+      "loss": 2.0215,
+      "step": 19150
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.8653629629629634e-05,
+      "loss": 1.9206,
+      "step": 19155
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.8650666666666666e-05,
+      "loss": 1.9868,
+      "step": 19160
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.864770370370371e-05,
+      "loss": 1.983,
+      "step": 19165
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.864474074074074e-05,
+      "loss": 1.9487,
+      "step": 19170
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.864177777777778e-05,
+      "loss": 1.8834,
+      "step": 19175
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.8638814814814814e-05,
+      "loss": 1.9915,
+      "step": 19180
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.863585185185185e-05,
+      "loss": 1.9366,
+      "step": 19185
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.863288888888889e-05,
+      "loss": 1.9728,
+      "step": 19190
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.862992592592593e-05,
+      "loss": 1.8977,
+      "step": 19195
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.862696296296296e-05,
+      "loss": 1.9914,
+      "step": 19200
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.8624e-05,
+      "loss": 1.9226,
+      "step": 19205
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.862103703703704e-05,
+      "loss": 1.9111,
+      "step": 19210
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.861807407407407e-05,
+      "loss": 2.0225,
+      "step": 19215
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.861511111111111e-05,
+      "loss": 1.9123,
+      "step": 19220
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.861214814814815e-05,
+      "loss": 1.9542,
+      "step": 19225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.860918518518519e-05,
+      "loss": 1.9121,
+      "step": 19230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.860622222222222e-05,
+      "loss": 1.7409,
+      "step": 19235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.860325925925926e-05,
+      "loss": 2.1915,
+      "step": 19240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.86002962962963e-05,
+      "loss": 1.7424,
+      "step": 19245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.859733333333334e-05,
+      "loss": 2.0072,
+      "step": 19250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.859437037037037e-05,
+      "loss": 1.8385,
+      "step": 19255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.859140740740741e-05,
+      "loss": 2.0189,
+      "step": 19260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.858844444444445e-05,
+      "loss": 1.8994,
+      "step": 19265
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8585481481481486e-05,
+      "loss": 1.9649,
+      "step": 19270
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.858251851851852e-05,
+      "loss": 2.0742,
+      "step": 19275
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.857955555555556e-05,
+      "loss": 1.8545,
+      "step": 19280
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.857659259259259e-05,
+      "loss": 1.9987,
+      "step": 19285
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8573629629629634e-05,
+      "loss": 1.963,
+      "step": 19290
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8570666666666666e-05,
+      "loss": 1.9324,
+      "step": 19295
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8567703703703705e-05,
+      "loss": 1.7869,
+      "step": 19300
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.856474074074074e-05,
+      "loss": 1.9619,
+      "step": 19305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.856177777777778e-05,
+      "loss": 2.0605,
+      "step": 19310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8558814814814815e-05,
+      "loss": 2.0127,
+      "step": 19315
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.8555851851851854e-05,
+      "loss": 1.9554,
+      "step": 19320
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.8552888888888886e-05,
+      "loss": 1.9327,
+      "step": 19325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.854992592592593e-05,
+      "loss": 1.9179,
+      "step": 19330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.8546962962962963e-05,
+      "loss": 2.0802,
+      "step": 19335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.8544e-05,
+      "loss": 2.0724,
+      "step": 19340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.8541037037037034e-05,
+      "loss": 2.0894,
+      "step": 19345
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.853807407407407e-05,
+      "loss": 2.0267,
+      "step": 19350
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.853511111111111e-05,
+      "loss": 1.964,
+      "step": 19355
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.853214814814815e-05,
+      "loss": 2.045,
+      "step": 19360
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.852918518518518e-05,
+      "loss": 1.8906,
+      "step": 19365
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.852622222222222e-05,
+      "loss": 2.0693,
+      "step": 19370
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.852325925925926e-05,
+      "loss": 1.8995,
+      "step": 19375
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.85202962962963e-05,
+      "loss": 2.1533,
+      "step": 19380
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.851733333333333e-05,
+      "loss": 1.8457,
+      "step": 19385
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.851437037037037e-05,
+      "loss": 2.0033,
+      "step": 19390
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.851140740740741e-05,
+      "loss": 1.9044,
+      "step": 19395
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.850844444444445e-05,
+      "loss": 1.8425,
+      "step": 19400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.850548148148148e-05,
+      "loss": 1.8263,
+      "step": 19405
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.850251851851852e-05,
+      "loss": 1.9774,
+      "step": 19410
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.849955555555556e-05,
+      "loss": 1.9672,
+      "step": 19415
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.8496592592592596e-05,
+      "loss": 1.7683,
+      "step": 19420
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.849362962962963e-05,
+      "loss": 2.0004,
+      "step": 19425
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.849066666666667e-05,
+      "loss": 2.1091,
+      "step": 19430
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8487703703703706e-05,
+      "loss": 1.8965,
+      "step": 19435
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8484740740740745e-05,
+      "loss": 1.8952,
+      "step": 19440
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.848177777777778e-05,
+      "loss": 1.8952,
+      "step": 19445
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8478814814814816e-05,
+      "loss": 1.9744,
+      "step": 19450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8475851851851855e-05,
+      "loss": 2.0103,
+      "step": 19455
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.847288888888889e-05,
+      "loss": 1.992,
+      "step": 19460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8469925925925925e-05,
+      "loss": 1.9147,
+      "step": 19465
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8466962962962964e-05,
+      "loss": 2.0468,
+      "step": 19470
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8464e-05,
+      "loss": 1.9042,
+      "step": 19475
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.846103703703704e-05,
+      "loss": 1.942,
+      "step": 19480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.8458074074074074e-05,
+      "loss": 1.8422,
+      "step": 19485
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.845511111111111e-05,
+      "loss": 1.9813,
+      "step": 19490
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.845214814814815e-05,
+      "loss": 2.0436,
+      "step": 19495
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.844918518518519e-05,
+      "loss": 1.9923,
+      "step": 19500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.844622222222222e-05,
+      "loss": 1.7841,
+      "step": 19505
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.844325925925926e-05,
+      "loss": 1.8681,
+      "step": 19510
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.844029629629629e-05,
+      "loss": 1.9571,
+      "step": 19515
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.843733333333334e-05,
+      "loss": 1.9107,
+      "step": 19520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.843437037037037e-05,
+      "loss": 1.9601,
+      "step": 19525
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.843140740740741e-05,
+      "loss": 2.0451,
+      "step": 19530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.842844444444444e-05,
+      "loss": 1.9737,
+      "step": 19535
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.842548148148149e-05,
+      "loss": 1.9557,
+      "step": 19540
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.842251851851852e-05,
+      "loss": 2.0266,
+      "step": 19545
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.841955555555556e-05,
+      "loss": 1.7994,
+      "step": 19550
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.841659259259259e-05,
+      "loss": 1.8544,
+      "step": 19555
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8413629629629636e-05,
+      "loss": 2.1024,
+      "step": 19560
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.841066666666667e-05,
+      "loss": 1.9766,
+      "step": 19565
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.840770370370371e-05,
+      "loss": 1.9036,
+      "step": 19570
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.840474074074074e-05,
+      "loss": 2.0583,
+      "step": 19575
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.840177777777778e-05,
+      "loss": 1.9418,
+      "step": 19580
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8398814814814817e-05,
+      "loss": 2.0346,
+      "step": 19585
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8395851851851855e-05,
+      "loss": 1.989,
+      "step": 19590
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.839288888888889e-05,
+      "loss": 2.02,
+      "step": 19595
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.8389925925925926e-05,
+      "loss": 1.7909,
+      "step": 19600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8386962962962965e-05,
+      "loss": 2.0601,
+      "step": 19605
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8384000000000004e-05,
+      "loss": 1.9828,
+      "step": 19610
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8381037037037036e-05,
+      "loss": 2.0091,
+      "step": 19615
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8378074074074075e-05,
+      "loss": 1.9604,
+      "step": 19620
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8375111111111114e-05,
+      "loss": 2.0798,
+      "step": 19625
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.837214814814815e-05,
+      "loss": 2.0536,
+      "step": 19630
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8369185185185184e-05,
+      "loss": 1.846,
+      "step": 19635
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.836622222222222e-05,
+      "loss": 1.957,
+      "step": 19640
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.836325925925926e-05,
+      "loss": 1.9611,
+      "step": 19645
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.83602962962963e-05,
+      "loss": 1.9307,
+      "step": 19650
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.835733333333333e-05,
+      "loss": 1.8096,
+      "step": 19655
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.835437037037037e-05,
+      "loss": 1.883,
+      "step": 19660
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.835140740740741e-05,
+      "loss": 1.9386,
+      "step": 19665
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.834844444444445e-05,
+      "loss": 1.9331,
+      "step": 19670
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.834548148148148e-05,
+      "loss": 1.9057,
+      "step": 19675
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.834251851851852e-05,
+      "loss": 1.8606,
+      "step": 19680
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.833955555555556e-05,
+      "loss": 2.0291,
+      "step": 19685
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.83365925925926e-05,
+      "loss": 1.9986,
+      "step": 19690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.833362962962963e-05,
+      "loss": 1.918,
+      "step": 19695
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.833066666666667e-05,
+      "loss": 2.0606,
+      "step": 19700
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.832770370370371e-05,
+      "loss": 2.0654,
+      "step": 19705
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8324740740740746e-05,
+      "loss": 1.9007,
+      "step": 19710
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.832177777777778e-05,
+      "loss": 1.8761,
+      "step": 19715
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.831881481481482e-05,
+      "loss": 2.009,
+      "step": 19720
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8315851851851856e-05,
+      "loss": 1.8596,
+      "step": 19725
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8312888888888895e-05,
+      "loss": 1.9878,
+      "step": 19730
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.830992592592593e-05,
+      "loss": 1.9412,
+      "step": 19735
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8306962962962966e-05,
+      "loss": 1.9634,
+      "step": 19740
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8304e-05,
+      "loss": 2.0041,
+      "step": 19745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8301037037037043e-05,
+      "loss": 1.9637,
+      "step": 19750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8298074074074075e-05,
+      "loss": 1.9728,
+      "step": 19755
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8295111111111114e-05,
+      "loss": 2.0181,
+      "step": 19760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.8292148148148146e-05,
+      "loss": 1.9003,
+      "step": 19765
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.828918518518519e-05,
+      "loss": 1.9066,
+      "step": 19770
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.8286222222222224e-05,
+      "loss": 2.0553,
+      "step": 19775
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.828325925925926e-05,
+      "loss": 1.9709,
+      "step": 19780
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.8280296296296295e-05,
+      "loss": 1.9395,
+      "step": 19785
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.827733333333334e-05,
+      "loss": 1.9761,
+      "step": 19790
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.827437037037037e-05,
+      "loss": 2.0571,
+      "step": 19795
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.827140740740741e-05,
+      "loss": 1.9006,
+      "step": 19800
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.826844444444444e-05,
+      "loss": 2.0519,
+      "step": 19805
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.826548148148148e-05,
+      "loss": 2.0014,
+      "step": 19810
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.826251851851852e-05,
+      "loss": 1.9408,
+      "step": 19815
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.825955555555556e-05,
+      "loss": 2.0233,
+      "step": 19820
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.825659259259259e-05,
+      "loss": 2.0875,
+      "step": 19825
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.825362962962963e-05,
+      "loss": 1.8947,
+      "step": 19830
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.825066666666667e-05,
+      "loss": 1.9824,
+      "step": 19835
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.824770370370371e-05,
+      "loss": 1.9553,
+      "step": 19840
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.824474074074074e-05,
+      "loss": 1.9245,
+      "step": 19845
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.824177777777778e-05,
+      "loss": 2.0935,
+      "step": 19850
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.823881481481482e-05,
+      "loss": 1.8683,
+      "step": 19855
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.823585185185186e-05,
+      "loss": 1.9085,
+      "step": 19860
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.823288888888889e-05,
+      "loss": 2.0078,
+      "step": 19865
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.822992592592593e-05,
+      "loss": 1.9467,
+      "step": 19870
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.8226962962962967e-05,
+      "loss": 1.929,
+      "step": 19875
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.8224000000000005e-05,
+      "loss": 1.9232,
+      "step": 19880
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.822103703703704e-05,
+      "loss": 1.9001,
+      "step": 19885
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8218074074074076e-05,
+      "loss": 1.7978,
+      "step": 19890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8215111111111115e-05,
+      "loss": 2.0708,
+      "step": 19895
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8212148148148154e-05,
+      "loss": 2.0007,
+      "step": 19900
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8209185185185186e-05,
+      "loss": 2.1008,
+      "step": 19905
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8206222222222225e-05,
+      "loss": 1.8995,
+      "step": 19910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8203259259259264e-05,
+      "loss": 1.9984,
+      "step": 19915
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.82002962962963e-05,
+      "loss": 2.049,
+      "step": 19920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8197333333333334e-05,
+      "loss": 2.0505,
+      "step": 19925
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.819437037037037e-05,
+      "loss": 1.9941,
+      "step": 19930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.819140740740741e-05,
+      "loss": 1.9888,
+      "step": 19935
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.818844444444445e-05,
+      "loss": 1.9068,
+      "step": 19940
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.818548148148148e-05,
+      "loss": 1.8618,
+      "step": 19945
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.818251851851852e-05,
+      "loss": 1.9384,
+      "step": 19950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.817955555555556e-05,
+      "loss": 1.9812,
+      "step": 19955
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.81765925925926e-05,
+      "loss": 1.9673,
+      "step": 19960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.817362962962963e-05,
+      "loss": 1.8537,
+      "step": 19965
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.817066666666667e-05,
+      "loss": 2.0594,
+      "step": 19970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.81677037037037e-05,
+      "loss": 1.9228,
+      "step": 19975
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.816474074074074e-05,
+      "loss": 1.9924,
+      "step": 19980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.816177777777778e-05,
+      "loss": 2.013,
+      "step": 19985
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.815881481481481e-05,
+      "loss": 1.9676,
+      "step": 19990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.815585185185185e-05,
+      "loss": 1.8215,
+      "step": 19995
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.815288888888889e-05,
+      "loss": 1.9639,
+      "step": 20000
+    },
+    {
+      "epoch": 3.56,
+      "eval_loss": 1.8181427717208862,
+      "eval_rouge2_fmeasure": 0.1771,
+      "eval_rouge2_precision": 0.2125,
+      "eval_rouge2_recall": 0.1606,
+      "eval_runtime": 36846.5383,
+      "eval_samples_per_second": 0.136,
+      "eval_steps_per_second": 0.068,
+      "step": 20000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.814992592592593e-05,
+      "loss": 1.9242,
+      "step": 20005
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.814696296296296e-05,
+      "loss": 1.8884,
+      "step": 20010
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.8144e-05,
+      "loss": 1.8602,
+      "step": 20015
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.814103703703704e-05,
+      "loss": 2.146,
+      "step": 20020
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.813807407407408e-05,
+      "loss": 1.914,
+      "step": 20025
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.813511111111111e-05,
+      "loss": 1.9613,
+      "step": 20030
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.813214814814815e-05,
+      "loss": 1.8841,
+      "step": 20035
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.812918518518519e-05,
+      "loss": 1.9845,
+      "step": 20040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.8126222222222226e-05,
+      "loss": 2.0209,
+      "step": 20045
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.812325925925926e-05,
+      "loss": 1.9506,
+      "step": 20050
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8120296296296296e-05,
+      "loss": 1.9947,
+      "step": 20055
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8117333333333335e-05,
+      "loss": 1.917,
+      "step": 20060
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8114370370370374e-05,
+      "loss": 1.986,
+      "step": 20065
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8111407407407406e-05,
+      "loss": 2.004,
+      "step": 20070
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8108444444444445e-05,
+      "loss": 1.9111,
+      "step": 20075
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8105481481481484e-05,
+      "loss": 1.8988,
+      "step": 20080
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.810251851851852e-05,
+      "loss": 1.9325,
+      "step": 20085
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8099555555555555e-05,
+      "loss": 1.9578,
+      "step": 20090
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.8096592592592593e-05,
+      "loss": 1.9725,
+      "step": 20095
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.809362962962963e-05,
+      "loss": 1.9984,
+      "step": 20100
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.809066666666667e-05,
+      "loss": 1.9189,
+      "step": 20105
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.80877037037037e-05,
+      "loss": 2.0368,
+      "step": 20110
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.808474074074074e-05,
+      "loss": 1.9537,
+      "step": 20115
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.8081777777777774e-05,
+      "loss": 2.0385,
+      "step": 20120
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.807881481481482e-05,
+      "loss": 1.9464,
+      "step": 20125
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.807585185185185e-05,
+      "loss": 1.928,
+      "step": 20130
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.807288888888889e-05,
+      "loss": 2.0644,
+      "step": 20135
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.806992592592592e-05,
+      "loss": 2.0066,
+      "step": 20140
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.806696296296297e-05,
+      "loss": 1.9024,
+      "step": 20145
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.8064e-05,
+      "loss": 2.1443,
+      "step": 20150
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.806103703703704e-05,
+      "loss": 1.9592,
+      "step": 20155
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.805807407407407e-05,
+      "loss": 1.9597,
+      "step": 20160
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.8055111111111117e-05,
+      "loss": 1.9378,
+      "step": 20165
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.805214814814815e-05,
+      "loss": 2.0411,
+      "step": 20170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.804918518518519e-05,
+      "loss": 2.0039,
+      "step": 20175
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.804622222222222e-05,
+      "loss": 2.0573,
+      "step": 20180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8043259259259265e-05,
+      "loss": 1.9441,
+      "step": 20185
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.80402962962963e-05,
+      "loss": 1.9339,
+      "step": 20190
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8037333333333336e-05,
+      "loss": 2.0266,
+      "step": 20195
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.803437037037037e-05,
+      "loss": 1.9893,
+      "step": 20200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.803140740740741e-05,
+      "loss": 2.0178,
+      "step": 20205
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8028444444444446e-05,
+      "loss": 1.9329,
+      "step": 20210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8025481481481484e-05,
+      "loss": 1.994,
+      "step": 20215
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.8022518518518517e-05,
+      "loss": 1.9525,
+      "step": 20220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8019555555555555e-05,
+      "loss": 1.7788,
+      "step": 20225
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8016592592592594e-05,
+      "loss": 2.0706,
+      "step": 20230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.801362962962963e-05,
+      "loss": 1.7772,
+      "step": 20235
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8010666666666665e-05,
+      "loss": 1.9923,
+      "step": 20240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8007703703703704e-05,
+      "loss": 1.9539,
+      "step": 20245
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.800474074074074e-05,
+      "loss": 1.9335,
+      "step": 20250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.800177777777778e-05,
+      "loss": 2.0497,
+      "step": 20255
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.7998814814814814e-05,
+      "loss": 1.9357,
+      "step": 20260
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.799585185185185e-05,
+      "loss": 2.0328,
+      "step": 20265
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.799288888888889e-05,
+      "loss": 1.9598,
+      "step": 20270
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.798992592592593e-05,
+      "loss": 1.936,
+      "step": 20275
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.798696296296296e-05,
+      "loss": 2.0266,
+      "step": 20280
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7984e-05,
+      "loss": 2.0422,
+      "step": 20285
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.798103703703704e-05,
+      "loss": 1.9786,
+      "step": 20290
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.797807407407408e-05,
+      "loss": 2.0306,
+      "step": 20295
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.797511111111111e-05,
+      "loss": 2.0431,
+      "step": 20300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.797214814814815e-05,
+      "loss": 1.9746,
+      "step": 20305
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.796918518518519e-05,
+      "loss": 1.9534,
+      "step": 20310
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.796622222222223e-05,
+      "loss": 1.9085,
+      "step": 20315
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.796325925925926e-05,
+      "loss": 2.0734,
+      "step": 20320
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.79602962962963e-05,
+      "loss": 1.8987,
+      "step": 20325
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.795733333333334e-05,
+      "loss": 1.9868,
+      "step": 20330
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7954370370370376e-05,
+      "loss": 1.9497,
+      "step": 20335
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.795140740740741e-05,
+      "loss": 2.0128,
+      "step": 20340
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7948444444444446e-05,
+      "loss": 1.9626,
+      "step": 20345
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.794548148148148e-05,
+      "loss": 1.9887,
+      "step": 20350
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7942518518518524e-05,
+      "loss": 2.0096,
+      "step": 20355
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7939555555555556e-05,
+      "loss": 1.9029,
+      "step": 20360
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7936592592592595e-05,
+      "loss": 1.9932,
+      "step": 20365
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.793362962962963e-05,
+      "loss": 1.9596,
+      "step": 20370
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.793066666666667e-05,
+      "loss": 1.9879,
+      "step": 20375
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7927703703703705e-05,
+      "loss": 1.8336,
+      "step": 20380
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7924740740740743e-05,
+      "loss": 1.9932,
+      "step": 20385
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7921777777777775e-05,
+      "loss": 1.9208,
+      "step": 20390
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.791881481481482e-05,
+      "loss": 1.8098,
+      "step": 20395
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.791585185185185e-05,
+      "loss": 1.9656,
+      "step": 20400
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.791288888888889e-05,
+      "loss": 1.9073,
+      "step": 20405
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7909925925925924e-05,
+      "loss": 2.0418,
+      "step": 20410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.790696296296297e-05,
+      "loss": 2.0456,
+      "step": 20415
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.7904e-05,
+      "loss": 1.9665,
+      "step": 20420
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.790103703703704e-05,
+      "loss": 1.8169,
+      "step": 20425
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.789807407407407e-05,
+      "loss": 1.9226,
+      "step": 20430
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.789511111111111e-05,
+      "loss": 2.0569,
+      "step": 20435
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.789214814814815e-05,
+      "loss": 1.9005,
+      "step": 20440
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.788918518518519e-05,
+      "loss": 2.1218,
+      "step": 20445
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.788622222222222e-05,
+      "loss": 1.886,
+      "step": 20450
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.788325925925926e-05,
+      "loss": 2.0779,
+      "step": 20455
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.78802962962963e-05,
+      "loss": 2.0185,
+      "step": 20460
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.787733333333334e-05,
+      "loss": 1.9625,
+      "step": 20465
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.787437037037037e-05,
+      "loss": 1.9478,
+      "step": 20470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.787140740740741e-05,
+      "loss": 2.0012,
+      "step": 20475
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.786844444444445e-05,
+      "loss": 2.0288,
+      "step": 20480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.7865481481481486e-05,
+      "loss": 1.9601,
+      "step": 20485
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.786251851851852e-05,
+      "loss": 2.0511,
+      "step": 20490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.785955555555556e-05,
+      "loss": 1.9344,
+      "step": 20495
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.7856592592592596e-05,
+      "loss": 1.9684,
+      "step": 20500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7853629629629635e-05,
+      "loss": 1.8574,
+      "step": 20505
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7850666666666667e-05,
+      "loss": 1.9739,
+      "step": 20510
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7847703703703705e-05,
+      "loss": 1.9307,
+      "step": 20515
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7844740740740744e-05,
+      "loss": 1.9406,
+      "step": 20520
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.784177777777778e-05,
+      "loss": 1.8881,
+      "step": 20525
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7838814814814815e-05,
+      "loss": 1.9621,
+      "step": 20530
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7835851851851854e-05,
+      "loss": 1.84,
+      "step": 20535
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.783288888888889e-05,
+      "loss": 1.8347,
+      "step": 20540
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.782992592592593e-05,
+      "loss": 1.8915,
+      "step": 20545
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7826962962962964e-05,
+      "loss": 1.8863,
+      "step": 20550
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.7824e-05,
+      "loss": 1.9382,
+      "step": 20555
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.782103703703704e-05,
+      "loss": 1.8947,
+      "step": 20560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.781807407407408e-05,
+      "loss": 1.9358,
+      "step": 20565
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.781511111111111e-05,
+      "loss": 1.9603,
+      "step": 20570
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.781214814814815e-05,
+      "loss": 1.8794,
+      "step": 20575
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.780918518518519e-05,
+      "loss": 1.9687,
+      "step": 20580
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.780622222222223e-05,
+      "loss": 1.9004,
+      "step": 20585
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.780325925925926e-05,
+      "loss": 2.0997,
+      "step": 20590
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.78002962962963e-05,
+      "loss": 1.9507,
+      "step": 20595
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.779733333333333e-05,
+      "loss": 1.9861,
+      "step": 20600
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.779437037037038e-05,
+      "loss": 2.0139,
+      "step": 20605
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.779140740740741e-05,
+      "loss": 2.0509,
+      "step": 20610
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.778844444444445e-05,
+      "loss": 1.9274,
+      "step": 20615
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.778548148148148e-05,
+      "loss": 2.0703,
+      "step": 20620
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7782518518518526e-05,
+      "loss": 2.0063,
+      "step": 20625
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.777955555555556e-05,
+      "loss": 1.9817,
+      "step": 20630
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7776592592592596e-05,
+      "loss": 2.1252,
+      "step": 20635
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.777362962962963e-05,
+      "loss": 1.9124,
+      "step": 20640
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7770666666666674e-05,
+      "loss": 1.94,
+      "step": 20645
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7767703703703706e-05,
+      "loss": 2.0087,
+      "step": 20650
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7764740740740745e-05,
+      "loss": 2.0185,
+      "step": 20655
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.776177777777778e-05,
+      "loss": 1.941,
+      "step": 20660
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7758814814814816e-05,
+      "loss": 1.8871,
+      "step": 20665
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.7755851851851855e-05,
+      "loss": 2.1713,
+      "step": 20670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7752888888888893e-05,
+      "loss": 2.1213,
+      "step": 20675
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7749925925925926e-05,
+      "loss": 1.8814,
+      "step": 20680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7746962962962964e-05,
+      "loss": 1.9309,
+      "step": 20685
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7744e-05,
+      "loss": 1.9567,
+      "step": 20690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.774103703703704e-05,
+      "loss": 1.9551,
+      "step": 20695
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.7738074074074074e-05,
+      "loss": 1.9598,
+      "step": 20700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.773511111111111e-05,
+      "loss": 1.8591,
+      "step": 20705
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.773214814814815e-05,
+      "loss": 1.9494,
+      "step": 20710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.772918518518519e-05,
+      "loss": 1.9957,
+      "step": 20715
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.772622222222222e-05,
+      "loss": 1.96,
+      "step": 20720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.772325925925926e-05,
+      "loss": 1.8708,
+      "step": 20725
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.77202962962963e-05,
+      "loss": 2.0776,
+      "step": 20730
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.771733333333334e-05,
+      "loss": 1.8553,
+      "step": 20735
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.771437037037037e-05,
+      "loss": 1.9028,
+      "step": 20740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.771140740740741e-05,
+      "loss": 2.0337,
+      "step": 20745
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.770844444444445e-05,
+      "loss": 1.9958,
+      "step": 20750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.770548148148148e-05,
+      "loss": 1.8491,
+      "step": 20755
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.770251851851852e-05,
+      "loss": 1.9595,
+      "step": 20760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.769955555555555e-05,
+      "loss": 1.9609,
+      "step": 20765
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.76965925925926e-05,
+      "loss": 1.9981,
+      "step": 20770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.769362962962963e-05,
+      "loss": 1.9176,
+      "step": 20775
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.769066666666667e-05,
+      "loss": 1.9321,
+      "step": 20780
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.76877037037037e-05,
+      "loss": 1.9501,
+      "step": 20785
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7684740740740746e-05,
+      "loss": 1.881,
+      "step": 20790
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.768177777777778e-05,
+      "loss": 1.9772,
+      "step": 20795
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.767881481481482e-05,
+      "loss": 1.9585,
+      "step": 20800
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.767585185185185e-05,
+      "loss": 1.8707,
+      "step": 20805
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7672888888888894e-05,
+      "loss": 1.8278,
+      "step": 20810
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7669925925925926e-05,
+      "loss": 2.0021,
+      "step": 20815
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7666962962962965e-05,
+      "loss": 1.9674,
+      "step": 20820
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7664e-05,
+      "loss": 2.0349,
+      "step": 20825
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7661037037037036e-05,
+      "loss": 1.9602,
+      "step": 20830
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7658074074074075e-05,
+      "loss": 2.0821,
+      "step": 20835
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.7655111111111114e-05,
+      "loss": 1.977,
+      "step": 20840
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7652148148148146e-05,
+      "loss": 1.9405,
+      "step": 20845
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7649185185185185e-05,
+      "loss": 1.8417,
+      "step": 20850
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.764622222222222e-05,
+      "loss": 1.9594,
+      "step": 20855
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.764325925925926e-05,
+      "loss": 2.0869,
+      "step": 20860
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.7640296296296294e-05,
+      "loss": 2.0832,
+      "step": 20865
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.763733333333333e-05,
+      "loss": 1.972,
+      "step": 20870
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.763437037037037e-05,
+      "loss": 1.9647,
+      "step": 20875
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.763140740740741e-05,
+      "loss": 2.0283,
+      "step": 20880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.762844444444444e-05,
+      "loss": 1.8196,
+      "step": 20885
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.762548148148148e-05,
+      "loss": 1.9829,
+      "step": 20890
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.762251851851852e-05,
+      "loss": 1.8277,
+      "step": 20895
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.761955555555556e-05,
+      "loss": 1.9101,
+      "step": 20900
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.761659259259259e-05,
+      "loss": 2.0292,
+      "step": 20905
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.761362962962963e-05,
+      "loss": 1.9849,
+      "step": 20910
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.761066666666667e-05,
+      "loss": 1.8762,
+      "step": 20915
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.760770370370371e-05,
+      "loss": 1.9212,
+      "step": 20920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.760474074074074e-05,
+      "loss": 1.9331,
+      "step": 20925
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.760177777777778e-05,
+      "loss": 1.9854,
+      "step": 20930
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.759881481481482e-05,
+      "loss": 1.9306,
+      "step": 20935
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.7595851851851856e-05,
+      "loss": 2.0732,
+      "step": 20940
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.759288888888889e-05,
+      "loss": 1.9525,
+      "step": 20945
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.758992592592593e-05,
+      "loss": 1.9798,
+      "step": 20950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7586962962962966e-05,
+      "loss": 2.0048,
+      "step": 20955
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7584000000000005e-05,
+      "loss": 2.086,
+      "step": 20960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.758103703703704e-05,
+      "loss": 2.0286,
+      "step": 20965
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7578074074074076e-05,
+      "loss": 1.9974,
+      "step": 20970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.757511111111111e-05,
+      "loss": 2.0797,
+      "step": 20975
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.757214814814815e-05,
+      "loss": 1.9855,
+      "step": 20980
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7569185185185185e-05,
+      "loss": 1.9892,
+      "step": 20985
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7566222222222224e-05,
+      "loss": 1.9774,
+      "step": 20990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7563259259259256e-05,
+      "loss": 2.0364,
+      "step": 20995
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.75602962962963e-05,
+      "loss": 1.9657,
+      "step": 21000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.7557333333333334e-05,
+      "loss": 1.8665,
+      "step": 21005
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.755437037037037e-05,
+      "loss": 1.879,
+      "step": 21010
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.7551407407407405e-05,
+      "loss": 1.9235,
+      "step": 21015
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.754844444444445e-05,
+      "loss": 2.0489,
+      "step": 21020
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.754548148148148e-05,
+      "loss": 2.0077,
+      "step": 21025
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.754251851851852e-05,
+      "loss": 1.8935,
+      "step": 21030
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.753955555555555e-05,
+      "loss": 1.9394,
+      "step": 21035
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.75365925925926e-05,
+      "loss": 1.8245,
+      "step": 21040
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.753362962962963e-05,
+      "loss": 1.9497,
+      "step": 21045
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.753066666666667e-05,
+      "loss": 1.8306,
+      "step": 21050
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.75277037037037e-05,
+      "loss": 1.9738,
+      "step": 21055
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.752474074074074e-05,
+      "loss": 2.0362,
+      "step": 21060
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.752177777777778e-05,
+      "loss": 1.9915,
+      "step": 21065
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.751881481481482e-05,
+      "loss": 2.0905,
+      "step": 21070
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.751585185185185e-05,
+      "loss": 1.9069,
+      "step": 21075
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.751288888888889e-05,
+      "loss": 1.8922,
+      "step": 21080
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.750992592592593e-05,
+      "loss": 1.764,
+      "step": 21085
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.750696296296297e-05,
+      "loss": 1.8953,
+      "step": 21090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7504e-05,
+      "loss": 2.0851,
+      "step": 21095
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.750103703703704e-05,
+      "loss": 2.0103,
+      "step": 21100
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7498074074074076e-05,
+      "loss": 1.9683,
+      "step": 21105
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7495111111111115e-05,
+      "loss": 1.9679,
+      "step": 21110
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.749214814814815e-05,
+      "loss": 1.9317,
+      "step": 21115
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.7489185185185186e-05,
+      "loss": 1.933,
+      "step": 21120
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7486222222222225e-05,
+      "loss": 2.0687,
+      "step": 21125
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7483259259259264e-05,
+      "loss": 2.0281,
+      "step": 21130
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7480296296296296e-05,
+      "loss": 2.0827,
+      "step": 21135
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7477333333333335e-05,
+      "loss": 1.8444,
+      "step": 21140
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.747437037037037e-05,
+      "loss": 2.0821,
+      "step": 21145
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.747140740740741e-05,
+      "loss": 1.8199,
+      "step": 21150
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.7468444444444444e-05,
+      "loss": 1.8965,
+      "step": 21155
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.746548148148148e-05,
+      "loss": 1.9435,
+      "step": 21160
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.746251851851852e-05,
+      "loss": 2.0255,
+      "step": 21165
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.745955555555556e-05,
+      "loss": 1.9893,
+      "step": 21170
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.745659259259259e-05,
+      "loss": 1.847,
+      "step": 21175
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.745362962962963e-05,
+      "loss": 2.0298,
+      "step": 21180
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.745066666666667e-05,
+      "loss": 1.891,
+      "step": 21185
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.744770370370371e-05,
+      "loss": 1.8723,
+      "step": 21190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.744474074074074e-05,
+      "loss": 1.9551,
+      "step": 21195
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.744177777777778e-05,
+      "loss": 1.9063,
+      "step": 21200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.743881481481481e-05,
+      "loss": 2.0195,
+      "step": 21205
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.743585185185186e-05,
+      "loss": 1.9634,
+      "step": 21210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.743288888888889e-05,
+      "loss": 2.0447,
+      "step": 21215
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.742992592592593e-05,
+      "loss": 2.0322,
+      "step": 21220
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.742696296296296e-05,
+      "loss": 2.0058,
+      "step": 21225
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.7424000000000006e-05,
+      "loss": 2.0062,
+      "step": 21230
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.742103703703704e-05,
+      "loss": 2.0324,
+      "step": 21235
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.741807407407408e-05,
+      "loss": 1.9296,
+      "step": 21240
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.741511111111111e-05,
+      "loss": 1.9292,
+      "step": 21245
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7412148148148155e-05,
+      "loss": 1.9179,
+      "step": 21250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.740918518518519e-05,
+      "loss": 2.0071,
+      "step": 21255
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7406222222222226e-05,
+      "loss": 1.9377,
+      "step": 21260
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.740325925925926e-05,
+      "loss": 2.0115,
+      "step": 21265
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.74002962962963e-05,
+      "loss": 2.0358,
+      "step": 21270
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7397333333333335e-05,
+      "loss": 1.9841,
+      "step": 21275
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7394370370370374e-05,
+      "loss": 2.0212,
+      "step": 21280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7391407407407406e-05,
+      "loss": 1.9864,
+      "step": 21285
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.7388444444444445e-05,
+      "loss": 1.9788,
+      "step": 21290
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7385481481481484e-05,
+      "loss": 1.9798,
+      "step": 21295
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.738251851851852e-05,
+      "loss": 1.9088,
+      "step": 21300
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7379555555555555e-05,
+      "loss": 1.8635,
+      "step": 21305
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.7376592592592594e-05,
+      "loss": 1.925,
+      "step": 21310
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.737362962962963e-05,
+      "loss": 2.0639,
+      "step": 21315
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.737066666666667e-05,
+      "loss": 1.9953,
+      "step": 21320
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.73677037037037e-05,
+      "loss": 1.9843,
+      "step": 21325
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.736474074074074e-05,
+      "loss": 1.9552,
+      "step": 21330
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.736177777777778e-05,
+      "loss": 1.9286,
+      "step": 21335
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.735881481481482e-05,
+      "loss": 1.8823,
+      "step": 21340
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.735585185185185e-05,
+      "loss": 2.0201,
+      "step": 21345
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.735288888888889e-05,
+      "loss": 1.9328,
+      "step": 21350
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.734992592592593e-05,
+      "loss": 1.9121,
+      "step": 21355
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.734696296296297e-05,
+      "loss": 2.0631,
+      "step": 21360
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7344e-05,
+      "loss": 1.8997,
+      "step": 21365
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.734103703703704e-05,
+      "loss": 2.0884,
+      "step": 21370
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.733807407407408e-05,
+      "loss": 1.8873,
+      "step": 21375
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.733511111111112e-05,
+      "loss": 1.9064,
+      "step": 21380
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.733214814814815e-05,
+      "loss": 1.9223,
+      "step": 21385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.732918518518519e-05,
+      "loss": 2.0491,
+      "step": 21390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7326222222222226e-05,
+      "loss": 2.0175,
+      "step": 21395
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.7323259259259265e-05,
+      "loss": 1.9188,
+      "step": 21400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.73202962962963e-05,
+      "loss": 1.9284,
+      "step": 21405
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7317333333333336e-05,
+      "loss": 1.7904,
+      "step": 21410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7314370370370375e-05,
+      "loss": 1.9965,
+      "step": 21415
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7311407407407414e-05,
+      "loss": 1.9632,
+      "step": 21420
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7308444444444446e-05,
+      "loss": 1.9321,
+      "step": 21425
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7305481481481485e-05,
+      "loss": 1.9903,
+      "step": 21430
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.730251851851852e-05,
+      "loss": 2.0091,
+      "step": 21435
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.729955555555556e-05,
+      "loss": 2.0615,
+      "step": 21440
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7296592592592594e-05,
+      "loss": 1.7409,
+      "step": 21445
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.729362962962963e-05,
+      "loss": 1.9476,
+      "step": 21450
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.7290666666666665e-05,
+      "loss": 1.7534,
+      "step": 21455
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.728770370370371e-05,
+      "loss": 2.0288,
+      "step": 21460
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.728474074074074e-05,
+      "loss": 1.8851,
+      "step": 21465
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.728177777777778e-05,
+      "loss": 2.092,
+      "step": 21470
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.7278814814814814e-05,
+      "loss": 2.0597,
+      "step": 21475
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.727585185185186e-05,
+      "loss": 1.9215,
+      "step": 21480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.727288888888889e-05,
+      "loss": 1.8503,
+      "step": 21485
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.726992592592593e-05,
+      "loss": 1.9161,
+      "step": 21490
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.726696296296296e-05,
+      "loss": 2.0338,
+      "step": 21495
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.726400000000001e-05,
+      "loss": 1.9394,
+      "step": 21500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.726103703703704e-05,
+      "loss": 1.997,
+      "step": 21505
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.725807407407408e-05,
+      "loss": 1.9885,
+      "step": 21510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.725511111111111e-05,
+      "loss": 1.893,
+      "step": 21515
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.725214814814815e-05,
+      "loss": 1.9817,
+      "step": 21520
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.724918518518519e-05,
+      "loss": 1.9804,
+      "step": 21525
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.724622222222222e-05,
+      "loss": 1.962,
+      "step": 21530
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.724325925925926e-05,
+      "loss": 1.9211,
+      "step": 21535
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.72402962962963e-05,
+      "loss": 2.0399,
+      "step": 21540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.723733333333334e-05,
+      "loss": 2.0076,
+      "step": 21545
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.723437037037037e-05,
+      "loss": 1.9601,
+      "step": 21550
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.723140740740741e-05,
+      "loss": 1.9727,
+      "step": 21555
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.7228444444444447e-05,
+      "loss": 1.9021,
+      "step": 21560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.7225481481481485e-05,
+      "loss": 1.9097,
+      "step": 21565
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.722251851851852e-05,
+      "loss": 2.0889,
+      "step": 21570
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7219555555555556e-05,
+      "loss": 2.0256,
+      "step": 21575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7216592592592595e-05,
+      "loss": 1.9004,
+      "step": 21580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7213629629629634e-05,
+      "loss": 1.9035,
+      "step": 21585
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7210666666666666e-05,
+      "loss": 2.0012,
+      "step": 21590
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7207703703703705e-05,
+      "loss": 1.8954,
+      "step": 21595
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.720474074074074e-05,
+      "loss": 1.8018,
+      "step": 21600
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.720177777777778e-05,
+      "loss": 1.8887,
+      "step": 21605
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7198814814814814e-05,
+      "loss": 1.8725,
+      "step": 21610
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.719585185185185e-05,
+      "loss": 2.031,
+      "step": 21615
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.7192888888888885e-05,
+      "loss": 1.9748,
+      "step": 21620
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.718992592592593e-05,
+      "loss": 1.9256,
+      "step": 21625
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.718696296296296e-05,
+      "loss": 1.9378,
+      "step": 21630
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7184e-05,
+      "loss": 1.9179,
+      "step": 21635
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.7181037037037034e-05,
+      "loss": 1.9444,
+      "step": 21640
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.717807407407408e-05,
+      "loss": 1.9535,
+      "step": 21645
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.717511111111111e-05,
+      "loss": 1.9446,
+      "step": 21650
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.717214814814815e-05,
+      "loss": 1.7618,
+      "step": 21655
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.716918518518518e-05,
+      "loss": 1.8998,
+      "step": 21660
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.716622222222222e-05,
+      "loss": 2.0236,
+      "step": 21665
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.716325925925926e-05,
+      "loss": 1.8581,
+      "step": 21670
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.71602962962963e-05,
+      "loss": 1.9523,
+      "step": 21675
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.715733333333333e-05,
+      "loss": 1.9489,
+      "step": 21680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.715437037037037e-05,
+      "loss": 2.0347,
+      "step": 21685
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.715140740740741e-05,
+      "loss": 2.0028,
+      "step": 21690
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.714844444444445e-05,
+      "loss": 1.8828,
+      "step": 21695
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.714548148148148e-05,
+      "loss": 1.8815,
+      "step": 21700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.714251851851852e-05,
+      "loss": 1.9018,
+      "step": 21705
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.713955555555556e-05,
+      "loss": 2.0293,
+      "step": 21710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7136592592592596e-05,
+      "loss": 2.035,
+      "step": 21715
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.713362962962963e-05,
+      "loss": 1.9395,
+      "step": 21720
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.713066666666667e-05,
+      "loss": 2.0302,
+      "step": 21725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7127703703703706e-05,
+      "loss": 1.9714,
+      "step": 21730
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7124740740740744e-05,
+      "loss": 1.8971,
+      "step": 21735
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.7121777777777776e-05,
+      "loss": 2.0585,
+      "step": 21740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7118814814814815e-05,
+      "loss": 1.925,
+      "step": 21745
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7115851851851854e-05,
+      "loss": 1.9618,
+      "step": 21750
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.711288888888889e-05,
+      "loss": 1.966,
+      "step": 21755
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7109925925925925e-05,
+      "loss": 2.0209,
+      "step": 21760
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7106962962962964e-05,
+      "loss": 1.8004,
+      "step": 21765
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7104e-05,
+      "loss": 2.0392,
+      "step": 21770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.710103703703704e-05,
+      "loss": 1.9251,
+      "step": 21775
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.7098074074074073e-05,
+      "loss": 1.9758,
+      "step": 21780
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.709511111111111e-05,
+      "loss": 1.8781,
+      "step": 21785
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.709214814814815e-05,
+      "loss": 1.95,
+      "step": 21790
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.708918518518519e-05,
+      "loss": 1.9053,
+      "step": 21795
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.708622222222222e-05,
+      "loss": 1.8633,
+      "step": 21800
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.708325925925926e-05,
+      "loss": 1.8152,
+      "step": 21805
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.70802962962963e-05,
+      "loss": 1.9397,
+      "step": 21810
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.707733333333334e-05,
+      "loss": 1.9326,
+      "step": 21815
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.707437037037037e-05,
+      "loss": 2.0459,
+      "step": 21820
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.707140740740741e-05,
+      "loss": 2.101,
+      "step": 21825
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.706844444444444e-05,
+      "loss": 1.8689,
+      "step": 21830
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.706548148148149e-05,
+      "loss": 1.9857,
+      "step": 21835
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.706251851851852e-05,
+      "loss": 1.8677,
+      "step": 21840
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.705955555555556e-05,
+      "loss": 1.8501,
+      "step": 21845
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.705659259259259e-05,
+      "loss": 1.9851,
+      "step": 21850
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7053629629629635e-05,
+      "loss": 2.0446,
+      "step": 21855
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.705066666666667e-05,
+      "loss": 2.0297,
+      "step": 21860
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7047703703703706e-05,
+      "loss": 1.9441,
+      "step": 21865
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.704474074074074e-05,
+      "loss": 1.951,
+      "step": 21870
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7041777777777784e-05,
+      "loss": 1.9539,
+      "step": 21875
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7038814814814816e-05,
+      "loss": 1.8907,
+      "step": 21880
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7035851851851855e-05,
+      "loss": 2.052,
+      "step": 21885
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.703288888888889e-05,
+      "loss": 1.9617,
+      "step": 21890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7029925925925926e-05,
+      "loss": 1.9081,
+      "step": 21895
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7026962962962964e-05,
+      "loss": 1.9695,
+      "step": 21900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.7024e-05,
+      "loss": 1.8691,
+      "step": 21905
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.7021037037037035e-05,
+      "loss": 1.9065,
+      "step": 21910
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.7018074074074074e-05,
+      "loss": 1.9233,
+      "step": 21915
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.701511111111111e-05,
+      "loss": 1.912,
+      "step": 21920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.701214814814815e-05,
+      "loss": 1.9988,
+      "step": 21925
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.7009185185185184e-05,
+      "loss": 1.9526,
+      "step": 21930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.700622222222222e-05,
+      "loss": 2.0439,
+      "step": 21935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.700325925925926e-05,
+      "loss": 2.021,
+      "step": 21940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.70002962962963e-05,
+      "loss": 2.0019,
+      "step": 21945
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.699733333333333e-05,
+      "loss": 1.9089,
+      "step": 21950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.699437037037037e-05,
+      "loss": 1.9837,
+      "step": 21955
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.699140740740741e-05,
+      "loss": 1.9148,
+      "step": 21960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.698844444444445e-05,
+      "loss": 1.967,
+      "step": 21965
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.698548148148148e-05,
+      "loss": 1.9087,
+      "step": 21970
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.698251851851852e-05,
+      "loss": 1.9309,
+      "step": 21975
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.697955555555556e-05,
+      "loss": 1.9613,
+      "step": 21980
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.69765925925926e-05,
+      "loss": 1.881,
+      "step": 21985
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.697362962962963e-05,
+      "loss": 1.9466,
+      "step": 21990
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.697066666666667e-05,
+      "loss": 1.9503,
+      "step": 21995
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.696770370370371e-05,
+      "loss": 2.1085,
+      "step": 22000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.6964740740740746e-05,
+      "loss": 1.9026,
+      "step": 22005
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.696177777777778e-05,
+      "loss": 2.0555,
+      "step": 22010
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.695881481481482e-05,
+      "loss": 1.8716,
+      "step": 22015
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.6955851851851856e-05,
+      "loss": 1.9142,
+      "step": 22020
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6952888888888894e-05,
+      "loss": 1.9085,
+      "step": 22025
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6949925925925926e-05,
+      "loss": 2.0172,
+      "step": 22030
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6946962962962965e-05,
+      "loss": 2.0271,
+      "step": 22035
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6944000000000004e-05,
+      "loss": 1.9035,
+      "step": 22040
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.694103703703704e-05,
+      "loss": 1.8703,
+      "step": 22045
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6938074074074075e-05,
+      "loss": 2.0103,
+      "step": 22050
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6935111111111114e-05,
+      "loss": 1.9128,
+      "step": 22055
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6932148148148146e-05,
+      "loss": 1.9627,
+      "step": 22060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.692918518518519e-05,
+      "loss": 1.973,
+      "step": 22065
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.6926222222222223e-05,
+      "loss": 2.063,
+      "step": 22070
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.692325925925926e-05,
+      "loss": 2.0007,
+      "step": 22075
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.6920296296296294e-05,
+      "loss": 2.0242,
+      "step": 22080
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.691733333333334e-05,
+      "loss": 2.0554,
+      "step": 22085
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.691437037037037e-05,
+      "loss": 2.0196,
+      "step": 22090
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.691140740740741e-05,
+      "loss": 2.1185,
+      "step": 22095
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.690844444444444e-05,
+      "loss": 1.8731,
+      "step": 22100
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.690548148148149e-05,
+      "loss": 1.9425,
+      "step": 22105
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.690251851851852e-05,
+      "loss": 1.8755,
+      "step": 22110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.689955555555556e-05,
+      "loss": 2.0374,
+      "step": 22115
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.689659259259259e-05,
+      "loss": 1.9534,
+      "step": 22120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.689362962962963e-05,
+      "loss": 2.0297,
+      "step": 22125
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.689066666666667e-05,
+      "loss": 2.0105,
+      "step": 22130
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.688770370370371e-05,
+      "loss": 1.8696,
+      "step": 22135
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.688474074074074e-05,
+      "loss": 1.8431,
+      "step": 22140
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.688177777777778e-05,
+      "loss": 1.9253,
+      "step": 22145
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.687881481481482e-05,
+      "loss": 1.8729,
+      "step": 22150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6875851851851856e-05,
+      "loss": 1.9633,
+      "step": 22155
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.687288888888889e-05,
+      "loss": 1.8941,
+      "step": 22160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.686992592592593e-05,
+      "loss": 1.8762,
+      "step": 22165
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6866962962962966e-05,
+      "loss": 1.999,
+      "step": 22170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6864000000000005e-05,
+      "loss": 1.879,
+      "step": 22175
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.686103703703704e-05,
+      "loss": 1.9317,
+      "step": 22180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6858074074074076e-05,
+      "loss": 1.8732,
+      "step": 22185
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.6855111111111115e-05,
+      "loss": 1.9577,
+      "step": 22190
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.685214814814815e-05,
+      "loss": 1.9193,
+      "step": 22195
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6849185185185185e-05,
+      "loss": 2.0029,
+      "step": 22200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6846222222222224e-05,
+      "loss": 1.9735,
+      "step": 22205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.684325925925926e-05,
+      "loss": 1.999,
+      "step": 22210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.68402962962963e-05,
+      "loss": 2.0577,
+      "step": 22215
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6837333333333334e-05,
+      "loss": 2.0269,
+      "step": 22220
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6834962962962964e-05,
+      "loss": 2.0091,
+      "step": 22225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.6832e-05,
+      "loss": 2.0179,
+      "step": 22230
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.682903703703704e-05,
+      "loss": 1.9184,
+      "step": 22235
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.682607407407407e-05,
+      "loss": 1.9849,
+      "step": 22240
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.682311111111111e-05,
+      "loss": 2.0387,
+      "step": 22245
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.682014814814815e-05,
+      "loss": 1.9514,
+      "step": 22250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.681718518518519e-05,
+      "loss": 1.9759,
+      "step": 22255
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.681422222222222e-05,
+      "loss": 2.0762,
+      "step": 22260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.681125925925926e-05,
+      "loss": 1.9457,
+      "step": 22265
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.68082962962963e-05,
+      "loss": 2.1658,
+      "step": 22270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.680533333333334e-05,
+      "loss": 2.0151,
+      "step": 22275
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.680237037037037e-05,
+      "loss": 2.0719,
+      "step": 22280
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.679940740740741e-05,
+      "loss": 1.9233,
+      "step": 22285
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.679644444444444e-05,
+      "loss": 1.894,
+      "step": 22290
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.679348148148149e-05,
+      "loss": 1.9605,
+      "step": 22295
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.679051851851852e-05,
+      "loss": 2.0719,
+      "step": 22300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.678755555555556e-05,
+      "loss": 1.9266,
+      "step": 22305
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.678459259259259e-05,
+      "loss": 1.9575,
+      "step": 22310
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6781629629629635e-05,
+      "loss": 1.912,
+      "step": 22315
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.677866666666667e-05,
+      "loss": 1.8704,
+      "step": 22320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6775703703703706e-05,
+      "loss": 1.914,
+      "step": 22325
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.677274074074074e-05,
+      "loss": 1.8973,
+      "step": 22330
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6769777777777784e-05,
+      "loss": 2.0735,
+      "step": 22335
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6766814814814816e-05,
+      "loss": 1.9187,
+      "step": 22340
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.6763851851851855e-05,
+      "loss": 2.0702,
+      "step": 22345
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.676088888888889e-05,
+      "loss": 1.928,
+      "step": 22350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.675792592592593e-05,
+      "loss": 1.7717,
+      "step": 22355
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6754962962962964e-05,
+      "loss": 2.1978,
+      "step": 22360
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6752e-05,
+      "loss": 2.0037,
+      "step": 22365
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6749037037037035e-05,
+      "loss": 1.9325,
+      "step": 22370
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6746074074074074e-05,
+      "loss": 2.0281,
+      "step": 22375
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.674311111111111e-05,
+      "loss": 2.0673,
+      "step": 22380
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.674014814814815e-05,
+      "loss": 1.8333,
+      "step": 22385
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.6737185185185184e-05,
+      "loss": 1.9408,
+      "step": 22390
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.673422222222222e-05,
+      "loss": 1.9469,
+      "step": 22395
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.673125925925926e-05,
+      "loss": 2.0528,
+      "step": 22400
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.67282962962963e-05,
+      "loss": 1.9336,
+      "step": 22405
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.672533333333333e-05,
+      "loss": 1.9752,
+      "step": 22410
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.672237037037037e-05,
+      "loss": 2.0135,
+      "step": 22415
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.671940740740741e-05,
+      "loss": 2.026,
+      "step": 22420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.671644444444445e-05,
+      "loss": 2.0687,
+      "step": 22425
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.671348148148148e-05,
+      "loss": 2.016,
+      "step": 22430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.671051851851852e-05,
+      "loss": 2.0342,
+      "step": 22435
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.670755555555556e-05,
+      "loss": 1.8784,
+      "step": 22440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.67045925925926e-05,
+      "loss": 1.8306,
+      "step": 22445
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.670162962962963e-05,
+      "loss": 1.9235,
+      "step": 22450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.669866666666667e-05,
+      "loss": 2.011,
+      "step": 22455
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.669570370370371e-05,
+      "loss": 1.9513,
+      "step": 22460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.6692740740740746e-05,
+      "loss": 2.0292,
+      "step": 22465
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.668977777777778e-05,
+      "loss": 1.7799,
+      "step": 22470
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.668681481481482e-05,
+      "loss": 1.912,
+      "step": 22475
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6683851851851855e-05,
+      "loss": 1.9549,
+      "step": 22480
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6680888888888894e-05,
+      "loss": 2.0827,
+      "step": 22485
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6677925925925926e-05,
+      "loss": 1.9584,
+      "step": 22490
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6674962962962965e-05,
+      "loss": 1.946,
+      "step": 22495
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6672000000000004e-05,
+      "loss": 1.9209,
+      "step": 22500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.666903703703704e-05,
+      "loss": 1.7846,
+      "step": 22505
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6666074074074075e-05,
+      "loss": 1.9142,
+      "step": 22510
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6663111111111114e-05,
+      "loss": 1.6891,
+      "step": 22515
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.6660148148148146e-05,
+      "loss": 1.7633,
+      "step": 22520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.665718518518519e-05,
+      "loss": 1.7224,
+      "step": 22525
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.665422222222222e-05,
+      "loss": 1.8495,
+      "step": 22530
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.665125925925926e-05,
+      "loss": 1.7849,
+      "step": 22535
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.6648296296296294e-05,
+      "loss": 1.784,
+      "step": 22540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.664533333333334e-05,
+      "loss": 1.8489,
+      "step": 22545
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.664237037037037e-05,
+      "loss": 1.8199,
+      "step": 22550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.663940740740741e-05,
+      "loss": 1.8022,
+      "step": 22555
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.663644444444444e-05,
+      "loss": 1.8269,
+      "step": 22560
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.663348148148149e-05,
+      "loss": 1.875,
+      "step": 22565
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.663051851851852e-05,
+      "loss": 1.8675,
+      "step": 22570
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.662755555555556e-05,
+      "loss": 1.7677,
+      "step": 22575
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.662459259259259e-05,
+      "loss": 1.6193,
+      "step": 22580
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.662162962962964e-05,
+      "loss": 1.7538,
+      "step": 22585
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.661866666666667e-05,
+      "loss": 1.9258,
+      "step": 22590
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.661570370370371e-05,
+      "loss": 1.8371,
+      "step": 22595
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.661274074074074e-05,
+      "loss": 1.7307,
+      "step": 22600
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.660977777777778e-05,
+      "loss": 1.8381,
+      "step": 22605
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.660681481481482e-05,
+      "loss": 1.777,
+      "step": 22610
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.6603851851851856e-05,
+      "loss": 1.8284,
+      "step": 22615
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.660088888888889e-05,
+      "loss": 1.7321,
+      "step": 22620
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.659792592592593e-05,
+      "loss": 1.7577,
+      "step": 22625
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.6594962962962966e-05,
+      "loss": 1.8464,
+      "step": 22630
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.6592000000000005e-05,
+      "loss": 1.8259,
+      "step": 22635
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.658903703703704e-05,
+      "loss": 1.7938,
+      "step": 22640
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6586074074074076e-05,
+      "loss": 1.8788,
+      "step": 22645
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6583111111111114e-05,
+      "loss": 1.8338,
+      "step": 22650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.658014814814815e-05,
+      "loss": 1.8972,
+      "step": 22655
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6577185185185185e-05,
+      "loss": 1.6231,
+      "step": 22660
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6574222222222224e-05,
+      "loss": 1.9287,
+      "step": 22665
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.657125925925926e-05,
+      "loss": 1.8256,
+      "step": 22670
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.65682962962963e-05,
+      "loss": 1.7844,
+      "step": 22675
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.6565333333333334e-05,
+      "loss": 1.8454,
+      "step": 22680
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.656237037037037e-05,
+      "loss": 1.806,
+      "step": 22685
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.655940740740741e-05,
+      "loss": 1.7871,
+      "step": 22690
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.655644444444445e-05,
+      "loss": 1.7787,
+      "step": 22695
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.655348148148148e-05,
+      "loss": 1.8291,
+      "step": 22700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.655051851851852e-05,
+      "loss": 1.8615,
+      "step": 22705
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.654755555555556e-05,
+      "loss": 1.8746,
+      "step": 22710
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.65445925925926e-05,
+      "loss": 1.664,
+      "step": 22715
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.654162962962963e-05,
+      "loss": 1.87,
+      "step": 22720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.653866666666667e-05,
+      "loss": 1.9176,
+      "step": 22725
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.653570370370371e-05,
+      "loss": 1.8991,
+      "step": 22730
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.653274074074075e-05,
+      "loss": 1.8384,
+      "step": 22735
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.652977777777778e-05,
+      "loss": 1.8157,
+      "step": 22740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.652681481481482e-05,
+      "loss": 1.8445,
+      "step": 22745
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.652385185185185e-05,
+      "loss": 1.8351,
+      "step": 22750
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6520888888888896e-05,
+      "loss": 1.7864,
+      "step": 22755
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.651792592592593e-05,
+      "loss": 1.7783,
+      "step": 22760
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.651496296296297e-05,
+      "loss": 1.8957,
+      "step": 22765
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6512e-05,
+      "loss": 1.9294,
+      "step": 22770
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.650903703703704e-05,
+      "loss": 1.8531,
+      "step": 22775
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6506074074074076e-05,
+      "loss": 1.9054,
+      "step": 22780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.650311111111111e-05,
+      "loss": 1.7755,
+      "step": 22785
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.650014814814815e-05,
+      "loss": 1.8592,
+      "step": 22790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6497185185185186e-05,
+      "loss": 1.8501,
+      "step": 22795
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.6494222222222225e-05,
+      "loss": 1.7843,
+      "step": 22800
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.649125925925926e-05,
+      "loss": 1.7665,
+      "step": 22805
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6488296296296296e-05,
+      "loss": 1.9824,
+      "step": 22810
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6485333333333335e-05,
+      "loss": 1.8467,
+      "step": 22815
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6482370370370373e-05,
+      "loss": 1.8848,
+      "step": 22820
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6479407407407405e-05,
+      "loss": 1.7562,
+      "step": 22825
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6476444444444444e-05,
+      "loss": 1.934,
+      "step": 22830
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.647348148148148e-05,
+      "loss": 1.8489,
+      "step": 22835
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.647051851851852e-05,
+      "loss": 1.8435,
+      "step": 22840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.6467555555555554e-05,
+      "loss": 1.6824,
+      "step": 22845
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.646459259259259e-05,
+      "loss": 1.8905,
+      "step": 22850
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.646162962962963e-05,
+      "loss": 1.812,
+      "step": 22855
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.645866666666667e-05,
+      "loss": 1.8359,
+      "step": 22860
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.64557037037037e-05,
+      "loss": 1.8117,
+      "step": 22865
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.645274074074074e-05,
+      "loss": 1.7905,
+      "step": 22870
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.644977777777778e-05,
+      "loss": 1.9334,
+      "step": 22875
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.644681481481482e-05,
+      "loss": 1.7589,
+      "step": 22880
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.644385185185185e-05,
+      "loss": 1.9271,
+      "step": 22885
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.644088888888889e-05,
+      "loss": 1.7317,
+      "step": 22890
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.643792592592593e-05,
+      "loss": 1.8193,
+      "step": 22895
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.643496296296297e-05,
+      "loss": 1.8843,
+      "step": 22900
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.6432e-05,
+      "loss": 1.8023,
+      "step": 22905
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.642903703703704e-05,
+      "loss": 1.837,
+      "step": 22910
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.642607407407407e-05,
+      "loss": 1.7608,
+      "step": 22915
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.6423111111111116e-05,
+      "loss": 1.9327,
+      "step": 22920
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.642014814814815e-05,
+      "loss": 1.8393,
+      "step": 22925
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.641718518518519e-05,
+      "loss": 1.8862,
+      "step": 22930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.641422222222222e-05,
+      "loss": 1.7127,
+      "step": 22935
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6411259259259264e-05,
+      "loss": 1.9046,
+      "step": 22940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6408296296296297e-05,
+      "loss": 1.8927,
+      "step": 22945
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6405333333333335e-05,
+      "loss": 1.9441,
+      "step": 22950
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.640237037037037e-05,
+      "loss": 1.8953,
+      "step": 22955
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.639940740740741e-05,
+      "loss": 1.7568,
+      "step": 22960
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6396444444444445e-05,
+      "loss": 1.7743,
+      "step": 22965
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6393481481481484e-05,
+      "loss": 1.8492,
+      "step": 22970
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6390518518518516e-05,
+      "loss": 1.8732,
+      "step": 22975
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6387555555555555e-05,
+      "loss": 1.8879,
+      "step": 22980
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6384592592592594e-05,
+      "loss": 1.7628,
+      "step": 22985
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.638162962962963e-05,
+      "loss": 1.8158,
+      "step": 22990
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6378666666666664e-05,
+      "loss": 1.8153,
+      "step": 22995
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.63757037037037e-05,
+      "loss": 1.8163,
+      "step": 23000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.637274074074074e-05,
+      "loss": 1.8222,
+      "step": 23005
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.636977777777778e-05,
+      "loss": 1.9337,
+      "step": 23010
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.636681481481481e-05,
+      "loss": 1.9534,
+      "step": 23015
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.636385185185185e-05,
+      "loss": 1.8636,
+      "step": 23020
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.636088888888889e-05,
+      "loss": 1.8016,
+      "step": 23025
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.635792592592593e-05,
+      "loss": 1.8423,
+      "step": 23030
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.635496296296296e-05,
+      "loss": 1.6686,
+      "step": 23035
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6352e-05,
+      "loss": 1.7344,
+      "step": 23040
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.634903703703704e-05,
+      "loss": 1.9504,
+      "step": 23045
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.634607407407408e-05,
+      "loss": 1.8448,
+      "step": 23050
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.634311111111111e-05,
+      "loss": 1.8852,
+      "step": 23055
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.634014814814815e-05,
+      "loss": 1.8572,
+      "step": 23060
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.633718518518519e-05,
+      "loss": 2.0033,
+      "step": 23065
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6334222222222226e-05,
+      "loss": 1.8127,
+      "step": 23070
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.633125925925926e-05,
+      "loss": 1.8652,
+      "step": 23075
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.63282962962963e-05,
+      "loss": 1.926,
+      "step": 23080
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6325333333333336e-05,
+      "loss": 1.8205,
+      "step": 23085
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6322370370370375e-05,
+      "loss": 1.9406,
+      "step": 23090
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.631940740740741e-05,
+      "loss": 1.7483,
+      "step": 23095
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6316444444444446e-05,
+      "loss": 1.7942,
+      "step": 23100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6313481481481485e-05,
+      "loss": 1.9152,
+      "step": 23105
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6310518518518523e-05,
+      "loss": 1.7856,
+      "step": 23110
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6307555555555555e-05,
+      "loss": 1.8829,
+      "step": 23115
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6304592592592594e-05,
+      "loss": 1.9182,
+      "step": 23120
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.630162962962963e-05,
+      "loss": 1.9467,
+      "step": 23125
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.629866666666667e-05,
+      "loss": 1.8157,
+      "step": 23130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6295703703703704e-05,
+      "loss": 1.9435,
+      "step": 23135
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.629274074074074e-05,
+      "loss": 1.9298,
+      "step": 23140
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.6289777777777775e-05,
+      "loss": 1.82,
+      "step": 23145
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.628681481481482e-05,
+      "loss": 1.7396,
+      "step": 23150
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.628385185185185e-05,
+      "loss": 1.8248,
+      "step": 23155
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.628088888888889e-05,
+      "loss": 1.8267,
+      "step": 23160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.627792592592592e-05,
+      "loss": 1.8337,
+      "step": 23165
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.627496296296297e-05,
+      "loss": 1.9368,
+      "step": 23170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.6272e-05,
+      "loss": 1.9274,
+      "step": 23175
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.626903703703704e-05,
+      "loss": 1.8147,
+      "step": 23180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.626607407407407e-05,
+      "loss": 1.8806,
+      "step": 23185
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.626311111111112e-05,
+      "loss": 1.8838,
+      "step": 23190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.626014814814815e-05,
+      "loss": 1.8962,
+      "step": 23195
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.625718518518519e-05,
+      "loss": 1.9163,
+      "step": 23200
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.625422222222222e-05,
+      "loss": 1.8,
+      "step": 23205
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.625125925925926e-05,
+      "loss": 1.8729,
+      "step": 23210
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.62482962962963e-05,
+      "loss": 1.7963,
+      "step": 23215
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.624533333333334e-05,
+      "loss": 1.9343,
+      "step": 23220
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.624237037037037e-05,
+      "loss": 1.8489,
+      "step": 23225
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.623940740740741e-05,
+      "loss": 1.8235,
+      "step": 23230
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6236444444444447e-05,
+      "loss": 2.0098,
+      "step": 23235
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6233481481481485e-05,
+      "loss": 1.9015,
+      "step": 23240
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.623051851851852e-05,
+      "loss": 1.9523,
+      "step": 23245
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6227555555555556e-05,
+      "loss": 1.8714,
+      "step": 23250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.6224592592592595e-05,
+      "loss": 1.788,
+      "step": 23255
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6221629629629634e-05,
+      "loss": 1.8974,
+      "step": 23260
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6218666666666666e-05,
+      "loss": 1.8034,
+      "step": 23265
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6215703703703705e-05,
+      "loss": 1.7945,
+      "step": 23270
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6212740740740744e-05,
+      "loss": 1.9759,
+      "step": 23275
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.620977777777778e-05,
+      "loss": 1.8251,
+      "step": 23280
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6206814814814814e-05,
+      "loss": 1.754,
+      "step": 23285
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.620385185185185e-05,
+      "loss": 1.8727,
+      "step": 23290
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.620088888888889e-05,
+      "loss": 1.8385,
+      "step": 23295
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.619792592592593e-05,
+      "loss": 1.9065,
+      "step": 23300
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.619496296296296e-05,
+      "loss": 1.9687,
+      "step": 23305
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.6192e-05,
+      "loss": 1.8316,
+      "step": 23310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.618903703703704e-05,
+      "loss": 1.9082,
+      "step": 23315
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.618607407407408e-05,
+      "loss": 1.6368,
+      "step": 23320
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.618311111111111e-05,
+      "loss": 1.7399,
+      "step": 23325
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.618014814814815e-05,
+      "loss": 1.8213,
+      "step": 23330
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.617718518518519e-05,
+      "loss": 1.7772,
+      "step": 23335
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.617422222222223e-05,
+      "loss": 1.8351,
+      "step": 23340
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.617125925925926e-05,
+      "loss": 1.8276,
+      "step": 23345
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.61682962962963e-05,
+      "loss": 1.7526,
+      "step": 23350
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.616533333333334e-05,
+      "loss": 1.905,
+      "step": 23355
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6162370370370376e-05,
+      "loss": 1.9396,
+      "step": 23360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.615940740740741e-05,
+      "loss": 1.8504,
+      "step": 23365
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.615644444444445e-05,
+      "loss": 1.7356,
+      "step": 23370
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.615348148148148e-05,
+      "loss": 1.8353,
+      "step": 23375
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6150518518518525e-05,
+      "loss": 1.7861,
+      "step": 23380
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.614755555555556e-05,
+      "loss": 1.7629,
+      "step": 23385
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6144592592592596e-05,
+      "loss": 2.0237,
+      "step": 23390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.614162962962963e-05,
+      "loss": 1.8321,
+      "step": 23395
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6138666666666673e-05,
+      "loss": 1.9313,
+      "step": 23400
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6135703703703706e-05,
+      "loss": 1.9016,
+      "step": 23405
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6132740740740744e-05,
+      "loss": 1.6967,
+      "step": 23410
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6129777777777776e-05,
+      "loss": 1.7836,
+      "step": 23415
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.612681481481482e-05,
+      "loss": 1.8029,
+      "step": 23420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6123851851851854e-05,
+      "loss": 1.8984,
+      "step": 23425
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.612088888888889e-05,
+      "loss": 1.7787,
+      "step": 23430
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.6117925925925925e-05,
+      "loss": 1.6863,
+      "step": 23435
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.6114962962962964e-05,
+      "loss": 1.8302,
+      "step": 23440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.6112e-05,
+      "loss": 1.8805,
+      "step": 23445
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.610903703703704e-05,
+      "loss": 1.6919,
+      "step": 23450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.6106074074074073e-05,
+      "loss": 1.8433,
+      "step": 23455
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.610311111111111e-05,
+      "loss": 1.8331,
+      "step": 23460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.610014814814815e-05,
+      "loss": 1.8952,
+      "step": 23465
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.609718518518519e-05,
+      "loss": 1.861,
+      "step": 23470
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.609422222222222e-05,
+      "loss": 1.9349,
+      "step": 23475
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.609125925925926e-05,
+      "loss": 1.8969,
+      "step": 23480
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.60882962962963e-05,
+      "loss": 1.9112,
+      "step": 23485
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.608533333333334e-05,
+      "loss": 1.9284,
+      "step": 23490
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.608237037037037e-05,
+      "loss": 1.8119,
+      "step": 23495
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.607940740740741e-05,
+      "loss": 1.899,
+      "step": 23500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.607644444444445e-05,
+      "loss": 1.796,
+      "step": 23505
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.607348148148149e-05,
+      "loss": 1.8193,
+      "step": 23510
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.607051851851852e-05,
+      "loss": 1.9024,
+      "step": 23515
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.606755555555556e-05,
+      "loss": 1.8121,
+      "step": 23520
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.6064592592592597e-05,
+      "loss": 1.8408,
+      "step": 23525
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.6061629629629635e-05,
+      "loss": 1.9334,
+      "step": 23530
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.605866666666667e-05,
+      "loss": 1.879,
+      "step": 23535
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.6055703703703706e-05,
+      "loss": 1.9985,
+      "step": 23540
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6052740740740745e-05,
+      "loss": 1.9182,
+      "step": 23545
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.604977777777778e-05,
+      "loss": 1.9913,
+      "step": 23550
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6046814814814816e-05,
+      "loss": 1.8068,
+      "step": 23555
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.604385185185185e-05,
+      "loss": 1.7911,
+      "step": 23560
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6040888888888894e-05,
+      "loss": 1.7813,
+      "step": 23565
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6037925925925926e-05,
+      "loss": 1.8827,
+      "step": 23570
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6034962962962964e-05,
+      "loss": 1.7944,
+      "step": 23575
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6031999999999997e-05,
+      "loss": 1.8562,
+      "step": 23580
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.602903703703704e-05,
+      "loss": 1.8451,
+      "step": 23585
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.6026074074074074e-05,
+      "loss": 1.8021,
+      "step": 23590
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.602311111111111e-05,
+      "loss": 1.866,
+      "step": 23595
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.6020148148148145e-05,
+      "loss": 1.934,
+      "step": 23600
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.6017185185185184e-05,
+      "loss": 1.9609,
+      "step": 23605
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.601422222222222e-05,
+      "loss": 1.8684,
+      "step": 23610
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.601125925925926e-05,
+      "loss": 1.9892,
+      "step": 23615
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.6008296296296294e-05,
+      "loss": 1.8274,
+      "step": 23620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.600533333333333e-05,
+      "loss": 1.7849,
+      "step": 23625
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.600237037037037e-05,
+      "loss": 1.872,
+      "step": 23630
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.599940740740741e-05,
+      "loss": 1.9701,
+      "step": 23635
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.599644444444444e-05,
+      "loss": 1.8229,
+      "step": 23640
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.599348148148148e-05,
+      "loss": 1.8587,
+      "step": 23645
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.599051851851852e-05,
+      "loss": 1.8681,
+      "step": 23650
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.598755555555556e-05,
+      "loss": 1.8745,
+      "step": 23655
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.598459259259259e-05,
+      "loss": 1.8708,
+      "step": 23660
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.598162962962963e-05,
+      "loss": 1.7522,
+      "step": 23665
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.597866666666667e-05,
+      "loss": 1.8429,
+      "step": 23670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.597570370370371e-05,
+      "loss": 1.8165,
+      "step": 23675
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.597274074074074e-05,
+      "loss": 1.9381,
+      "step": 23680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.596977777777778e-05,
+      "loss": 1.7509,
+      "step": 23685
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.596681481481482e-05,
+      "loss": 1.8546,
+      "step": 23690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.5963851851851856e-05,
+      "loss": 1.9515,
+      "step": 23695
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.596088888888889e-05,
+      "loss": 1.7258,
+      "step": 23700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.5957925925925926e-05,
+      "loss": 1.8776,
+      "step": 23705
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5954962962962965e-05,
+      "loss": 1.9044,
+      "step": 23710
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5952000000000004e-05,
+      "loss": 1.8833,
+      "step": 23715
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5949037037037036e-05,
+      "loss": 1.8041,
+      "step": 23720
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5946074074074075e-05,
+      "loss": 1.8098,
+      "step": 23725
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5943111111111114e-05,
+      "loss": 1.8424,
+      "step": 23730
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.594014814814815e-05,
+      "loss": 1.838,
+      "step": 23735
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5937185185185185e-05,
+      "loss": 1.8764,
+      "step": 23740
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.5934222222222223e-05,
+      "loss": 1.8344,
+      "step": 23745
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.593125925925926e-05,
+      "loss": 1.8099,
+      "step": 23750
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.59282962962963e-05,
+      "loss": 1.9737,
+      "step": 23755
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.592533333333333e-05,
+      "loss": 1.9193,
+      "step": 23760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.592237037037037e-05,
+      "loss": 1.7806,
+      "step": 23765
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.5919407407407404e-05,
+      "loss": 1.8442,
+      "step": 23770
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.591644444444445e-05,
+      "loss": 1.8861,
+      "step": 23775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.591348148148148e-05,
+      "loss": 1.8891,
+      "step": 23780
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.591051851851852e-05,
+      "loss": 1.9153,
+      "step": 23785
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.590755555555555e-05,
+      "loss": 1.8506,
+      "step": 23790
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.59045925925926e-05,
+      "loss": 1.8424,
+      "step": 23795
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.590162962962963e-05,
+      "loss": 1.9049,
+      "step": 23800
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.589866666666667e-05,
+      "loss": 1.873,
+      "step": 23805
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.58957037037037e-05,
+      "loss": 1.9368,
+      "step": 23810
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.589274074074075e-05,
+      "loss": 1.7873,
+      "step": 23815
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.588977777777778e-05,
+      "loss": 1.7702,
+      "step": 23820
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.588681481481482e-05,
+      "loss": 1.7825,
+      "step": 23825
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.588385185185185e-05,
+      "loss": 1.9163,
+      "step": 23830
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.588088888888889e-05,
+      "loss": 1.7773,
+      "step": 23835
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.587792592592593e-05,
+      "loss": 1.8207,
+      "step": 23840
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5874962962962966e-05,
+      "loss": 1.7184,
+      "step": 23845
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5872e-05,
+      "loss": 1.7885,
+      "step": 23850
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.586903703703704e-05,
+      "loss": 1.9153,
+      "step": 23855
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5866074074074076e-05,
+      "loss": 1.8651,
+      "step": 23860
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5863111111111115e-05,
+      "loss": 1.8613,
+      "step": 23865
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5860148148148147e-05,
+      "loss": 1.9139,
+      "step": 23870
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.5857185185185185e-05,
+      "loss": 1.7422,
+      "step": 23875
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.5854222222222224e-05,
+      "loss": 1.8131,
+      "step": 23880
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.585125925925926e-05,
+      "loss": 1.7602,
+      "step": 23885
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.5848296296296295e-05,
+      "loss": 1.7918,
+      "step": 23890
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.5845333333333334e-05,
+      "loss": 1.7561,
+      "step": 23895
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.584237037037037e-05,
+      "loss": 1.9091,
+      "step": 23900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.583940740740741e-05,
+      "loss": 1.8036,
+      "step": 23905
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.5836444444444444e-05,
+      "loss": 1.7554,
+      "step": 23910
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.583348148148148e-05,
+      "loss": 1.7602,
+      "step": 23915
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.583051851851852e-05,
+      "loss": 1.7864,
+      "step": 23920
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.582755555555556e-05,
+      "loss": 1.9324,
+      "step": 23925
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.582459259259259e-05,
+      "loss": 1.8254,
+      "step": 23930
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.582162962962963e-05,
+      "loss": 1.8523,
+      "step": 23935
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.581866666666667e-05,
+      "loss": 1.7055,
+      "step": 23940
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.581570370370371e-05,
+      "loss": 1.8094,
+      "step": 23945
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.581274074074074e-05,
+      "loss": 1.776,
+      "step": 23950
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.580977777777778e-05,
+      "loss": 1.7461,
+      "step": 23955
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.580681481481482e-05,
+      "loss": 1.9923,
+      "step": 23960
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.580385185185186e-05,
+      "loss": 1.8857,
+      "step": 23965
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.580088888888889e-05,
+      "loss": 1.7751,
+      "step": 23970
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.579792592592593e-05,
+      "loss": 1.8931,
+      "step": 23975
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.579496296296297e-05,
+      "loss": 1.839,
+      "step": 23980
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.5792000000000006e-05,
+      "loss": 1.9552,
+      "step": 23985
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 3.578903703703704e-05,
+      "loss": 1.7468,
+      "step": 23990
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5786074074074076e-05,
+      "loss": 1.8894,
+      "step": 23995
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.578311111111111e-05,
+      "loss": 1.9389,
+      "step": 24000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5780148148148154e-05,
+      "loss": 1.7804,
+      "step": 24005
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5777185185185186e-05,
+      "loss": 1.927,
+      "step": 24010
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5774222222222225e-05,
+      "loss": 1.8599,
+      "step": 24015
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.577125925925926e-05,
+      "loss": 1.7356,
+      "step": 24020
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.57682962962963e-05,
+      "loss": 1.8256,
+      "step": 24025
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5765333333333335e-05,
+      "loss": 1.9001,
+      "step": 24030
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5762370370370374e-05,
+      "loss": 1.9931,
+      "step": 24035
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.5759407407407406e-05,
+      "loss": 1.721,
+      "step": 24040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.575644444444445e-05,
+      "loss": 1.8939,
+      "step": 24045
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.575348148148148e-05,
+      "loss": 1.8057,
+      "step": 24050
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.575051851851852e-05,
+      "loss": 1.8283,
+      "step": 24055
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.5747555555555554e-05,
+      "loss": 1.7916,
+      "step": 24060
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.574459259259259e-05,
+      "loss": 1.8491,
+      "step": 24065
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.574162962962963e-05,
+      "loss": 1.7988,
+      "step": 24070
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.573866666666667e-05,
+      "loss": 2.0095,
+      "step": 24075
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.57357037037037e-05,
+      "loss": 2.0253,
+      "step": 24080
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.573274074074074e-05,
+      "loss": 1.8337,
+      "step": 24085
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.572977777777778e-05,
+      "loss": 1.9918,
+      "step": 24090
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.572681481481482e-05,
+      "loss": 1.9256,
+      "step": 24095
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 3.572385185185185e-05,
+      "loss": 1.9391,
+      "step": 24100
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.572088888888889e-05,
+      "loss": 1.9269,
+      "step": 24105
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.571792592592593e-05,
+      "loss": 1.9112,
+      "step": 24110
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.571496296296297e-05,
+      "loss": 1.9586,
+      "step": 24115
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5712e-05,
+      "loss": 1.7859,
+      "step": 24120
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.570903703703704e-05,
+      "loss": 1.7673,
+      "step": 24125
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.570607407407408e-05,
+      "loss": 1.8856,
+      "step": 24130
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5703111111111116e-05,
+      "loss": 1.9015,
+      "step": 24135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.570014814814815e-05,
+      "loss": 1.8433,
+      "step": 24140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.569718518518519e-05,
+      "loss": 1.9985,
+      "step": 24145
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5694222222222226e-05,
+      "loss": 1.9534,
+      "step": 24150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.5691259259259265e-05,
+      "loss": 1.8793,
+      "step": 24155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.56882962962963e-05,
+      "loss": 1.7488,
+      "step": 24160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5685333333333335e-05,
+      "loss": 1.8579,
+      "step": 24165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5682370370370374e-05,
+      "loss": 1.8537,
+      "step": 24170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.567940740740741e-05,
+      "loss": 1.807,
+      "step": 24175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5676444444444445e-05,
+      "loss": 1.6736,
+      "step": 24180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5673481481481484e-05,
+      "loss": 1.9001,
+      "step": 24185
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.567051851851852e-05,
+      "loss": 1.8104,
+      "step": 24190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.566755555555556e-05,
+      "loss": 1.9599,
+      "step": 24195
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.5664592592592594e-05,
+      "loss": 1.8312,
+      "step": 24200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.566162962962963e-05,
+      "loss": 1.8425,
+      "step": 24205
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.565866666666667e-05,
+      "loss": 1.853,
+      "step": 24210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 3.565570370370371e-05,
+      "loss": 1.7619,
+      "step": 24215
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.565274074074074e-05,
+      "loss": 1.9584,
+      "step": 24220
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.564977777777778e-05,
+      "loss": 1.8885,
+      "step": 24225
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.564681481481481e-05,
+      "loss": 1.8371,
+      "step": 24230
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.564385185185186e-05,
+      "loss": 1.9163,
+      "step": 24235
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.564088888888889e-05,
+      "loss": 1.8696,
+      "step": 24240
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.563792592592593e-05,
+      "loss": 1.7983,
+      "step": 24245
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.563496296296296e-05,
+      "loss": 1.8355,
+      "step": 24250
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.563200000000001e-05,
+      "loss": 1.8772,
+      "step": 24255
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.562903703703704e-05,
+      "loss": 1.8363,
+      "step": 24260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.562607407407408e-05,
+      "loss": 1.8263,
+      "step": 24265
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 3.562311111111111e-05,
+      "loss": 1.9177,
+      "step": 24270
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5620148148148156e-05,
+      "loss": 1.856,
+      "step": 24275
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.561718518518519e-05,
+      "loss": 1.7943,
+      "step": 24280
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5614222222222227e-05,
+      "loss": 1.8397,
+      "step": 24285
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.561125925925926e-05,
+      "loss": 1.7794,
+      "step": 24290
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.56082962962963e-05,
+      "loss": 1.9025,
+      "step": 24295
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5605333333333336e-05,
+      "loss": 1.9026,
+      "step": 24300
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5602370370370375e-05,
+      "loss": 1.837,
+      "step": 24305
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.559940740740741e-05,
+      "loss": 1.733,
+      "step": 24310
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5596444444444446e-05,
+      "loss": 1.889,
+      "step": 24315
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.5593481481481485e-05,
+      "loss": 1.8402,
+      "step": 24320
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.559051851851852e-05,
+      "loss": 1.9087,
+      "step": 24325
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5587555555555556e-05,
+      "loss": 1.8417,
+      "step": 24330
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5584592592592594e-05,
+      "loss": 1.9214,
+      "step": 24335
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.558162962962963e-05,
+      "loss": 1.9137,
+      "step": 24340
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5578666666666665e-05,
+      "loss": 1.9687,
+      "step": 24345
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5575703703703704e-05,
+      "loss": 1.8883,
+      "step": 24350
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.557274074074074e-05,
+      "loss": 1.8356,
+      "step": 24355
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.556977777777778e-05,
+      "loss": 1.9336,
+      "step": 24360
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5566814814814814e-05,
+      "loss": 1.8689,
+      "step": 24365
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.556385185185185e-05,
+      "loss": 1.809,
+      "step": 24370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.5560888888888885e-05,
+      "loss": 1.8831,
+      "step": 24375
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 3.555792592592593e-05,
+      "loss": 1.8179,
+      "step": 24380
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.555496296296296e-05,
+      "loss": 1.7819,
+      "step": 24385
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.5552e-05,
+      "loss": 1.8614,
+      "step": 24390
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.554903703703703e-05,
+      "loss": 1.8621,
+      "step": 24395
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.554607407407408e-05,
+      "loss": 1.8766,
+      "step": 24400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.554311111111111e-05,
+      "loss": 2.0427,
+      "step": 24405
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.554014814814815e-05,
+      "loss": 1.8638,
+      "step": 24410
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.553777777777778e-05,
+      "loss": 1.8285,
+      "step": 24415
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.553481481481482e-05,
+      "loss": 1.9399,
+      "step": 24420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.553185185185186e-05,
+      "loss": 1.8146,
+      "step": 24425
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.552888888888889e-05,
+      "loss": 1.9894,
+      "step": 24430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.552592592592593e-05,
+      "loss": 1.727,
+      "step": 24435
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.552296296296297e-05,
+      "loss": 1.8678,
+      "step": 24440
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 1.7672,
+      "step": 24445
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.551703703703704e-05,
+      "loss": 1.8745,
+      "step": 24450
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5514074074074076e-05,
+      "loss": 1.9035,
+      "step": 24455
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.551111111111111e-05,
+      "loss": 1.8315,
+      "step": 24460
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5508148148148154e-05,
+      "loss": 1.7804,
+      "step": 24465
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5505185185185186e-05,
+      "loss": 1.8774,
+      "step": 24470
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5502222222222225e-05,
+      "loss": 1.8165,
+      "step": 24475
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.549925925925926e-05,
+      "loss": 1.8997,
+      "step": 24480
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.54962962962963e-05,
+      "loss": 1.896,
+      "step": 24485
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5493333333333335e-05,
+      "loss": 1.9262,
+      "step": 24490
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 3.5490370370370373e-05,
+      "loss": 1.9757,
+      "step": 24495
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.5487407407407405e-05,
+      "loss": 1.8168,
+      "step": 24500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.548444444444445e-05,
+      "loss": 1.9046,
+      "step": 24505
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.548148148148148e-05,
+      "loss": 1.761,
+      "step": 24510
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.547851851851852e-05,
+      "loss": 1.944,
+      "step": 24515
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.5475555555555554e-05,
+      "loss": 1.9702,
+      "step": 24520
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.547259259259259e-05,
+      "loss": 1.7807,
+      "step": 24525
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.546962962962963e-05,
+      "loss": 1.8418,
+      "step": 24530
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.546666666666667e-05,
+      "loss": 1.7804,
+      "step": 24535
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.54637037037037e-05,
+      "loss": 1.8409,
+      "step": 24540
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.546074074074074e-05,
+      "loss": 1.8596,
+      "step": 24545
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.545777777777778e-05,
+      "loss": 1.8124,
+      "step": 24550
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.545481481481482e-05,
+      "loss": 1.8125,
+      "step": 24555
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.545185185185185e-05,
+      "loss": 1.8544,
+      "step": 24560
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544888888888889e-05,
+      "loss": 1.8986,
+      "step": 24565
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544592592592593e-05,
+      "loss": 1.8437,
+      "step": 24570
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544296296296297e-05,
+      "loss": 1.7703,
+      "step": 24575
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.544e-05,
+      "loss": 1.9501,
+      "step": 24580
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.543703703703704e-05,
+      "loss": 1.7988,
+      "step": 24585
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.543407407407408e-05,
+      "loss": 1.83,
+      "step": 24590
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.5431111111111116e-05,
+      "loss": 1.882,
+      "step": 24595
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.542814814814815e-05,
+      "loss": 1.7785,
+      "step": 24600
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 3.542518518518519e-05,
+      "loss": 1.862,
+      "step": 24605
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5422222222222226e-05,
+      "loss": 2.0215,
+      "step": 24610
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5419259259259264e-05,
+      "loss": 1.7936,
+      "step": 24615
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5416296296296297e-05,
+      "loss": 2.037,
+      "step": 24620
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5413333333333335e-05,
+      "loss": 1.841,
+      "step": 24625
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5410370370370374e-05,
+      "loss": 1.693,
+      "step": 24630
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.540740740740741e-05,
+      "loss": 1.851,
+      "step": 24635
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5404444444444445e-05,
+      "loss": 1.8352,
+      "step": 24640
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5401481481481484e-05,
+      "loss": 1.9591,
+      "step": 24645
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.539851851851852e-05,
+      "loss": 1.7345,
+      "step": 24650
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.539555555555556e-05,
+      "loss": 1.7922,
+      "step": 24655
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.5392592592592594e-05,
+      "loss": 1.8394,
+      "step": 24660
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.538962962962963e-05,
+      "loss": 1.7019,
+      "step": 24665
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.538666666666667e-05,
+      "loss": 1.8921,
+      "step": 24670
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.538370370370371e-05,
+      "loss": 1.9013,
+      "step": 24675
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.538074074074074e-05,
+      "loss": 1.8075,
+      "step": 24680
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.537777777777778e-05,
+      "loss": 1.9012,
+      "step": 24685
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.537481481481481e-05,
+      "loss": 1.9693,
+      "step": 24690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.537185185185186e-05,
+      "loss": 1.9659,
+      "step": 24695
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.536888888888889e-05,
+      "loss": 1.8073,
+      "step": 24700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.536592592592593e-05,
+      "loss": 1.8324,
+      "step": 24705
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.536296296296296e-05,
+      "loss": 1.7607,
+      "step": 24710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.536000000000001e-05,
+      "loss": 1.8187,
+      "step": 24715
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.535703703703704e-05,
+      "loss": 1.8858,
+      "step": 24720
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.535407407407408e-05,
+      "loss": 1.8929,
+      "step": 24725
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.535111111111111e-05,
+      "loss": 1.793,
+      "step": 24730
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5348148148148156e-05,
+      "loss": 1.8317,
+      "step": 24735
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.534518518518519e-05,
+      "loss": 1.8341,
+      "step": 24740
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5342222222222226e-05,
+      "loss": 1.9416,
+      "step": 24745
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.533925925925926e-05,
+      "loss": 1.7878,
+      "step": 24750
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.53362962962963e-05,
+      "loss": 1.8886,
+      "step": 24755
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5333333333333336e-05,
+      "loss": 1.863,
+      "step": 24760
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5330370370370375e-05,
+      "loss": 1.926,
+      "step": 24765
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.532740740740741e-05,
+      "loss": 1.9467,
+      "step": 24770
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.5324444444444446e-05,
+      "loss": 1.886,
+      "step": 24775
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5321481481481485e-05,
+      "loss": 1.9764,
+      "step": 24780
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.531851851851852e-05,
+      "loss": 1.7238,
+      "step": 24785
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5315555555555555e-05,
+      "loss": 1.7047,
+      "step": 24790
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5312592592592594e-05,
+      "loss": 1.8313,
+      "step": 24795
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.530962962962963e-05,
+      "loss": 1.835,
+      "step": 24800
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5306666666666665e-05,
+      "loss": 2.0512,
+      "step": 24805
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5303703703703704e-05,
+      "loss": 1.8739,
+      "step": 24810
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.530074074074074e-05,
+      "loss": 1.8078,
+      "step": 24815
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.529777777777778e-05,
+      "loss": 1.8331,
+      "step": 24820
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5294814814814814e-05,
+      "loss": 1.8031,
+      "step": 24825
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.529185185185185e-05,
+      "loss": 1.8622,
+      "step": 24830
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.528888888888889e-05,
+      "loss": 1.942,
+      "step": 24835
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.528592592592593e-05,
+      "loss": 1.8436,
+      "step": 24840
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.528296296296296e-05,
+      "loss": 1.9,
+      "step": 24845
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.528e-05,
+      "loss": 1.8748,
+      "step": 24850
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.527703703703703e-05,
+      "loss": 1.8249,
+      "step": 24855
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.527407407407408e-05,
+      "loss": 1.8156,
+      "step": 24860
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.527111111111111e-05,
+      "loss": 1.8637,
+      "step": 24865
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.526814814814815e-05,
+      "loss": 1.742,
+      "step": 24870
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.526518518518518e-05,
+      "loss": 1.8675,
+      "step": 24875
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.526222222222223e-05,
+      "loss": 1.8444,
+      "step": 24880
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.525925925925926e-05,
+      "loss": 1.8077,
+      "step": 24885
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.52562962962963e-05,
+      "loss": 1.8205,
+      "step": 24890
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.525333333333333e-05,
+      "loss": 1.8491,
+      "step": 24895
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5250370370370376e-05,
+      "loss": 1.8633,
+      "step": 24900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.524740740740741e-05,
+      "loss": 2.0354,
+      "step": 24905
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5244444444444447e-05,
+      "loss": 1.9558,
+      "step": 24910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.524148148148148e-05,
+      "loss": 1.9479,
+      "step": 24915
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.523851851851852e-05,
+      "loss": 1.9407,
+      "step": 24920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5235555555555556e-05,
+      "loss": 1.7419,
+      "step": 24925
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5232592592592595e-05,
+      "loss": 1.8207,
+      "step": 24930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.522962962962963e-05,
+      "loss": 1.8884,
+      "step": 24935
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5226666666666666e-05,
+      "loss": 1.8133,
+      "step": 24940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.5223703703703705e-05,
+      "loss": 1.9704,
+      "step": 24945
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.5220740740740744e-05,
+      "loss": 1.8793,
+      "step": 24950
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.5217777777777776e-05,
+      "loss": 1.8465,
+      "step": 24955
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.5214814814814814e-05,
+      "loss": 1.89,
+      "step": 24960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.521185185185185e-05,
+      "loss": 1.8096,
+      "step": 24965
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.520888888888889e-05,
+      "loss": 1.919,
+      "step": 24970
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.5205925925925924e-05,
+      "loss": 1.9718,
+      "step": 24975
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.520296296296296e-05,
+      "loss": 1.7884,
+      "step": 24980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.52e-05,
+      "loss": 1.7779,
+      "step": 24985
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.519703703703704e-05,
+      "loss": 1.8378,
+      "step": 24990
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.519407407407407e-05,
+      "loss": 1.8966,
+      "step": 24995
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.519111111111111e-05,
+      "loss": 1.8504,
+      "step": 25000
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.518814814814815e-05,
+      "loss": 1.9341,
+      "step": 25005
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 1.901,
+      "step": 25010
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.518222222222222e-05,
+      "loss": 1.9542,
+      "step": 25015
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.517925925925926e-05,
+      "loss": 1.9702,
+      "step": 25020
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.51762962962963e-05,
+      "loss": 1.8339,
+      "step": 25025
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.517333333333334e-05,
+      "loss": 1.9315,
+      "step": 25030
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.517037037037037e-05,
+      "loss": 1.7635,
+      "step": 25035
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.516740740740741e-05,
+      "loss": 1.8099,
+      "step": 25040
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.516444444444445e-05,
+      "loss": 1.8775,
+      "step": 25045
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.5161481481481486e-05,
+      "loss": 1.8339,
+      "step": 25050
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.515851851851852e-05,
+      "loss": 1.8749,
+      "step": 25055
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.515555555555556e-05,
+      "loss": 1.8664,
+      "step": 25060
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5152592592592596e-05,
+      "loss": 1.863,
+      "step": 25065
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5149629629629635e-05,
+      "loss": 1.9182,
+      "step": 25070
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.514666666666667e-05,
+      "loss": 1.9756,
+      "step": 25075
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5143703703703706e-05,
+      "loss": 1.8123,
+      "step": 25080
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.514074074074074e-05,
+      "loss": 1.9715,
+      "step": 25085
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.513777777777778e-05,
+      "loss": 1.945,
+      "step": 25090
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5134814814814815e-05,
+      "loss": 1.8166,
+      "step": 25095
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5131851851851854e-05,
+      "loss": 1.9568,
+      "step": 25100
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5128888888888886e-05,
+      "loss": 1.9139,
+      "step": 25105
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.512592592592593e-05,
+      "loss": 1.7323,
+      "step": 25110
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.5122962962962964e-05,
+      "loss": 1.841,
+      "step": 25115
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.512e-05,
+      "loss": 1.796,
+      "step": 25120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.5117037037037035e-05,
+      "loss": 1.9153,
+      "step": 25125
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.511407407407408e-05,
+      "loss": 1.931,
+      "step": 25130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.511111111111111e-05,
+      "loss": 1.9465,
+      "step": 25135
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.510814814814815e-05,
+      "loss": 1.956,
+      "step": 25140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.510518518518518e-05,
+      "loss": 1.8571,
+      "step": 25145
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.510222222222222e-05,
+      "loss": 1.8822,
+      "step": 25150
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.509925925925926e-05,
+      "loss": 1.8245,
+      "step": 25155
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.50962962962963e-05,
+      "loss": 1.7957,
+      "step": 25160
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.509333333333333e-05,
+      "loss": 1.7799,
+      "step": 25165
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 3.509037037037037e-05,
+      "loss": 1.9367,
+      "step": 25170
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.508740740740741e-05,
+      "loss": 1.8052,
+      "step": 25175
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.508444444444445e-05,
+      "loss": 1.7642,
+      "step": 25180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.508148148148148e-05,
+      "loss": 1.8024,
+      "step": 25185
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.507851851851852e-05,
+      "loss": 1.8394,
+      "step": 25190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.507555555555556e-05,
+      "loss": 1.9051,
+      "step": 25195
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.5072592592592597e-05,
+      "loss": 1.7754,
+      "step": 25200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.506962962962963e-05,
+      "loss": 1.8441,
+      "step": 25205
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.506666666666667e-05,
+      "loss": 1.8541,
+      "step": 25210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.5063703703703706e-05,
+      "loss": 1.7804,
+      "step": 25215
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.5060740740740745e-05,
+      "loss": 1.9259,
+      "step": 25220
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.505777777777778e-05,
+      "loss": 1.9042,
+      "step": 25225
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5054814814814816e-05,
+      "loss": 1.7943,
+      "step": 25230
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5051851851851855e-05,
+      "loss": 1.7717,
+      "step": 25235
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5048888888888894e-05,
+      "loss": 1.7454,
+      "step": 25240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5045925925925926e-05,
+      "loss": 1.8993,
+      "step": 25245
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5042962962962965e-05,
+      "loss": 1.8667,
+      "step": 25250
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.504e-05,
+      "loss": 1.7998,
+      "step": 25255
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.503703703703704e-05,
+      "loss": 1.9124,
+      "step": 25260
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.5034074074074074e-05,
+      "loss": 1.9739,
+      "step": 25265
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.503111111111111e-05,
+      "loss": 1.8649,
+      "step": 25270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.502814814814815e-05,
+      "loss": 1.9104,
+      "step": 25275
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.502518518518519e-05,
+      "loss": 1.899,
+      "step": 25280
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.502222222222222e-05,
+      "loss": 1.7956,
+      "step": 25285
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.501925925925926e-05,
+      "loss": 1.8665,
+      "step": 25290
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.50162962962963e-05,
+      "loss": 1.8491,
+      "step": 25295
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.501333333333334e-05,
+      "loss": 1.8148,
+      "step": 25300
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.501037037037037e-05,
+      "loss": 1.7376,
+      "step": 25305
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.500740740740741e-05,
+      "loss": 1.876,
+      "step": 25310
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.500444444444444e-05,
+      "loss": 1.6976,
+      "step": 25315
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.500148148148149e-05,
+      "loss": 1.8688,
+      "step": 25320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.499851851851852e-05,
+      "loss": 1.9759,
+      "step": 25325
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.499555555555556e-05,
+      "loss": 1.7414,
+      "step": 25330
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.499259259259259e-05,
+      "loss": 1.8782,
+      "step": 25335
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.4989629629629636e-05,
+      "loss": 1.8686,
+      "step": 25340
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.498666666666667e-05,
+      "loss": 1.8583,
+      "step": 25345
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.498370370370371e-05,
+      "loss": 1.9514,
+      "step": 25350
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.498074074074074e-05,
+      "loss": 1.8586,
+      "step": 25355
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4977777777777785e-05,
+      "loss": 1.7982,
+      "step": 25360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.497481481481482e-05,
+      "loss": 1.9058,
+      "step": 25365
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4971851851851856e-05,
+      "loss": 1.8518,
+      "step": 25370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.496888888888889e-05,
+      "loss": 1.9264,
+      "step": 25375
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4965925925925926e-05,
+      "loss": 1.9888,
+      "step": 25380
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4962962962962965e-05,
+      "loss": 1.9091,
+      "step": 25385
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4960000000000004e-05,
+      "loss": 1.8697,
+      "step": 25390
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.4957037037037036e-05,
+      "loss": 1.9636,
+      "step": 25395
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4954074074074075e-05,
+      "loss": 1.7825,
+      "step": 25400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4951111111111114e-05,
+      "loss": 1.7781,
+      "step": 25405
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.494814814814815e-05,
+      "loss": 1.8524,
+      "step": 25410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4945185185185185e-05,
+      "loss": 1.8798,
+      "step": 25415
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.4942222222222223e-05,
+      "loss": 1.7901,
+      "step": 25420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.493925925925926e-05,
+      "loss": 1.8895,
+      "step": 25425
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.49362962962963e-05,
+      "loss": 1.8999,
+      "step": 25430
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.493333333333333e-05,
+      "loss": 1.7997,
+      "step": 25435
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.493037037037037e-05,
+      "loss": 1.9247,
+      "step": 25440
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.492740740740741e-05,
+      "loss": 1.7978,
+      "step": 25445
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.492444444444445e-05,
+      "loss": 1.9353,
+      "step": 25450
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.492148148148148e-05,
+      "loss": 1.9127,
+      "step": 25455
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.491851851851852e-05,
+      "loss": 1.9625,
+      "step": 25460
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.491555555555556e-05,
+      "loss": 1.8562,
+      "step": 25465
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.49125925925926e-05,
+      "loss": 1.7948,
+      "step": 25470
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.490962962962963e-05,
+      "loss": 1.9687,
+      "step": 25475
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.490666666666667e-05,
+      "loss": 1.883,
+      "step": 25480
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.490370370370371e-05,
+      "loss": 1.8448,
+      "step": 25485
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.490074074074075e-05,
+      "loss": 1.8739,
+      "step": 25490
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.489777777777778e-05,
+      "loss": 1.8471,
+      "step": 25495
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.489481481481482e-05,
+      "loss": 1.94,
+      "step": 25500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.4891851851851856e-05,
+      "loss": 1.8707,
+      "step": 25505
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4888888888888895e-05,
+      "loss": 1.9176,
+      "step": 25510
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.488592592592593e-05,
+      "loss": 1.9169,
+      "step": 25515
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4882962962962966e-05,
+      "loss": 1.8741,
+      "step": 25520
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 1.7539,
+      "step": 25525
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4877037037037044e-05,
+      "loss": 1.9299,
+      "step": 25530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4874074074074076e-05,
+      "loss": 1.8688,
+      "step": 25535
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4871111111111115e-05,
+      "loss": 1.8637,
+      "step": 25540
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4868148148148147e-05,
+      "loss": 1.8973,
+      "step": 25545
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.486518518518519e-05,
+      "loss": 1.9224,
+      "step": 25550
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4862222222222224e-05,
+      "loss": 1.903,
+      "step": 25555
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4859259259259256e-05,
+      "loss": 1.7937,
+      "step": 25560
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.4856296296296295e-05,
+      "loss": 1.7909,
+      "step": 25565
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4853333333333334e-05,
+      "loss": 1.8075,
+      "step": 25570
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.485037037037037e-05,
+      "loss": 1.9302,
+      "step": 25575
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4847407407407405e-05,
+      "loss": 1.8582,
+      "step": 25580
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4844444444444444e-05,
+      "loss": 1.8397,
+      "step": 25585
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.484148148148148e-05,
+      "loss": 1.9057,
+      "step": 25590
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.483851851851852e-05,
+      "loss": 1.8557,
+      "step": 25595
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.483555555555555e-05,
+      "loss": 1.9313,
+      "step": 25600
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.483259259259259e-05,
+      "loss": 1.9711,
+      "step": 25605
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.482962962962963e-05,
+      "loss": 1.9934,
+      "step": 25610
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.482666666666667e-05,
+      "loss": 1.8104,
+      "step": 25615
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.48237037037037e-05,
+      "loss": 1.7785,
+      "step": 25620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.482074074074074e-05,
+      "loss": 1.9117,
+      "step": 25625
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.481777777777778e-05,
+      "loss": 1.7903,
+      "step": 25630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.481481481481482e-05,
+      "loss": 1.8321,
+      "step": 25635
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.481185185185185e-05,
+      "loss": 1.9943,
+      "step": 25640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.480888888888889e-05,
+      "loss": 1.8763,
+      "step": 25645
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.480592592592593e-05,
+      "loss": 1.7984,
+      "step": 25650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.480296296296297e-05,
+      "loss": 1.903,
+      "step": 25655
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.48e-05,
+      "loss": 1.9589,
+      "step": 25660
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.479703703703704e-05,
+      "loss": 1.9972,
+      "step": 25665
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4794074074074076e-05,
+      "loss": 1.8235,
+      "step": 25670
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4791111111111115e-05,
+      "loss": 2.0638,
+      "step": 25675
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4788740740740745e-05,
+      "loss": 1.8662,
+      "step": 25680
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.478577777777778e-05,
+      "loss": 1.8202,
+      "step": 25685
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4782814814814816e-05,
+      "loss": 1.7592,
+      "step": 25690
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4779851851851855e-05,
+      "loss": 1.8934,
+      "step": 25695
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4776888888888894e-05,
+      "loss": 1.8565,
+      "step": 25700
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4773925925925926e-05,
+      "loss": 1.6515,
+      "step": 25705
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4770962962962964e-05,
+      "loss": 1.8836,
+      "step": 25710
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4768e-05,
+      "loss": 1.8516,
+      "step": 25715
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.476503703703704e-05,
+      "loss": 1.8772,
+      "step": 25720
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.4762074074074074e-05,
+      "loss": 1.8507,
+      "step": 25725
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.475911111111111e-05,
+      "loss": 1.7597,
+      "step": 25730
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.475614814814815e-05,
+      "loss": 1.9444,
+      "step": 25735
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.475318518518519e-05,
+      "loss": 1.7096,
+      "step": 25740
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.475022222222222e-05,
+      "loss": 1.8264,
+      "step": 25745
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.474725925925926e-05,
+      "loss": 1.8842,
+      "step": 25750
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.47442962962963e-05,
+      "loss": 1.983,
+      "step": 25755
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.474133333333334e-05,
+      "loss": 1.7881,
+      "step": 25760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.473837037037037e-05,
+      "loss": 1.8182,
+      "step": 25765
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.473540740740741e-05,
+      "loss": 1.9144,
+      "step": 25770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.473244444444444e-05,
+      "loss": 1.8981,
+      "step": 25775
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.472948148148149e-05,
+      "loss": 2.0234,
+      "step": 25780
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.472651851851852e-05,
+      "loss": 1.8354,
+      "step": 25785
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.472355555555556e-05,
+      "loss": 2.0328,
+      "step": 25790
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.472059259259259e-05,
+      "loss": 1.9785,
+      "step": 25795
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4717629629629636e-05,
+      "loss": 1.857,
+      "step": 25800
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.471466666666667e-05,
+      "loss": 2.021,
+      "step": 25805
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.471170370370371e-05,
+      "loss": 1.9032,
+      "step": 25810
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.470874074074074e-05,
+      "loss": 1.9665,
+      "step": 25815
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4705777777777785e-05,
+      "loss": 1.8891,
+      "step": 25820
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.470281481481482e-05,
+      "loss": 1.8862,
+      "step": 25825
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4699851851851855e-05,
+      "loss": 1.907,
+      "step": 25830
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.469688888888889e-05,
+      "loss": 1.7811,
+      "step": 25835
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4693925925925926e-05,
+      "loss": 1.8134,
+      "step": 25840
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4690962962962965e-05,
+      "loss": 1.803,
+      "step": 25845
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4688000000000004e-05,
+      "loss": 1.8969,
+      "step": 25850
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4685037037037036e-05,
+      "loss": 1.8,
+      "step": 25855
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4682074074074075e-05,
+      "loss": 1.9012,
+      "step": 25860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4679111111111114e-05,
+      "loss": 1.843,
+      "step": 25865
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.467614814814815e-05,
+      "loss": 1.8891,
+      "step": 25870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.4673185185185185e-05,
+      "loss": 1.8011,
+      "step": 25875
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.467022222222222e-05,
+      "loss": 1.7436,
+      "step": 25880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.466725925925926e-05,
+      "loss": 1.9957,
+      "step": 25885
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.46642962962963e-05,
+      "loss": 1.8581,
+      "step": 25890
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.466133333333333e-05,
+      "loss": 1.8335,
+      "step": 25895
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.465837037037037e-05,
+      "loss": 1.9573,
+      "step": 25900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.465540740740741e-05,
+      "loss": 1.8825,
+      "step": 25905
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.465244444444445e-05,
+      "loss": 1.9169,
+      "step": 25910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.464948148148148e-05,
+      "loss": 1.7855,
+      "step": 25915
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.464651851851852e-05,
+      "loss": 1.8015,
+      "step": 25920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.464355555555556e-05,
+      "loss": 1.9257,
+      "step": 25925
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.46405925925926e-05,
+      "loss": 1.8334,
+      "step": 25930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.463762962962963e-05,
+      "loss": 1.8031,
+      "step": 25935
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.463466666666667e-05,
+      "loss": 1.8825,
+      "step": 25940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.463170370370371e-05,
+      "loss": 1.7468,
+      "step": 25945
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.4628740740740747e-05,
+      "loss": 1.8302,
+      "step": 25950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.462577777777778e-05,
+      "loss": 1.6789,
+      "step": 25955
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.462281481481482e-05,
+      "loss": 1.7938,
+      "step": 25960
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4619851851851856e-05,
+      "loss": 1.9825,
+      "step": 25965
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4616888888888895e-05,
+      "loss": 2.0325,
+      "step": 25970
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.461392592592593e-05,
+      "loss": 1.8721,
+      "step": 25975
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4610962962962966e-05,
+      "loss": 1.856,
+      "step": 25980
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4608000000000005e-05,
+      "loss": 1.8112,
+      "step": 25985
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4605037037037044e-05,
+      "loss": 1.8795,
+      "step": 25990
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4602074074074076e-05,
+      "loss": 1.9922,
+      "step": 25995
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4599111111111114e-05,
+      "loss": 1.9082,
+      "step": 26000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4596148148148146e-05,
+      "loss": 1.7745,
+      "step": 26005
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.459318518518519e-05,
+      "loss": 1.7269,
+      "step": 26010
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.4590222222222224e-05,
+      "loss": 1.8373,
+      "step": 26015
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.458725925925926e-05,
+      "loss": 1.9266,
+      "step": 26020
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4584296296296295e-05,
+      "loss": 1.8215,
+      "step": 26025
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4581333333333334e-05,
+      "loss": 1.8742,
+      "step": 26030
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.457837037037037e-05,
+      "loss": 1.8247,
+      "step": 26035
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4575407407407405e-05,
+      "loss": 1.9136,
+      "step": 26040
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.4572444444444444e-05,
+      "loss": 1.8285,
+      "step": 26045
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.456948148148148e-05,
+      "loss": 1.8065,
+      "step": 26050
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.456651851851852e-05,
+      "loss": 1.8805,
+      "step": 26055
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.456355555555555e-05,
+      "loss": 1.8095,
+      "step": 26060
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.456059259259259e-05,
+      "loss": 1.7353,
+      "step": 26065
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 3.455762962962963e-05,
+      "loss": 1.9693,
+      "step": 26070
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.455466666666667e-05,
+      "loss": 1.8052,
+      "step": 26075
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.45517037037037e-05,
+      "loss": 1.8169,
+      "step": 26080
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.454874074074074e-05,
+      "loss": 1.9702,
+      "step": 26085
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.454577777777778e-05,
+      "loss": 1.7609,
+      "step": 26090
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.454281481481482e-05,
+      "loss": 1.8585,
+      "step": 26095
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.453985185185185e-05,
+      "loss": 2.0404,
+      "step": 26100
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.453688888888889e-05,
+      "loss": 1.8763,
+      "step": 26105
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.453392592592593e-05,
+      "loss": 1.9617,
+      "step": 26110
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.453096296296297e-05,
+      "loss": 1.9698,
+      "step": 26115
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.4528e-05,
+      "loss": 1.8195,
+      "step": 26120
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 3.452503703703704e-05,
+      "loss": 1.9747,
+      "step": 26125
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4522074074074076e-05,
+      "loss": 1.8068,
+      "step": 26130
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4519111111111115e-05,
+      "loss": 1.921,
+      "step": 26135
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.451614814814815e-05,
+      "loss": 1.7984,
+      "step": 26140
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4513185185185186e-05,
+      "loss": 1.7852,
+      "step": 26145
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4510222222222225e-05,
+      "loss": 1.8887,
+      "step": 26150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4507259259259264e-05,
+      "loss": 1.9726,
+      "step": 26155
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4504296296296296e-05,
+      "loss": 1.823,
+      "step": 26160
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4501333333333335e-05,
+      "loss": 1.8372,
+      "step": 26165
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.449837037037037e-05,
+      "loss": 1.8759,
+      "step": 26170
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.449540740740741e-05,
+      "loss": 1.7618,
+      "step": 26175
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 3.4492444444444444e-05,
+      "loss": 1.7007,
+      "step": 26180
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.448948148148148e-05,
+      "loss": 1.9698,
+      "step": 26185
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4486518518518515e-05,
+      "loss": 1.9448,
+      "step": 26190
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.448355555555556e-05,
+      "loss": 1.9044,
+      "step": 26195
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.448059259259259e-05,
+      "loss": 1.8554,
+      "step": 26200
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.447762962962963e-05,
+      "loss": 1.8383,
+      "step": 26205
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.4474666666666664e-05,
+      "loss": 1.795,
+      "step": 26210
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.447170370370371e-05,
+      "loss": 1.9453,
+      "step": 26215
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.446874074074074e-05,
+      "loss": 1.8724,
+      "step": 26220
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.446577777777778e-05,
+      "loss": 2.0048,
+      "step": 26225
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.446281481481481e-05,
+      "loss": 1.8792,
+      "step": 26230
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.445985185185185e-05,
+      "loss": 1.8202,
+      "step": 26235
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 3.445688888888889e-05,
+      "loss": 1.991,
+      "step": 26240
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.445392592592593e-05,
+      "loss": 1.8573,
+      "step": 26245
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.445096296296296e-05,
+      "loss": 1.8389,
+      "step": 26250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4448e-05,
+      "loss": 1.859,
+      "step": 26255
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.444503703703704e-05,
+      "loss": 1.88,
+      "step": 26260
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.444207407407408e-05,
+      "loss": 1.8078,
+      "step": 26265
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.443911111111111e-05,
+      "loss": 1.92,
+      "step": 26270
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.443614814814815e-05,
+      "loss": 1.9742,
+      "step": 26275
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.443318518518519e-05,
+      "loss": 1.977,
+      "step": 26280
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4430222222222226e-05,
+      "loss": 1.8195,
+      "step": 26285
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.442725925925926e-05,
+      "loss": 2.011,
+      "step": 26290
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 3.4424296296296297e-05,
+      "loss": 1.8426,
+      "step": 26295
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4421333333333335e-05,
+      "loss": 1.7877,
+      "step": 26300
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4418370370370374e-05,
+      "loss": 1.8484,
+      "step": 26305
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4415407407407406e-05,
+      "loss": 1.8728,
+      "step": 26310
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4412444444444445e-05,
+      "loss": 1.9012,
+      "step": 26315
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4409481481481484e-05,
+      "loss": 1.8064,
+      "step": 26320
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.440651851851852e-05,
+      "loss": 1.7642,
+      "step": 26325
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4403555555555555e-05,
+      "loss": 1.8113,
+      "step": 26330
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.4400592592592594e-05,
+      "loss": 1.7067,
+      "step": 26335
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.439762962962963e-05,
+      "loss": 1.9396,
+      "step": 26340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.439466666666667e-05,
+      "loss": 1.8967,
+      "step": 26345
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 3.43917037037037e-05,
+      "loss": 1.8276,
+      "step": 26350
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.438874074074074e-05,
+      "loss": 1.7048,
+      "step": 26355
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.438577777777778e-05,
+      "loss": 1.7739,
+      "step": 26360
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.438281481481482e-05,
+      "loss": 1.9196,
+      "step": 26365
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.437985185185185e-05,
+      "loss": 1.9543,
+      "step": 26370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.437688888888889e-05,
+      "loss": 1.7592,
+      "step": 26375
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.437392592592593e-05,
+      "loss": 1.8271,
+      "step": 26380
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.437096296296297e-05,
+      "loss": 1.7922,
+      "step": 26385
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.4368e-05,
+      "loss": 1.867,
+      "step": 26390
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.436503703703704e-05,
+      "loss": 1.8377,
+      "step": 26395
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.436207407407407e-05,
+      "loss": 1.9104,
+      "step": 26400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 3.435911111111112e-05,
+      "loss": 1.8955,
+      "step": 26405
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.435614814814815e-05,
+      "loss": 1.7656,
+      "step": 26410
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.435318518518519e-05,
+      "loss": 1.8218,
+      "step": 26415
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.435022222222222e-05,
+      "loss": 1.7283,
+      "step": 26420
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4347259259259265e-05,
+      "loss": 1.855,
+      "step": 26425
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.43442962962963e-05,
+      "loss": 1.8763,
+      "step": 26430
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4341333333333336e-05,
+      "loss": 1.8621,
+      "step": 26435
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.433837037037037e-05,
+      "loss": 1.7818,
+      "step": 26440
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4335407407407414e-05,
+      "loss": 1.8186,
+      "step": 26445
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4332444444444446e-05,
+      "loss": 1.838,
+      "step": 26450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4329481481481485e-05,
+      "loss": 1.8215,
+      "step": 26455
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.432651851851852e-05,
+      "loss": 1.7418,
+      "step": 26460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 3.4323555555555556e-05,
+      "loss": 1.8272,
+      "step": 26465
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.4320592592592594e-05,
+      "loss": 1.8945,
+      "step": 26470
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.431762962962963e-05,
+      "loss": 1.7323,
+      "step": 26475
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.4314666666666665e-05,
+      "loss": 1.7787,
+      "step": 26480
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.4311703703703704e-05,
+      "loss": 1.8821,
+      "step": 26485
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.430874074074074e-05,
+      "loss": 1.9165,
+      "step": 26490
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.430577777777778e-05,
+      "loss": 1.8994,
+      "step": 26495
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.4302814814814814e-05,
+      "loss": 1.8609,
+      "step": 26500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.429985185185185e-05,
+      "loss": 1.8666,
+      "step": 26505
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.429688888888889e-05,
+      "loss": 1.9216,
+      "step": 26510
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.429392592592593e-05,
+      "loss": 1.842,
+      "step": 26515
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 3.429096296296296e-05,
+      "loss": 1.8254,
+      "step": 26520
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.4288e-05,
+      "loss": 1.7813,
+      "step": 26525
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.428503703703704e-05,
+      "loss": 1.8412,
+      "step": 26530
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.428207407407408e-05,
+      "loss": 1.8566,
+      "step": 26535
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.427911111111111e-05,
+      "loss": 1.8186,
+      "step": 26540
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.427614814814815e-05,
+      "loss": 1.8927,
+      "step": 26545
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.427318518518519e-05,
+      "loss": 1.8554,
+      "step": 26550
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.427022222222223e-05,
+      "loss": 1.9565,
+      "step": 26555
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.426725925925926e-05,
+      "loss": 1.844,
+      "step": 26560
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.42642962962963e-05,
+      "loss": 1.9028,
+      "step": 26565
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.426133333333334e-05,
+      "loss": 1.8434,
+      "step": 26570
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 3.4258370370370376e-05,
+      "loss": 1.808,
+      "step": 26575
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.425540740740741e-05,
+      "loss": 2.0082,
+      "step": 26580
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4252444444444447e-05,
+      "loss": 1.7569,
+      "step": 26585
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4249481481481485e-05,
+      "loss": 1.9194,
+      "step": 26590
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4246518518518524e-05,
+      "loss": 1.8615,
+      "step": 26595
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4243555555555556e-05,
+      "loss": 1.8587,
+      "step": 26600
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4240592592592595e-05,
+      "loss": 1.8431,
+      "step": 26605
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4237629629629634e-05,
+      "loss": 1.9205,
+      "step": 26610
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.423466666666667e-05,
+      "loss": 1.8673,
+      "step": 26615
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4231703703703705e-05,
+      "loss": 1.7975,
+      "step": 26620
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4228740740740744e-05,
+      "loss": 1.8357,
+      "step": 26625
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.4225777777777776e-05,
+      "loss": 2.0048,
+      "step": 26630
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.422281481481482e-05,
+      "loss": 1.9088,
+      "step": 26635
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.421985185185185e-05,
+      "loss": 1.8599,
+      "step": 26640
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.421688888888889e-05,
+      "loss": 1.878,
+      "step": 26645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4213925925925924e-05,
+      "loss": 1.9824,
+      "step": 26650
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.421096296296297e-05,
+      "loss": 1.7353,
+      "step": 26655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4208e-05,
+      "loss": 1.8436,
+      "step": 26660
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.420503703703704e-05,
+      "loss": 1.8544,
+      "step": 26665
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.420207407407407e-05,
+      "loss": 1.9418,
+      "step": 26670
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.419911111111112e-05,
+      "loss": 1.8483,
+      "step": 26675
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.419614814814815e-05,
+      "loss": 1.8695,
+      "step": 26680
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.419318518518519e-05,
+      "loss": 1.8626,
+      "step": 26685
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.419022222222222e-05,
+      "loss": 1.9012,
+      "step": 26690
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.418725925925926e-05,
+      "loss": 2.0329,
+      "step": 26695
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.41842962962963e-05,
+      "loss": 1.8436,
+      "step": 26700
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.418133333333334e-05,
+      "loss": 1.8494,
+      "step": 26705
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.417837037037037e-05,
+      "loss": 1.8044,
+      "step": 26710
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.417540740740741e-05,
+      "loss": 1.8807,
+      "step": 26715
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.417244444444445e-05,
+      "loss": 1.8754,
+      "step": 26720
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4169481481481486e-05,
+      "loss": 1.8899,
+      "step": 26725
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.416651851851852e-05,
+      "loss": 1.7409,
+      "step": 26730
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.416355555555556e-05,
+      "loss": 1.9378,
+      "step": 26735
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4160592592592596e-05,
+      "loss": 1.8474,
+      "step": 26740
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4157629629629635e-05,
+      "loss": 1.837,
+      "step": 26745
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.415466666666667e-05,
+      "loss": 2.0005,
+      "step": 26750
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4151703703703706e-05,
+      "loss": 1.9536,
+      "step": 26755
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4148740740740744e-05,
+      "loss": 1.7999,
+      "step": 26760
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.414577777777778e-05,
+      "loss": 1.9324,
+      "step": 26765
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4142814814814815e-05,
+      "loss": 1.8026,
+      "step": 26770
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4139851851851854e-05,
+      "loss": 1.7909,
+      "step": 26775
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.413688888888889e-05,
+      "loss": 1.9094,
+      "step": 26780
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.413392592592593e-05,
+      "loss": 1.9107,
+      "step": 26785
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4130962962962964e-05,
+      "loss": 1.8998,
+      "step": 26790
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.4127999999999996e-05,
+      "loss": 1.8793,
+      "step": 26795
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.412503703703704e-05,
+      "loss": 1.8062,
+      "step": 26800
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.4122074074074073e-05,
+      "loss": 1.8257,
+      "step": 26805
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.411911111111111e-05,
+      "loss": 1.9068,
+      "step": 26810
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.4116148148148144e-05,
+      "loss": 1.833,
+      "step": 26815
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.411318518518519e-05,
+      "loss": 1.9013,
+      "step": 26820
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.411022222222222e-05,
+      "loss": 1.9421,
+      "step": 26825
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.410725925925926e-05,
+      "loss": 1.7704,
+      "step": 26830
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.410429629629629e-05,
+      "loss": 1.8683,
+      "step": 26835
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.410133333333334e-05,
+      "loss": 2.0467,
+      "step": 26840
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.409837037037037e-05,
+      "loss": 1.979,
+      "step": 26845
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.409540740740741e-05,
+      "loss": 1.77,
+      "step": 26850
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 3.409244444444444e-05,
+      "loss": 1.8678,
+      "step": 26855
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.408948148148148e-05,
+      "loss": 1.8441,
+      "step": 26860
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.408651851851852e-05,
+      "loss": 1.7521,
+      "step": 26865
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.408355555555556e-05,
+      "loss": 1.8018,
+      "step": 26870
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.408059259259259e-05,
+      "loss": 1.8317,
+      "step": 26875
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.407762962962963e-05,
+      "loss": 1.8124,
+      "step": 26880
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.407466666666667e-05,
+      "loss": 1.8533,
+      "step": 26885
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.4071703703703706e-05,
+      "loss": 1.8812,
+      "step": 26890
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.406874074074074e-05,
+      "loss": 1.9247,
+      "step": 26895
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.406577777777778e-05,
+      "loss": 1.7035,
+      "step": 26900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.4062814814814816e-05,
+      "loss": 1.8284,
+      "step": 26905
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.4059851851851855e-05,
+      "loss": 2.0453,
+      "step": 26910
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 3.405688888888889e-05,
+      "loss": 2.0362,
+      "step": 26915
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4053925925925926e-05,
+      "loss": 1.9043,
+      "step": 26920
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4050962962962965e-05,
+      "loss": 1.833,
+      "step": 26925
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4048e-05,
+      "loss": 1.8868,
+      "step": 26930
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4045037037037035e-05,
+      "loss": 1.8316,
+      "step": 26935
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4042074074074074e-05,
+      "loss": 1.9401,
+      "step": 26940
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.403911111111111e-05,
+      "loss": 1.7539,
+      "step": 26945
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.403614814814815e-05,
+      "loss": 1.8662,
+      "step": 26950
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.4033185185185184e-05,
+      "loss": 1.8887,
+      "step": 26955
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.403022222222222e-05,
+      "loss": 1.9022,
+      "step": 26960
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.402725925925926e-05,
+      "loss": 1.8331,
+      "step": 26965
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 3.40242962962963e-05,
+      "loss": 1.9568,
+      "step": 26970
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.402133333333333e-05,
+      "loss": 1.8852,
+      "step": 26975
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.401837037037037e-05,
+      "loss": 1.7597,
+      "step": 26980
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.401540740740741e-05,
+      "loss": 1.8942,
+      "step": 26985
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.401244444444445e-05,
+      "loss": 1.8976,
+      "step": 26990
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.400948148148148e-05,
+      "loss": 1.8534,
+      "step": 26995
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.400651851851852e-05,
+      "loss": 1.8821,
+      "step": 27000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.400355555555555e-05,
+      "loss": 1.8543,
+      "step": 27005
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.40005925925926e-05,
+      "loss": 1.9833,
+      "step": 27010
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.399762962962963e-05,
+      "loss": 1.823,
+      "step": 27015
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.399466666666667e-05,
+      "loss": 1.8767,
+      "step": 27020
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.39917037037037e-05,
+      "loss": 1.769,
+      "step": 27025
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3988740740740746e-05,
+      "loss": 1.8142,
+      "step": 27030
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.398577777777778e-05,
+      "loss": 1.9023,
+      "step": 27035
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.398281481481482e-05,
+      "loss": 1.8469,
+      "step": 27040
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.397985185185185e-05,
+      "loss": 1.994,
+      "step": 27045
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3976888888888894e-05,
+      "loss": 1.8669,
+      "step": 27050
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3973925925925926e-05,
+      "loss": 1.86,
+      "step": 27055
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3970962962962965e-05,
+      "loss": 1.7664,
+      "step": 27060
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3968e-05,
+      "loss": 1.9338,
+      "step": 27065
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.396503703703704e-05,
+      "loss": 1.8918,
+      "step": 27070
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3962074074074075e-05,
+      "loss": 1.934,
+      "step": 27075
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.3959111111111114e-05,
+      "loss": 1.9046,
+      "step": 27080
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3956148148148146e-05,
+      "loss": 1.9348,
+      "step": 27085
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3953185185185185e-05,
+      "loss": 1.9185,
+      "step": 27090
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3950222222222223e-05,
+      "loss": 1.8127,
+      "step": 27095
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.394725925925926e-05,
+      "loss": 1.9062,
+      "step": 27100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.3944296296296294e-05,
+      "loss": 1.7339,
+      "step": 27105
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.394133333333333e-05,
+      "loss": 1.7953,
+      "step": 27110
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.393837037037037e-05,
+      "loss": 1.8504,
+      "step": 27115
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.393540740740741e-05,
+      "loss": 1.8805,
+      "step": 27120
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.393244444444444e-05,
+      "loss": 1.954,
+      "step": 27125
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.392948148148148e-05,
+      "loss": 1.8441,
+      "step": 27130
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.392651851851852e-05,
+      "loss": 1.8299,
+      "step": 27135
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.392355555555556e-05,
+      "loss": 1.9614,
+      "step": 27140
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.392059259259259e-05,
+      "loss": 1.9247,
+      "step": 27145
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.391762962962963e-05,
+      "loss": 1.8856,
+      "step": 27150
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.391466666666667e-05,
+      "loss": 1.8852,
+      "step": 27155
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.391170370370371e-05,
+      "loss": 1.8622,
+      "step": 27160
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.390874074074074e-05,
+      "loss": 1.8509,
+      "step": 27165
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.390577777777778e-05,
+      "loss": 1.8655,
+      "step": 27170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.390281481481482e-05,
+      "loss": 1.8659,
+      "step": 27175
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.3899851851851856e-05,
+      "loss": 1.9569,
+      "step": 27180
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.389688888888889e-05,
+      "loss": 1.8286,
+      "step": 27185
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.389392592592593e-05,
+      "loss": 1.7826,
+      "step": 27190
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.3890962962962966e-05,
+      "loss": 1.8569,
+      "step": 27195
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3888000000000005e-05,
+      "loss": 1.9217,
+      "step": 27200
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.388503703703704e-05,
+      "loss": 1.8555,
+      "step": 27205
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3882074074074076e-05,
+      "loss": 1.9139,
+      "step": 27210
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3879111111111115e-05,
+      "loss": 1.8997,
+      "step": 27215
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.387614814814815e-05,
+      "loss": 1.8336,
+      "step": 27220
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3873185185185185e-05,
+      "loss": 1.9609,
+      "step": 27225
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3870222222222224e-05,
+      "loss": 1.8004,
+      "step": 27230
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.386725925925926e-05,
+      "loss": 1.8281,
+      "step": 27235
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.38642962962963e-05,
+      "loss": 1.8662,
+      "step": 27240
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.3861333333333334e-05,
+      "loss": 1.81,
+      "step": 27245
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.385837037037037e-05,
+      "loss": 1.8333,
+      "step": 27250
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.3855407407407405e-05,
+      "loss": 1.7969,
+      "step": 27255
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.385244444444445e-05,
+      "loss": 1.7766,
+      "step": 27260
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.384948148148148e-05,
+      "loss": 1.8643,
+      "step": 27265
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.384651851851852e-05,
+      "loss": 1.7737,
+      "step": 27270
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.384355555555555e-05,
+      "loss": 1.8939,
+      "step": 27275
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.38405925925926e-05,
+      "loss": 1.8586,
+      "step": 27280
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.383762962962963e-05,
+      "loss": 1.8959,
+      "step": 27285
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.383466666666667e-05,
+      "loss": 1.8894,
+      "step": 27290
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.38317037037037e-05,
+      "loss": 1.8193,
+      "step": 27295
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.382874074074075e-05,
+      "loss": 1.8419,
+      "step": 27300
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.382577777777778e-05,
+      "loss": 1.9208,
+      "step": 27305
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.382281481481482e-05,
+      "loss": 1.8542,
+      "step": 27310
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.381985185185185e-05,
+      "loss": 1.8298,
+      "step": 27315
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.381688888888889e-05,
+      "loss": 1.894,
+      "step": 27320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.381392592592593e-05,
+      "loss": 1.8889,
+      "step": 27325
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.381096296296297e-05,
+      "loss": 1.7362,
+      "step": 27330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3808e-05,
+      "loss": 1.9045,
+      "step": 27335
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.380503703703704e-05,
+      "loss": 1.9487,
+      "step": 27340
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3802074074074077e-05,
+      "loss": 1.7598,
+      "step": 27345
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3799111111111115e-05,
+      "loss": 1.9082,
+      "step": 27350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.379614814814815e-05,
+      "loss": 1.8697,
+      "step": 27355
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3793185185185186e-05,
+      "loss": 1.829,
+      "step": 27360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.3790222222222225e-05,
+      "loss": 1.7843,
+      "step": 27365
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3787259259259264e-05,
+      "loss": 1.8943,
+      "step": 27370
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3784296296296296e-05,
+      "loss": 1.7697,
+      "step": 27375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3781333333333335e-05,
+      "loss": 1.9667,
+      "step": 27380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3778370370370374e-05,
+      "loss": 1.9397,
+      "step": 27385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.377540740740741e-05,
+      "loss": 1.8692,
+      "step": 27390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3772444444444444e-05,
+      "loss": 1.7121,
+      "step": 27395
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.376948148148148e-05,
+      "loss": 1.7247,
+      "step": 27400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.376651851851852e-05,
+      "loss": 1.8749,
+      "step": 27405
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.376355555555556e-05,
+      "loss": 1.9242,
+      "step": 27410
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.376059259259259e-05,
+      "loss": 1.8277,
+      "step": 27415
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.375762962962963e-05,
+      "loss": 1.8604,
+      "step": 27420
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.375466666666667e-05,
+      "loss": 1.7147,
+      "step": 27425
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.375170370370371e-05,
+      "loss": 1.7693,
+      "step": 27430
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.374874074074074e-05,
+      "loss": 1.8308,
+      "step": 27435
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.374577777777778e-05,
+      "loss": 1.9276,
+      "step": 27440
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.374281481481482e-05,
+      "loss": 1.8627,
+      "step": 27445
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.373985185185186e-05,
+      "loss": 1.8388,
+      "step": 27450
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.373688888888889e-05,
+      "loss": 1.917,
+      "step": 27455
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.373392592592593e-05,
+      "loss": 1.9011,
+      "step": 27460
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.373096296296297e-05,
+      "loss": 1.8276,
+      "step": 27465
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.3728000000000006e-05,
+      "loss": 1.8683,
+      "step": 27470
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.372503703703704e-05,
+      "loss": 1.986,
+      "step": 27475
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.372207407407408e-05,
+      "loss": 1.851,
+      "step": 27480
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.371911111111111e-05,
+      "loss": 1.8596,
+      "step": 27485
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3716148148148155e-05,
+      "loss": 1.8237,
+      "step": 27490
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.371318518518519e-05,
+      "loss": 1.7493,
+      "step": 27495
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3710222222222226e-05,
+      "loss": 1.8453,
+      "step": 27500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.370725925925926e-05,
+      "loss": 1.9803,
+      "step": 27505
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3704296296296303e-05,
+      "loss": 1.8217,
+      "step": 27510
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3701333333333335e-05,
+      "loss": 1.7674,
+      "step": 27515
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3698370370370374e-05,
+      "loss": 1.9307,
+      "step": 27520
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.3695407407407406e-05,
+      "loss": 1.7862,
+      "step": 27525
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.369244444444445e-05,
+      "loss": 1.8071,
+      "step": 27530
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3689481481481484e-05,
+      "loss": 1.9153,
+      "step": 27535
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.368651851851852e-05,
+      "loss": 1.9317,
+      "step": 27540
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3683555555555555e-05,
+      "loss": 1.8791,
+      "step": 27545
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.3680592592592594e-05,
+      "loss": 1.7685,
+      "step": 27550
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.367762962962963e-05,
+      "loss": 1.8431,
+      "step": 27555
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.367466666666667e-05,
+      "loss": 1.7817,
+      "step": 27560
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.36717037037037e-05,
+      "loss": 1.8976,
+      "step": 27565
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.366874074074074e-05,
+      "loss": 1.7995,
+      "step": 27570
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.366577777777778e-05,
+      "loss": 1.8784,
+      "step": 27575
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.366281481481481e-05,
+      "loss": 1.8824,
+      "step": 27580
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.365985185185185e-05,
+      "loss": 1.8377,
+      "step": 27585
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.365688888888889e-05,
+      "loss": 1.9117,
+      "step": 27590
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.365392592592593e-05,
+      "loss": 1.8219,
+      "step": 27595
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.365096296296296e-05,
+      "loss": 1.8808,
+      "step": 27600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3648e-05,
+      "loss": 1.975,
+      "step": 27605
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.364503703703704e-05,
+      "loss": 1.9522,
+      "step": 27610
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.364207407407408e-05,
+      "loss": 1.8265,
+      "step": 27615
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363911111111111e-05,
+      "loss": 1.8612,
+      "step": 27620
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363614814814815e-05,
+      "loss": 1.8166,
+      "step": 27625
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.363318518518518e-05,
+      "loss": 1.9841,
+      "step": 27630
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.3630222222222227e-05,
+      "loss": 1.865,
+      "step": 27635
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.362725925925926e-05,
+      "loss": 1.8827,
+      "step": 27640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.36242962962963e-05,
+      "loss": 1.8036,
+      "step": 27645
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.362133333333333e-05,
+      "loss": 1.8639,
+      "step": 27650
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3618370370370375e-05,
+      "loss": 1.9511,
+      "step": 27655
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.361540740740741e-05,
+      "loss": 1.7709,
+      "step": 27660
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3612444444444446e-05,
+      "loss": 1.9322,
+      "step": 27665
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.360948148148148e-05,
+      "loss": 1.9328,
+      "step": 27670
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3606518518518524e-05,
+      "loss": 1.8236,
+      "step": 27675
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3603555555555556e-05,
+      "loss": 1.8497,
+      "step": 27680
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3600592592592594e-05,
+      "loss": 1.9246,
+      "step": 27685
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3597629629629626e-05,
+      "loss": 1.7647,
+      "step": 27690
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.359466666666667e-05,
+      "loss": 1.9265,
+      "step": 27695
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3591703703703704e-05,
+      "loss": 1.9041,
+      "step": 27700
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.358874074074074e-05,
+      "loss": 1.9264,
+      "step": 27705
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3585777777777775e-05,
+      "loss": 1.9545,
+      "step": 27710
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3582814814814814e-05,
+      "loss": 1.8655,
+      "step": 27715
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.357985185185185e-05,
+      "loss": 1.8238,
+      "step": 27720
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.357688888888889e-05,
+      "loss": 1.7974,
+      "step": 27725
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3573925925925924e-05,
+      "loss": 1.9911,
+      "step": 27730
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.357096296296296e-05,
+      "loss": 1.8882,
+      "step": 27735
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.3568e-05,
+      "loss": 1.8791,
+      "step": 27740
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.356503703703704e-05,
+      "loss": 1.7675,
+      "step": 27745
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.356207407407407e-05,
+      "loss": 1.9004,
+      "step": 27750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.355911111111111e-05,
+      "loss": 1.6786,
+      "step": 27755
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.355614814814815e-05,
+      "loss": 1.9158,
+      "step": 27760
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.355318518518519e-05,
+      "loss": 1.9172,
+      "step": 27765
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.355022222222222e-05,
+      "loss": 1.903,
+      "step": 27770
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.354725925925926e-05,
+      "loss": 1.8954,
+      "step": 27775
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.35442962962963e-05,
+      "loss": 1.8498,
+      "step": 27780
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.354133333333334e-05,
+      "loss": 1.7382,
+      "step": 27785
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.353837037037037e-05,
+      "loss": 1.8835,
+      "step": 27790
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.353540740740741e-05,
+      "loss": 1.7171,
+      "step": 27795
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.353244444444445e-05,
+      "loss": 1.8001,
+      "step": 27800
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.3529481481481486e-05,
+      "loss": 1.8424,
+      "step": 27805
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.352651851851852e-05,
+      "loss": 1.9593,
+      "step": 27810
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.3523555555555556e-05,
+      "loss": 1.8342,
+      "step": 27815
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3520592592592595e-05,
+      "loss": 1.827,
+      "step": 27820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3517629629629634e-05,
+      "loss": 1.9964,
+      "step": 27825
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3514666666666666e-05,
+      "loss": 1.803,
+      "step": 27830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3511703703703705e-05,
+      "loss": 1.9362,
+      "step": 27835
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3508740740740744e-05,
+      "loss": 1.8935,
+      "step": 27840
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.350577777777778e-05,
+      "loss": 1.8737,
+      "step": 27845
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3502814814814815e-05,
+      "loss": 1.8276,
+      "step": 27850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3499851851851853e-05,
+      "loss": 1.7986,
+      "step": 27855
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.3496888888888885e-05,
+      "loss": 1.7834,
+      "step": 27860
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.349392592592593e-05,
+      "loss": 1.8473,
+      "step": 27865
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.349096296296296e-05,
+      "loss": 1.756,
+      "step": 27870
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3488e-05,
+      "loss": 1.9552,
+      "step": 27875
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3485037037037034e-05,
+      "loss": 1.8291,
+      "step": 27880
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.348207407407408e-05,
+      "loss": 1.8651,
+      "step": 27885
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.347911111111111e-05,
+      "loss": 1.7846,
+      "step": 27890
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.347614814814815e-05,
+      "loss": 1.6562,
+      "step": 27895
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.347318518518518e-05,
+      "loss": 1.9841,
+      "step": 27900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.347022222222223e-05,
+      "loss": 1.8572,
+      "step": 27905
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.346725925925926e-05,
+      "loss": 1.885,
+      "step": 27910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.34642962962963e-05,
+      "loss": 1.9045,
+      "step": 27915
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.346133333333333e-05,
+      "loss": 1.8631,
+      "step": 27920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.3458370370370377e-05,
+      "loss": 1.7185,
+      "step": 27925
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.345540740740741e-05,
+      "loss": 1.9752,
+      "step": 27930
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.345244444444445e-05,
+      "loss": 1.8501,
+      "step": 27935
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.344948148148148e-05,
+      "loss": 1.9054,
+      "step": 27940
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.344651851851852e-05,
+      "loss": 1.8105,
+      "step": 27945
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.344355555555556e-05,
+      "loss": 1.687,
+      "step": 27950
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3440592592592596e-05,
+      "loss": 1.8352,
+      "step": 27955
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.343762962962963e-05,
+      "loss": 1.8621,
+      "step": 27960
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.343466666666667e-05,
+      "loss": 1.9309,
+      "step": 27965
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3431703703703706e-05,
+      "loss": 1.9289,
+      "step": 27970
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3428740740740744e-05,
+      "loss": 1.7891,
+      "step": 27975
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.3425777777777777e-05,
+      "loss": 1.9466,
+      "step": 27980
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3422814814814815e-05,
+      "loss": 1.9184,
+      "step": 27985
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3419851851851854e-05,
+      "loss": 1.9668,
+      "step": 27990
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.341688888888889e-05,
+      "loss": 1.9515,
+      "step": 27995
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3413925925925925e-05,
+      "loss": 1.8021,
+      "step": 28000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3410962962962964e-05,
+      "loss": 1.8978,
+      "step": 28005
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3408e-05,
+      "loss": 1.7877,
+      "step": 28010
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.340503703703704e-05,
+      "loss": 1.8504,
+      "step": 28015
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3402074074074074e-05,
+      "loss": 1.8294,
+      "step": 28020
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339911111111111e-05,
+      "loss": 1.928,
+      "step": 28025
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339614814814815e-05,
+      "loss": 1.8477,
+      "step": 28030
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339318518518519e-05,
+      "loss": 1.9393,
+      "step": 28035
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.339022222222222e-05,
+      "loss": 1.9513,
+      "step": 28040
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.338725925925926e-05,
+      "loss": 1.8603,
+      "step": 28045
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.33842962962963e-05,
+      "loss": 1.8251,
+      "step": 28050
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.338133333333334e-05,
+      "loss": 1.8296,
+      "step": 28055
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.337837037037037e-05,
+      "loss": 1.7772,
+      "step": 28060
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.337540740740741e-05,
+      "loss": 1.9128,
+      "step": 28065
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.337244444444445e-05,
+      "loss": 1.8094,
+      "step": 28070
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.336948148148149e-05,
+      "loss": 2.0236,
+      "step": 28075
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.336651851851852e-05,
+      "loss": 1.9778,
+      "step": 28080
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.336355555555556e-05,
+      "loss": 1.7681,
+      "step": 28085
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.336059259259259e-05,
+      "loss": 1.8195,
+      "step": 28090
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3357629629629636e-05,
+      "loss": 1.7828,
+      "step": 28095
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.335466666666667e-05,
+      "loss": 1.902,
+      "step": 28100
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3351703703703706e-05,
+      "loss": 1.9494,
+      "step": 28105
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.334874074074074e-05,
+      "loss": 1.8356,
+      "step": 28110
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3345777777777784e-05,
+      "loss": 1.9337,
+      "step": 28115
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3342814814814816e-05,
+      "loss": 1.829,
+      "step": 28120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3339851851851855e-05,
+      "loss": 1.8372,
+      "step": 28125
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.333688888888889e-05,
+      "loss": 1.7224,
+      "step": 28130
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.333392592592593e-05,
+      "loss": 1.6538,
+      "step": 28135
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3330962962962965e-05,
+      "loss": 1.5821,
+      "step": 28140
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3328000000000003e-05,
+      "loss": 1.6281,
+      "step": 28145
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.3325037037037036e-05,
+      "loss": 1.7006,
+      "step": 28150
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.332207407407408e-05,
+      "loss": 1.782,
+      "step": 28155
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.331911111111111e-05,
+      "loss": 1.7907,
+      "step": 28160
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.331614814814815e-05,
+      "loss": 1.7706,
+      "step": 28165
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.3313185185185184e-05,
+      "loss": 1.7231,
+      "step": 28170
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.331022222222222e-05,
+      "loss": 1.8103,
+      "step": 28175
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.330725925925926e-05,
+      "loss": 1.7197,
+      "step": 28180
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.33042962962963e-05,
+      "loss": 1.6846,
+      "step": 28185
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.330133333333333e-05,
+      "loss": 1.8474,
+      "step": 28190
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.329837037037037e-05,
+      "loss": 1.7646,
+      "step": 28195
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.329540740740741e-05,
+      "loss": 1.6594,
+      "step": 28200
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.329244444444445e-05,
+      "loss": 1.7398,
+      "step": 28205
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.328948148148148e-05,
+      "loss": 1.8003,
+      "step": 28210
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.328651851851852e-05,
+      "loss": 1.7009,
+      "step": 28215
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.328355555555556e-05,
+      "loss": 1.5595,
+      "step": 28220
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.32805925925926e-05,
+      "loss": 1.6855,
+      "step": 28225
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.327762962962963e-05,
+      "loss": 1.6388,
+      "step": 28230
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.327466666666667e-05,
+      "loss": 1.7706,
+      "step": 28235
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.327170370370371e-05,
+      "loss": 1.8324,
+      "step": 28240
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.3268740740740746e-05,
+      "loss": 1.898,
+      "step": 28245
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.326577777777778e-05,
+      "loss": 1.8067,
+      "step": 28250
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.326281481481482e-05,
+      "loss": 1.6449,
+      "step": 28255
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.3259851851851856e-05,
+      "loss": 1.8342,
+      "step": 28260
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.3256888888888895e-05,
+      "loss": 1.7177,
+      "step": 28265
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3253925925925927e-05,
+      "loss": 1.6746,
+      "step": 28270
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3250962962962965e-05,
+      "loss": 1.7529,
+      "step": 28275
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3248000000000004e-05,
+      "loss": 1.6488,
+      "step": 28280
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.324503703703704e-05,
+      "loss": 1.6773,
+      "step": 28285
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3242074074074075e-05,
+      "loss": 1.742,
+      "step": 28290
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3239111111111114e-05,
+      "loss": 1.7488,
+      "step": 28295
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.323614814814815e-05,
+      "loss": 1.7793,
+      "step": 28300
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.323318518518519e-05,
+      "loss": 1.7276,
+      "step": 28305
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3230222222222224e-05,
+      "loss": 1.7337,
+      "step": 28310
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.322725925925926e-05,
+      "loss": 1.6659,
+      "step": 28315
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.3224296296296294e-05,
+      "loss": 1.8224,
+      "step": 28320
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.322133333333334e-05,
+      "loss": 1.6992,
+      "step": 28325
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.321837037037037e-05,
+      "loss": 1.7376,
+      "step": 28330
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.321540740740741e-05,
+      "loss": 1.6674,
+      "step": 28335
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.321244444444444e-05,
+      "loss": 1.7425,
+      "step": 28340
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.320948148148148e-05,
+      "loss": 1.7201,
+      "step": 28345
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.320651851851852e-05,
+      "loss": 1.7534,
+      "step": 28350
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.320355555555555e-05,
+      "loss": 1.7864,
+      "step": 28355
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.320059259259259e-05,
+      "loss": 1.7196,
+      "step": 28360
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.319762962962963e-05,
+      "loss": 1.7655,
+      "step": 28365
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.319466666666667e-05,
+      "loss": 1.6174,
+      "step": 28370
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.31917037037037e-05,
+      "loss": 1.7523,
+      "step": 28375
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.318874074074074e-05,
+      "loss": 1.7561,
+      "step": 28380
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.318577777777778e-05,
+      "loss": 1.7134,
+      "step": 28385
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.318281481481482e-05,
+      "loss": 1.7702,
+      "step": 28390
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.317985185185185e-05,
+      "loss": 1.7156,
+      "step": 28395
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.317688888888889e-05,
+      "loss": 1.7965,
+      "step": 28400
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.317392592592593e-05,
+      "loss": 1.7134,
+      "step": 28405
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3170962962962966e-05,
+      "loss": 1.8202,
+      "step": 28410
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3168e-05,
+      "loss": 1.6275,
+      "step": 28415
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.316503703703704e-05,
+      "loss": 1.7416,
+      "step": 28420
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3162074074074076e-05,
+      "loss": 1.744,
+      "step": 28425
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.3159111111111115e-05,
+      "loss": 1.8075,
+      "step": 28430
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.315614814814815e-05,
+      "loss": 1.6238,
+      "step": 28435
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3153185185185186e-05,
+      "loss": 1.7057,
+      "step": 28440
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3150222222222224e-05,
+      "loss": 1.7718,
+      "step": 28445
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.314725925925926e-05,
+      "loss": 1.7203,
+      "step": 28450
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3144296296296295e-05,
+      "loss": 1.7554,
+      "step": 28455
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3141333333333334e-05,
+      "loss": 1.8198,
+      "step": 28460
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.313837037037037e-05,
+      "loss": 1.8542,
+      "step": 28465
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.313540740740741e-05,
+      "loss": 1.8241,
+      "step": 28470
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3132444444444444e-05,
+      "loss": 1.8091,
+      "step": 28475
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.312948148148148e-05,
+      "loss": 1.7722,
+      "step": 28480
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.3126518518518515e-05,
+      "loss": 1.776,
+      "step": 28485
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.312355555555556e-05,
+      "loss": 1.7729,
+      "step": 28490
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.312059259259259e-05,
+      "loss": 1.7194,
+      "step": 28495
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.311762962962963e-05,
+      "loss": 1.8304,
+      "step": 28500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.311466666666666e-05,
+      "loss": 1.7313,
+      "step": 28505
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.311170370370371e-05,
+      "loss": 1.7868,
+      "step": 28510
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.310874074074074e-05,
+      "loss": 1.5741,
+      "step": 28515
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.310577777777778e-05,
+      "loss": 1.6619,
+      "step": 28520
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.310281481481481e-05,
+      "loss": 1.7637,
+      "step": 28525
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.309985185185186e-05,
+      "loss": 1.8242,
+      "step": 28530
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.309688888888889e-05,
+      "loss": 1.7894,
+      "step": 28535
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.309392592592593e-05,
+      "loss": 1.7902,
+      "step": 28540
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.309096296296296e-05,
+      "loss": 1.7513,
+      "step": 28545
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3088e-05,
+      "loss": 1.7672,
+      "step": 28550
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.308503703703704e-05,
+      "loss": 1.7964,
+      "step": 28555
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.308207407407408e-05,
+      "loss": 1.6526,
+      "step": 28560
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.307911111111111e-05,
+      "loss": 1.7298,
+      "step": 28565
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.307614814814815e-05,
+      "loss": 1.7679,
+      "step": 28570
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3073185185185186e-05,
+      "loss": 1.7826,
+      "step": 28575
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3070222222222225e-05,
+      "loss": 1.7892,
+      "step": 28580
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.306725925925926e-05,
+      "loss": 1.6299,
+      "step": 28585
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3064296296296296e-05,
+      "loss": 1.7188,
+      "step": 28590
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3061333333333335e-05,
+      "loss": 1.6656,
+      "step": 28595
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.3058370370370374e-05,
+      "loss": 1.8235,
+      "step": 28600
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3055407407407406e-05,
+      "loss": 1.6673,
+      "step": 28605
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3052444444444445e-05,
+      "loss": 1.8927,
+      "step": 28610
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.304948148148148e-05,
+      "loss": 1.7311,
+      "step": 28615
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.304651851851852e-05,
+      "loss": 1.7159,
+      "step": 28620
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.3043555555555554e-05,
+      "loss": 1.8549,
+      "step": 28625
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.304059259259259e-05,
+      "loss": 1.6865,
+      "step": 28630
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.303762962962963e-05,
+      "loss": 1.7174,
+      "step": 28635
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.303466666666667e-05,
+      "loss": 1.6539,
+      "step": 28640
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.30317037037037e-05,
+      "loss": 1.8361,
+      "step": 28645
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.302874074074074e-05,
+      "loss": 1.6395,
+      "step": 28650
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.302577777777778e-05,
+      "loss": 1.6991,
+      "step": 28655
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.302281481481482e-05,
+      "loss": 1.7888,
+      "step": 28660
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.301985185185185e-05,
+      "loss": 1.6486,
+      "step": 28665
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.301688888888889e-05,
+      "loss": 1.827,
+      "step": 28670
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.301392592592593e-05,
+      "loss": 1.8384,
+      "step": 28675
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.301096296296297e-05,
+      "loss": 1.8186,
+      "step": 28680
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.3008e-05,
+      "loss": 1.7747,
+      "step": 28685
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.300503703703704e-05,
+      "loss": 1.7714,
+      "step": 28690
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.300207407407408e-05,
+      "loss": 1.6875,
+      "step": 28695
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.2999111111111116e-05,
+      "loss": 1.72,
+      "step": 28700
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.299614814814815e-05,
+      "loss": 1.7074,
+      "step": 28705
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.299318518518519e-05,
+      "loss": 1.6847,
+      "step": 28710
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.299022222222222e-05,
+      "loss": 1.7436,
+      "step": 28715
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2987259259259265e-05,
+      "loss": 2.0024,
+      "step": 28720
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.29842962962963e-05,
+      "loss": 1.6868,
+      "step": 28725
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2981333333333336e-05,
+      "loss": 1.648,
+      "step": 28730
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.297837037037037e-05,
+      "loss": 1.7467,
+      "step": 28735
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.297540740740741e-05,
+      "loss": 1.7308,
+      "step": 28740
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2972444444444445e-05,
+      "loss": 1.6924,
+      "step": 28745
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2969481481481484e-05,
+      "loss": 1.773,
+      "step": 28750
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2966518518518516e-05,
+      "loss": 1.739,
+      "step": 28755
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.296355555555556e-05,
+      "loss": 1.8963,
+      "step": 28760
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.2960592592592594e-05,
+      "loss": 1.7628,
+      "step": 28765
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.295762962962963e-05,
+      "loss": 1.7016,
+      "step": 28770
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.2954666666666665e-05,
+      "loss": 1.764,
+      "step": 28775
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.2951703703703703e-05,
+      "loss": 1.6707,
+      "step": 28780
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.294874074074074e-05,
+      "loss": 1.8278,
+      "step": 28785
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.294577777777778e-05,
+      "loss": 1.6476,
+      "step": 28790
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.294281481481481e-05,
+      "loss": 1.7054,
+      "step": 28795
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293985185185185e-05,
+      "loss": 1.8137,
+      "step": 28800
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293688888888889e-05,
+      "loss": 1.6914,
+      "step": 28805
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293392592592593e-05,
+      "loss": 1.9189,
+      "step": 28810
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.293096296296296e-05,
+      "loss": 1.7599,
+      "step": 28815
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.2928e-05,
+      "loss": 1.7007,
+      "step": 28820
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.292503703703704e-05,
+      "loss": 1.8185,
+      "step": 28825
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.292207407407408e-05,
+      "loss": 1.8087,
+      "step": 28830
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.291911111111111e-05,
+      "loss": 2.0938,
+      "step": 28835
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.291614814814815e-05,
+      "loss": 1.6221,
+      "step": 28840
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.291318518518519e-05,
+      "loss": 1.7422,
+      "step": 28845
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.291022222222223e-05,
+      "loss": 1.8171,
+      "step": 28850
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.290725925925926e-05,
+      "loss": 1.8412,
+      "step": 28855
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.29042962962963e-05,
+      "loss": 1.762,
+      "step": 28860
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.2901333333333336e-05,
+      "loss": 1.6639,
+      "step": 28865
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.2898370370370375e-05,
+      "loss": 1.7134,
+      "step": 28870
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.289540740740741e-05,
+      "loss": 1.8787,
+      "step": 28875
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.2892444444444446e-05,
+      "loss": 1.7999,
+      "step": 28880
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2889481481481485e-05,
+      "loss": 1.8612,
+      "step": 28885
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2886518518518524e-05,
+      "loss": 1.7203,
+      "step": 28890
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2883555555555556e-05,
+      "loss": 1.6524,
+      "step": 28895
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2880592592592595e-05,
+      "loss": 1.7538,
+      "step": 28900
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.287762962962963e-05,
+      "loss": 1.7661,
+      "step": 28905
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.287466666666667e-05,
+      "loss": 1.7724,
+      "step": 28910
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.2871703703703704e-05,
+      "loss": 1.7852,
+      "step": 28915
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.286874074074074e-05,
+      "loss": 1.817,
+      "step": 28920
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.286577777777778e-05,
+      "loss": 1.7295,
+      "step": 28925
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.286281481481482e-05,
+      "loss": 1.7138,
+      "step": 28930
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.285985185185185e-05,
+      "loss": 1.8018,
+      "step": 28935
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.285688888888889e-05,
+      "loss": 1.8042,
+      "step": 28940
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.2853925925925924e-05,
+      "loss": 1.709,
+      "step": 28945
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.285096296296297e-05,
+      "loss": 1.7333,
+      "step": 28950
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.2848e-05,
+      "loss": 1.8335,
+      "step": 28955
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.284503703703704e-05,
+      "loss": 1.7092,
+      "step": 28960
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.284207407407407e-05,
+      "loss": 1.8359,
+      "step": 28965
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.283911111111112e-05,
+      "loss": 1.7295,
+      "step": 28970
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.283614814814815e-05,
+      "loss": 1.7539,
+      "step": 28975
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.283318518518519e-05,
+      "loss": 1.6472,
+      "step": 28980
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.283022222222222e-05,
+      "loss": 1.8571,
+      "step": 28985
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.2827259259259266e-05,
+      "loss": 1.8143,
+      "step": 28990
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.28242962962963e-05,
+      "loss": 1.6736,
+      "step": 28995
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.282133333333334e-05,
+      "loss": 1.6895,
+      "step": 29000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.281837037037037e-05,
+      "loss": 1.7298,
+      "step": 29005
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.281540740740741e-05,
+      "loss": 1.7215,
+      "step": 29010
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.281244444444445e-05,
+      "loss": 1.7344,
+      "step": 29015
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2809481481481486e-05,
+      "loss": 1.7413,
+      "step": 29020
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.280651851851852e-05,
+      "loss": 1.7335,
+      "step": 29025
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2803555555555557e-05,
+      "loss": 1.6782,
+      "step": 29030
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2800592592592595e-05,
+      "loss": 1.8319,
+      "step": 29035
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2797629629629634e-05,
+      "loss": 1.8867,
+      "step": 29040
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2794666666666666e-05,
+      "loss": 1.8898,
+      "step": 29045
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.2791703703703705e-05,
+      "loss": 1.6628,
+      "step": 29050
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2788740740740744e-05,
+      "loss": 1.6802,
+      "step": 29055
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.278577777777778e-05,
+      "loss": 1.8035,
+      "step": 29060
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2782814814814815e-05,
+      "loss": 1.7856,
+      "step": 29065
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2779851851851854e-05,
+      "loss": 1.8685,
+      "step": 29070
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.277688888888889e-05,
+      "loss": 1.6967,
+      "step": 29075
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.277392592592593e-05,
+      "loss": 1.7145,
+      "step": 29080
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.277096296296296e-05,
+      "loss": 1.7407,
+      "step": 29085
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.2768e-05,
+      "loss": 1.8242,
+      "step": 29090
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.276503703703704e-05,
+      "loss": 1.8387,
+      "step": 29095
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.276207407407408e-05,
+      "loss": 1.7806,
+      "step": 29100
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.275911111111111e-05,
+      "loss": 1.8899,
+      "step": 29105
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.275614814814815e-05,
+      "loss": 1.7004,
+      "step": 29110
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.275318518518519e-05,
+      "loss": 1.6837,
+      "step": 29115
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.275022222222222e-05,
+      "loss": 1.7514,
+      "step": 29120
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.274725925925926e-05,
+      "loss": 1.7557,
+      "step": 29125
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.274429629629629e-05,
+      "loss": 1.7695,
+      "step": 29130
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.274133333333334e-05,
+      "loss": 1.6763,
+      "step": 29135
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.273837037037037e-05,
+      "loss": 1.8425,
+      "step": 29140
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.273540740740741e-05,
+      "loss": 1.802,
+      "step": 29145
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.273244444444444e-05,
+      "loss": 1.7162,
+      "step": 29150
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.2729481481481486e-05,
+      "loss": 1.825,
+      "step": 29155
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.272651851851852e-05,
+      "loss": 1.6796,
+      "step": 29160
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.272355555555556e-05,
+      "loss": 1.755,
+      "step": 29165
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.272059259259259e-05,
+      "loss": 1.7298,
+      "step": 29170
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.271762962962963e-05,
+      "loss": 1.9278,
+      "step": 29175
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.271466666666667e-05,
+      "loss": 1.821,
+      "step": 29180
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2711703703703706e-05,
+      "loss": 1.846,
+      "step": 29185
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.270874074074074e-05,
+      "loss": 1.6878,
+      "step": 29190
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.270577777777778e-05,
+      "loss": 1.7034,
+      "step": 29195
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2702814814814815e-05,
+      "loss": 1.6669,
+      "step": 29200
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2699851851851854e-05,
+      "loss": 1.8904,
+      "step": 29205
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2696888888888886e-05,
+      "loss": 1.679,
+      "step": 29210
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2693925925925925e-05,
+      "loss": 1.7235,
+      "step": 29215
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.2690962962962964e-05,
+      "loss": 1.7319,
+      "step": 29220
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.2688e-05,
+      "loss": 1.579,
+      "step": 29225
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.2685037037037035e-05,
+      "loss": 1.8418,
+      "step": 29230
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.2682074074074074e-05,
+      "loss": 1.657,
+      "step": 29235
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.267911111111111e-05,
+      "loss": 1.8681,
+      "step": 29240
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.267614814814815e-05,
+      "loss": 1.7843,
+      "step": 29245
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.267318518518518e-05,
+      "loss": 1.744,
+      "step": 29250
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.267022222222222e-05,
+      "loss": 1.8594,
+      "step": 29255
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.266725925925926e-05,
+      "loss": 1.8169,
+      "step": 29260
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.26642962962963e-05,
+      "loss": 1.7859,
+      "step": 29265
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.266133333333333e-05,
+      "loss": 1.6908,
+      "step": 29270
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.265837037037037e-05,
+      "loss": 1.815,
+      "step": 29275
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.265540740740741e-05,
+      "loss": 1.7012,
+      "step": 29280
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.265244444444445e-05,
+      "loss": 1.7931,
+      "step": 29285
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.264948148148148e-05,
+      "loss": 1.761,
+      "step": 29290
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.264651851851852e-05,
+      "loss": 1.7149,
+      "step": 29295
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.264355555555556e-05,
+      "loss": 1.8547,
+      "step": 29300
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.26405925925926e-05,
+      "loss": 1.8682,
+      "step": 29305
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.263762962962963e-05,
+      "loss": 1.7597,
+      "step": 29310
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.263466666666667e-05,
+      "loss": 1.7386,
+      "step": 29315
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.2631703703703707e-05,
+      "loss": 1.7125,
+      "step": 29320
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.2628740740740745e-05,
+      "loss": 1.7047,
+      "step": 29325
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.262577777777778e-05,
+      "loss": 1.8127,
+      "step": 29330
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2622814814814816e-05,
+      "loss": 1.6905,
+      "step": 29335
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.261985185185185e-05,
+      "loss": 1.8548,
+      "step": 29340
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2616888888888894e-05,
+      "loss": 1.7607,
+      "step": 29345
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2613925925925926e-05,
+      "loss": 1.797,
+      "step": 29350
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2610962962962965e-05,
+      "loss": 1.661,
+      "step": 29355
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2608e-05,
+      "loss": 1.7261,
+      "step": 29360
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.260503703703704e-05,
+      "loss": 1.8225,
+      "step": 29365
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2602074074074074e-05,
+      "loss": 1.6582,
+      "step": 29370
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.259911111111111e-05,
+      "loss": 1.8001,
+      "step": 29375
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.2596148148148145e-05,
+      "loss": 1.7668,
+      "step": 29380
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.259318518518519e-05,
+      "loss": 1.7052,
+      "step": 29385
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.259022222222222e-05,
+      "loss": 1.8592,
+      "step": 29390
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.258725925925926e-05,
+      "loss": 1.7894,
+      "step": 29395
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.2584296296296294e-05,
+      "loss": 1.7153,
+      "step": 29400
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.258133333333333e-05,
+      "loss": 1.7068,
+      "step": 29405
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.257837037037037e-05,
+      "loss": 1.7645,
+      "step": 29410
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.257540740740741e-05,
+      "loss": 1.7908,
+      "step": 29415
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.257244444444444e-05,
+      "loss": 1.8108,
+      "step": 29420
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.256948148148148e-05,
+      "loss": 1.7633,
+      "step": 29425
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.256651851851852e-05,
+      "loss": 1.8961,
+      "step": 29430
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.256355555555556e-05,
+      "loss": 1.8886,
+      "step": 29435
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.256059259259259e-05,
+      "loss": 1.7095,
+      "step": 29440
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.255762962962963e-05,
+      "loss": 1.5951,
+      "step": 29445
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.255466666666667e-05,
+      "loss": 1.7643,
+      "step": 29450
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.255170370370371e-05,
+      "loss": 1.629,
+      "step": 29455
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.254874074074074e-05,
+      "loss": 1.8029,
+      "step": 29460
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.254577777777778e-05,
+      "loss": 1.5463,
+      "step": 29465
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.254281481481482e-05,
+      "loss": 1.6309,
+      "step": 29470
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.2539851851851856e-05,
+      "loss": 1.821,
+      "step": 29475
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.253688888888889e-05,
+      "loss": 1.6447,
+      "step": 29480
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.253392592592593e-05,
+      "loss": 1.7735,
+      "step": 29485
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.2530962962962966e-05,
+      "loss": 1.8529,
+      "step": 29490
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.2528000000000004e-05,
+      "loss": 1.7299,
+      "step": 29495
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.2525037037037036e-05,
+      "loss": 1.9048,
+      "step": 29500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2522074074074075e-05,
+      "loss": 1.6739,
+      "step": 29505
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2519111111111114e-05,
+      "loss": 1.7896,
+      "step": 29510
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.251614814814815e-05,
+      "loss": 1.7819,
+      "step": 29515
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2513185185185185e-05,
+      "loss": 1.774,
+      "step": 29520
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2510222222222224e-05,
+      "loss": 1.6446,
+      "step": 29525
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.250725925925926e-05,
+      "loss": 1.7756,
+      "step": 29530
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.25042962962963e-05,
+      "loss": 1.7319,
+      "step": 29535
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.2501333333333333e-05,
+      "loss": 1.9038,
+      "step": 29540
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.249837037037037e-05,
+      "loss": 1.7563,
+      "step": 29545
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.249540740740741e-05,
+      "loss": 1.8106,
+      "step": 29550
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.249244444444445e-05,
+      "loss": 1.8337,
+      "step": 29555
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.248948148148148e-05,
+      "loss": 1.8299,
+      "step": 29560
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.248651851851852e-05,
+      "loss": 1.7799,
+      "step": 29565
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.248355555555555e-05,
+      "loss": 1.7407,
+      "step": 29570
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.24805925925926e-05,
+      "loss": 1.7248,
+      "step": 29575
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.247762962962963e-05,
+      "loss": 1.8263,
+      "step": 29580
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.247466666666667e-05,
+      "loss": 1.7576,
+      "step": 29585
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.24717037037037e-05,
+      "loss": 1.8666,
+      "step": 29590
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.246874074074075e-05,
+      "loss": 1.725,
+      "step": 29595
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.246577777777778e-05,
+      "loss": 1.6978,
+      "step": 29600
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.246281481481482e-05,
+      "loss": 1.6997,
+      "step": 29605
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.245985185185185e-05,
+      "loss": 1.7745,
+      "step": 29610
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.2457481481481486e-05,
+      "loss": 1.7577,
+      "step": 29615
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.245451851851852e-05,
+      "loss": 1.8617,
+      "step": 29620
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.245155555555556e-05,
+      "loss": 1.8229,
+      "step": 29625
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.244859259259259e-05,
+      "loss": 1.7525,
+      "step": 29630
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.244562962962963e-05,
+      "loss": 1.8071,
+      "step": 29635
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.244266666666667e-05,
+      "loss": 1.9536,
+      "step": 29640
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2439703703703706e-05,
+      "loss": 1.7349,
+      "step": 29645
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.243674074074074e-05,
+      "loss": 1.7494,
+      "step": 29650
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2433777777777777e-05,
+      "loss": 1.7886,
+      "step": 29655
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2430814814814815e-05,
+      "loss": 1.7791,
+      "step": 29660
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2427851851851854e-05,
+      "loss": 1.7325,
+      "step": 29665
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.2424888888888886e-05,
+      "loss": 1.8129,
+      "step": 29670
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2421925925925925e-05,
+      "loss": 1.8459,
+      "step": 29675
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2418962962962964e-05,
+      "loss": 1.8665,
+      "step": 29680
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2416e-05,
+      "loss": 1.8058,
+      "step": 29685
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2413037037037035e-05,
+      "loss": 1.8305,
+      "step": 29690
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.2410074074074074e-05,
+      "loss": 1.8658,
+      "step": 29695
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.240711111111111e-05,
+      "loss": 1.8145,
+      "step": 29700
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.240414814814815e-05,
+      "loss": 1.9193,
+      "step": 29705
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.240118518518518e-05,
+      "loss": 1.8542,
+      "step": 29710
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.239822222222222e-05,
+      "loss": 1.9819,
+      "step": 29715
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.239525925925926e-05,
+      "loss": 1.7057,
+      "step": 29720
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.23922962962963e-05,
+      "loss": 1.7611,
+      "step": 29725
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.238933333333333e-05,
+      "loss": 1.7125,
+      "step": 29730
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.238637037037037e-05,
+      "loss": 1.7924,
+      "step": 29735
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.238340740740741e-05,
+      "loss": 1.6958,
+      "step": 29740
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.238044444444445e-05,
+      "loss": 1.7958,
+      "step": 29745
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.237748148148148e-05,
+      "loss": 1.7368,
+      "step": 29750
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.237451851851852e-05,
+      "loss": 1.777,
+      "step": 29755
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.237155555555556e-05,
+      "loss": 1.7095,
+      "step": 29760
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.23685925925926e-05,
+      "loss": 1.6856,
+      "step": 29765
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.236562962962963e-05,
+      "loss": 1.7666,
+      "step": 29770
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.236266666666667e-05,
+      "loss": 1.8457,
+      "step": 29775
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.2359703703703706e-05,
+      "loss": 1.7137,
+      "step": 29780
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2356740740740745e-05,
+      "loss": 1.6178,
+      "step": 29785
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.235377777777778e-05,
+      "loss": 1.8272,
+      "step": 29790
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2350814814814816e-05,
+      "loss": 1.8348,
+      "step": 29795
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.234785185185185e-05,
+      "loss": 1.7047,
+      "step": 29800
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2344888888888894e-05,
+      "loss": 1.8682,
+      "step": 29805
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2341925925925926e-05,
+      "loss": 1.7249,
+      "step": 29810
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2338962962962965e-05,
+      "loss": 1.7136,
+      "step": 29815
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2336e-05,
+      "loss": 1.7797,
+      "step": 29820
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.233303703703704e-05,
+      "loss": 1.7417,
+      "step": 29825
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2330074074074074e-05,
+      "loss": 1.7154,
+      "step": 29830
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.232711111111111e-05,
+      "loss": 1.8332,
+      "step": 29835
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.2324148148148145e-05,
+      "loss": 1.8463,
+      "step": 29840
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.232118518518519e-05,
+      "loss": 1.7096,
+      "step": 29845
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.231822222222222e-05,
+      "loss": 1.8184,
+      "step": 29850
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.231525925925926e-05,
+      "loss": 1.6954,
+      "step": 29855
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.2312296296296294e-05,
+      "loss": 1.6259,
+      "step": 29860
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.230933333333333e-05,
+      "loss": 1.759,
+      "step": 29865
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.230637037037037e-05,
+      "loss": 1.8674,
+      "step": 29870
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.230340740740741e-05,
+      "loss": 1.7582,
+      "step": 29875
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.230044444444444e-05,
+      "loss": 1.682,
+      "step": 29880
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.229748148148148e-05,
+      "loss": 1.856,
+      "step": 29885
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.229451851851852e-05,
+      "loss": 1.7464,
+      "step": 29890
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.229155555555556e-05,
+      "loss": 1.8445,
+      "step": 29895
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.228859259259259e-05,
+      "loss": 1.7165,
+      "step": 29900
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.228562962962963e-05,
+      "loss": 1.806,
+      "step": 29905
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.228266666666667e-05,
+      "loss": 1.736,
+      "step": 29910
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.227970370370371e-05,
+      "loss": 1.7156,
+      "step": 29915
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.227674074074074e-05,
+      "loss": 1.7369,
+      "step": 29920
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.227377777777778e-05,
+      "loss": 1.7532,
+      "step": 29925
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.227081481481482e-05,
+      "loss": 1.7148,
+      "step": 29930
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.2267851851851856e-05,
+      "loss": 1.6292,
+      "step": 29935
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.226488888888889e-05,
+      "loss": 1.7505,
+      "step": 29940
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.2261925925925927e-05,
+      "loss": 1.9294,
+      "step": 29945
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.2258962962962965e-05,
+      "loss": 1.772,
+      "step": 29950
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2256000000000004e-05,
+      "loss": 1.8099,
+      "step": 29955
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2253037037037036e-05,
+      "loss": 1.7111,
+      "step": 29960
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2250074074074075e-05,
+      "loss": 1.8077,
+      "step": 29965
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2247111111111114e-05,
+      "loss": 1.7618,
+      "step": 29970
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.224414814814815e-05,
+      "loss": 1.8128,
+      "step": 29975
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2241185185185185e-05,
+      "loss": 1.7034,
+      "step": 29980
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.2238222222222224e-05,
+      "loss": 1.8062,
+      "step": 29985
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.223525925925926e-05,
+      "loss": 1.7719,
+      "step": 29990
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.22322962962963e-05,
+      "loss": 1.7577,
+      "step": 29995
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.222933333333333e-05,
+      "loss": 1.6706,
+      "step": 30000
+    },
+    {
+      "epoch": 5.33,
+      "eval_loss": 1.6660683155059814,
+      "eval_rouge2_fmeasure": 0.1858,
+      "eval_rouge2_precision": 0.212,
+      "eval_rouge2_recall": 0.1754,
+      "eval_runtime": 40154.3823,
+      "eval_samples_per_second": 0.125,
+      "eval_steps_per_second": 0.062,
+      "step": 30000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.222637037037037e-05,
+      "loss": 1.7674,
+      "step": 30005
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.222340740740741e-05,
+      "loss": 1.7299,
+      "step": 30010
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.222044444444445e-05,
+      "loss": 1.7153,
+      "step": 30015
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.221748148148148e-05,
+      "loss": 1.814,
+      "step": 30020
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.221451851851852e-05,
+      "loss": 1.7316,
+      "step": 30025
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.221155555555555e-05,
+      "loss": 1.7726,
+      "step": 30030
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.22085925925926e-05,
+      "loss": 1.6476,
+      "step": 30035
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.220562962962963e-05,
+      "loss": 1.6778,
+      "step": 30040
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.220266666666667e-05,
+      "loss": 1.8191,
+      "step": 30045
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.21997037037037e-05,
+      "loss": 1.7427,
+      "step": 30050
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.219674074074075e-05,
+      "loss": 1.8601,
+      "step": 30055
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.219377777777778e-05,
+      "loss": 1.6535,
+      "step": 30060
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.219081481481482e-05,
+      "loss": 1.8686,
+      "step": 30065
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.218785185185185e-05,
+      "loss": 1.7833,
+      "step": 30070
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2184888888888895e-05,
+      "loss": 1.6951,
+      "step": 30075
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.218192592592593e-05,
+      "loss": 1.6929,
+      "step": 30080
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2178962962962966e-05,
+      "loss": 1.7882,
+      "step": 30085
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2176e-05,
+      "loss": 1.7619,
+      "step": 30090
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.217303703703704e-05,
+      "loss": 1.6784,
+      "step": 30095
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2170074074074076e-05,
+      "loss": 1.7072,
+      "step": 30100
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2167111111111115e-05,
+      "loss": 1.9223,
+      "step": 30105
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.216414814814815e-05,
+      "loss": 1.5836,
+      "step": 30110
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2161185185185186e-05,
+      "loss": 1.7562,
+      "step": 30115
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.2158222222222224e-05,
+      "loss": 1.7978,
+      "step": 30120
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.215525925925926e-05,
+      "loss": 1.8594,
+      "step": 30125
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2152296296296295e-05,
+      "loss": 1.777,
+      "step": 30130
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2149333333333334e-05,
+      "loss": 1.8381,
+      "step": 30135
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.214637037037037e-05,
+      "loss": 1.6528,
+      "step": 30140
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.214340740740741e-05,
+      "loss": 1.7207,
+      "step": 30145
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.2140444444444444e-05,
+      "loss": 1.8969,
+      "step": 30150
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.213748148148148e-05,
+      "loss": 1.7894,
+      "step": 30155
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.213451851851852e-05,
+      "loss": 1.7771,
+      "step": 30160
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.213155555555556e-05,
+      "loss": 1.6601,
+      "step": 30165
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.212859259259259e-05,
+      "loss": 1.7616,
+      "step": 30170
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.212562962962963e-05,
+      "loss": 1.9105,
+      "step": 30175
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.212266666666667e-05,
+      "loss": 1.8659,
+      "step": 30180
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.211970370370371e-05,
+      "loss": 1.7576,
+      "step": 30185
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.211674074074074e-05,
+      "loss": 1.7892,
+      "step": 30190
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.211377777777778e-05,
+      "loss": 1.6566,
+      "step": 30195
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.211081481481482e-05,
+      "loss": 1.8304,
+      "step": 30200
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.210785185185186e-05,
+      "loss": 1.7859,
+      "step": 30205
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.210488888888889e-05,
+      "loss": 1.8121,
+      "step": 30210
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.210192592592593e-05,
+      "loss": 1.764,
+      "step": 30215
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.209896296296297e-05,
+      "loss": 1.7455,
+      "step": 30220
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.2096000000000006e-05,
+      "loss": 1.7184,
+      "step": 30225
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.209303703703704e-05,
+      "loss": 1.8512,
+      "step": 30230
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.209007407407408e-05,
+      "loss": 1.7478,
+      "step": 30235
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2087111111111115e-05,
+      "loss": 1.8058,
+      "step": 30240
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2084148148148154e-05,
+      "loss": 1.6179,
+      "step": 30245
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2081185185185186e-05,
+      "loss": 1.6414,
+      "step": 30250
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2078222222222225e-05,
+      "loss": 1.745,
+      "step": 30255
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.207525925925926e-05,
+      "loss": 1.6527,
+      "step": 30260
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.20722962962963e-05,
+      "loss": 1.7201,
+      "step": 30265
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2069333333333335e-05,
+      "loss": 1.7982,
+      "step": 30270
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2066370370370374e-05,
+      "loss": 1.7573,
+      "step": 30275
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2063407407407406e-05,
+      "loss": 1.8131,
+      "step": 30280
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.206044444444445e-05,
+      "loss": 1.7896,
+      "step": 30285
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.205748148148148e-05,
+      "loss": 1.7856,
+      "step": 30290
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.205451851851852e-05,
+      "loss": 1.8124,
+      "step": 30295
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2051555555555554e-05,
+      "loss": 1.9074,
+      "step": 30300
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.20485925925926e-05,
+      "loss": 1.8047,
+      "step": 30305
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.204562962962963e-05,
+      "loss": 1.6993,
+      "step": 30310
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.204266666666667e-05,
+      "loss": 1.694,
+      "step": 30315
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.20397037037037e-05,
+      "loss": 1.695,
+      "step": 30320
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.203674074074075e-05,
+      "loss": 1.7901,
+      "step": 30325
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.203377777777778e-05,
+      "loss": 1.8551,
+      "step": 30330
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.203081481481482e-05,
+      "loss": 1.9053,
+      "step": 30335
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.202785185185185e-05,
+      "loss": 1.8342,
+      "step": 30340
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.202488888888889e-05,
+      "loss": 1.8835,
+      "step": 30345
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.202192592592593e-05,
+      "loss": 1.7006,
+      "step": 30350
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.201896296296297e-05,
+      "loss": 1.8549,
+      "step": 30355
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.2016e-05,
+      "loss": 1.7715,
+      "step": 30360
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.201303703703704e-05,
+      "loss": 1.7962,
+      "step": 30365
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.201007407407408e-05,
+      "loss": 1.7828,
+      "step": 30370
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.200711111111111e-05,
+      "loss": 1.7275,
+      "step": 30375
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.200414814814815e-05,
+      "loss": 1.7349,
+      "step": 30380
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.200118518518519e-05,
+      "loss": 1.7595,
+      "step": 30385
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.1998222222222226e-05,
+      "loss": 1.8099,
+      "step": 30390
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.199525925925926e-05,
+      "loss": 1.8064,
+      "step": 30395
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.19922962962963e-05,
+      "loss": 1.8302,
+      "step": 30400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1989333333333336e-05,
+      "loss": 1.6796,
+      "step": 30405
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1986370370370374e-05,
+      "loss": 1.7358,
+      "step": 30410
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1983407407407406e-05,
+      "loss": 1.8466,
+      "step": 30415
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1980444444444445e-05,
+      "loss": 1.9074,
+      "step": 30420
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.197748148148148e-05,
+      "loss": 1.8118,
+      "step": 30425
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.197451851851852e-05,
+      "loss": 1.6472,
+      "step": 30430
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1971555555555555e-05,
+      "loss": 1.7588,
+      "step": 30435
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1968592592592594e-05,
+      "loss": 1.8209,
+      "step": 30440
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1965629629629626e-05,
+      "loss": 1.8238,
+      "step": 30445
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.196266666666667e-05,
+      "loss": 1.7621,
+      "step": 30450
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.1959703703703703e-05,
+      "loss": 1.8016,
+      "step": 30455
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.195674074074074e-05,
+      "loss": 1.6624,
+      "step": 30460
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.1953777777777774e-05,
+      "loss": 1.7246,
+      "step": 30465
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.195081481481482e-05,
+      "loss": 1.7926,
+      "step": 30470
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.194785185185185e-05,
+      "loss": 1.7122,
+      "step": 30475
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.194488888888889e-05,
+      "loss": 1.742,
+      "step": 30480
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.194192592592592e-05,
+      "loss": 1.6996,
+      "step": 30485
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.193896296296296e-05,
+      "loss": 1.8148,
+      "step": 30490
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.1936e-05,
+      "loss": 1.7629,
+      "step": 30495
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.193303703703704e-05,
+      "loss": 1.6853,
+      "step": 30500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.193007407407407e-05,
+      "loss": 1.8395,
+      "step": 30505
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.192711111111111e-05,
+      "loss": 1.8557,
+      "step": 30510
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.192414814814815e-05,
+      "loss": 1.7458,
+      "step": 30515
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.192118518518519e-05,
+      "loss": 1.8935,
+      "step": 30520
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.191822222222222e-05,
+      "loss": 1.6818,
+      "step": 30525
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.191525925925926e-05,
+      "loss": 1.9193,
+      "step": 30530
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.19122962962963e-05,
+      "loss": 1.7242,
+      "step": 30535
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1909333333333336e-05,
+      "loss": 1.7063,
+      "step": 30540
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.190637037037037e-05,
+      "loss": 1.725,
+      "step": 30545
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.190340740740741e-05,
+      "loss": 1.7652,
+      "step": 30550
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1900444444444446e-05,
+      "loss": 1.821,
+      "step": 30555
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1897481481481485e-05,
+      "loss": 1.8678,
+      "step": 30560
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.189451851851852e-05,
+      "loss": 1.6798,
+      "step": 30565
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.1891555555555556e-05,
+      "loss": 1.7284,
+      "step": 30570
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.1888592592592595e-05,
+      "loss": 1.7557,
+      "step": 30575
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.1885629629629633e-05,
+      "loss": 1.7892,
+      "step": 30580
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.1882666666666665e-05,
+      "loss": 1.7837,
+      "step": 30585
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.1879703703703704e-05,
+      "loss": 1.7782,
+      "step": 30590
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.187674074074074e-05,
+      "loss": 1.8331,
+      "step": 30595
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.187377777777778e-05,
+      "loss": 1.842,
+      "step": 30600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.1870814814814814e-05,
+      "loss": 1.7656,
+      "step": 30605
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.186785185185185e-05,
+      "loss": 1.828,
+      "step": 30610
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.186488888888889e-05,
+      "loss": 1.8331,
+      "step": 30615
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.186192592592593e-05,
+      "loss": 1.7244,
+      "step": 30620
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.185896296296296e-05,
+      "loss": 1.8241,
+      "step": 30625
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1856e-05,
+      "loss": 1.7786,
+      "step": 30630
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.185303703703704e-05,
+      "loss": 1.6925,
+      "step": 30635
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.185007407407408e-05,
+      "loss": 1.7422,
+      "step": 30640
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.184711111111111e-05,
+      "loss": 1.7634,
+      "step": 30645
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.184414814814815e-05,
+      "loss": 1.6899,
+      "step": 30650
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.184118518518518e-05,
+      "loss": 1.9626,
+      "step": 30655
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.183822222222223e-05,
+      "loss": 1.6883,
+      "step": 30660
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.183525925925926e-05,
+      "loss": 1.7763,
+      "step": 30665
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.18322962962963e-05,
+      "loss": 1.8264,
+      "step": 30670
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.182933333333333e-05,
+      "loss": 1.7097,
+      "step": 30675
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1826370370370376e-05,
+      "loss": 1.7946,
+      "step": 30680
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.182340740740741e-05,
+      "loss": 1.7882,
+      "step": 30685
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.182044444444445e-05,
+      "loss": 1.788,
+      "step": 30690
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.181748148148148e-05,
+      "loss": 1.8282,
+      "step": 30695
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1814518518518524e-05,
+      "loss": 1.8171,
+      "step": 30700
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1811555555555557e-05,
+      "loss": 1.8122,
+      "step": 30705
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1808592592592595e-05,
+      "loss": 1.8627,
+      "step": 30710
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.180562962962963e-05,
+      "loss": 1.6132,
+      "step": 30715
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1802666666666666e-05,
+      "loss": 1.8072,
+      "step": 30720
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1799703703703705e-05,
+      "loss": 1.7519,
+      "step": 30725
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1796740740740744e-05,
+      "loss": 1.8369,
+      "step": 30730
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1793777777777776e-05,
+      "loss": 1.7252,
+      "step": 30735
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1790814814814815e-05,
+      "loss": 1.725,
+      "step": 30740
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.1787851851851854e-05,
+      "loss": 1.7855,
+      "step": 30745
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.178488888888889e-05,
+      "loss": 1.8013,
+      "step": 30750
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.1781925925925924e-05,
+      "loss": 1.6371,
+      "step": 30755
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.177896296296296e-05,
+      "loss": 1.9044,
+      "step": 30760
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.1776e-05,
+      "loss": 1.7048,
+      "step": 30765
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.177303703703704e-05,
+      "loss": 1.8366,
+      "step": 30770
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.177007407407407e-05,
+      "loss": 1.7882,
+      "step": 30775
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.176711111111111e-05,
+      "loss": 1.7665,
+      "step": 30780
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.176414814814815e-05,
+      "loss": 1.8495,
+      "step": 30785
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.176118518518519e-05,
+      "loss": 1.8446,
+      "step": 30790
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.175822222222222e-05,
+      "loss": 1.6982,
+      "step": 30795
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.175525925925926e-05,
+      "loss": 1.7092,
+      "step": 30800
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.17522962962963e-05,
+      "loss": 1.7658,
+      "step": 30805
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.174933333333334e-05,
+      "loss": 1.7615,
+      "step": 30810
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.174637037037037e-05,
+      "loss": 1.7814,
+      "step": 30815
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.174340740740741e-05,
+      "loss": 1.7661,
+      "step": 30820
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.174044444444445e-05,
+      "loss": 1.8778,
+      "step": 30825
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.1737481481481486e-05,
+      "loss": 1.6932,
+      "step": 30830
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.173451851851852e-05,
+      "loss": 1.7178,
+      "step": 30835
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.173155555555556e-05,
+      "loss": 1.7099,
+      "step": 30840
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.1728592592592596e-05,
+      "loss": 1.9026,
+      "step": 30845
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.1725629629629635e-05,
+      "loss": 1.7083,
+      "step": 30850
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.172266666666667e-05,
+      "loss": 1.7845,
+      "step": 30855
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1719703703703706e-05,
+      "loss": 1.7638,
+      "step": 30860
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1716740740740745e-05,
+      "loss": 1.7684,
+      "step": 30865
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1713777777777783e-05,
+      "loss": 1.6788,
+      "step": 30870
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1710814814814815e-05,
+      "loss": 1.867,
+      "step": 30875
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1707851851851854e-05,
+      "loss": 1.7258,
+      "step": 30880
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1704888888888886e-05,
+      "loss": 1.7607,
+      "step": 30885
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.170192592592593e-05,
+      "loss": 1.6763,
+      "step": 30890
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1698962962962964e-05,
+      "loss": 1.7192,
+      "step": 30895
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1696e-05,
+      "loss": 1.7689,
+      "step": 30900
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1693037037037035e-05,
+      "loss": 1.7727,
+      "step": 30905
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.169007407407408e-05,
+      "loss": 1.7891,
+      "step": 30910
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.168711111111111e-05,
+      "loss": 1.821,
+      "step": 30915
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.168414814814815e-05,
+      "loss": 1.6491,
+      "step": 30920
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.168118518518518e-05,
+      "loss": 1.8538,
+      "step": 30925
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.167822222222223e-05,
+      "loss": 1.8045,
+      "step": 30930
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.167525925925926e-05,
+      "loss": 1.7913,
+      "step": 30935
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.16722962962963e-05,
+      "loss": 1.7949,
+      "step": 30940
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.166933333333333e-05,
+      "loss": 1.7217,
+      "step": 30945
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.166637037037037e-05,
+      "loss": 1.975,
+      "step": 30950
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.166340740740741e-05,
+      "loss": 1.6561,
+      "step": 30955
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.166044444444445e-05,
+      "loss": 1.9199,
+      "step": 30960
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.165748148148148e-05,
+      "loss": 1.7394,
+      "step": 30965
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.165451851851852e-05,
+      "loss": 1.8672,
+      "step": 30970
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.165155555555556e-05,
+      "loss": 1.6846,
+      "step": 30975
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.16485925925926e-05,
+      "loss": 1.7104,
+      "step": 30980
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.164562962962963e-05,
+      "loss": 1.7971,
+      "step": 30985
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.164266666666667e-05,
+      "loss": 1.6928,
+      "step": 30990
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1639703703703707e-05,
+      "loss": 1.7726,
+      "step": 30995
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1636740740740745e-05,
+      "loss": 1.7499,
+      "step": 31000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.163377777777778e-05,
+      "loss": 1.8686,
+      "step": 31005
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1630814814814816e-05,
+      "loss": 1.7976,
+      "step": 31010
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1627851851851855e-05,
+      "loss": 1.8947,
+      "step": 31015
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1624888888888894e-05,
+      "loss": 1.838,
+      "step": 31020
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1621925925925926e-05,
+      "loss": 1.7313,
+      "step": 31025
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1618962962962965e-05,
+      "loss": 1.8511,
+      "step": 31030
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1616000000000004e-05,
+      "loss": 1.6926,
+      "step": 31035
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.161303703703704e-05,
+      "loss": 1.8433,
+      "step": 31040
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1610666666666665e-05,
+      "loss": 1.7327,
+      "step": 31045
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1607703703703704e-05,
+      "loss": 1.7641,
+      "step": 31050
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.160474074074074e-05,
+      "loss": 1.8921,
+      "step": 31055
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.160177777777778e-05,
+      "loss": 1.8823,
+      "step": 31060
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1598814814814814e-05,
+      "loss": 1.8443,
+      "step": 31065
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.159585185185185e-05,
+      "loss": 1.8559,
+      "step": 31070
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.159288888888889e-05,
+      "loss": 1.7812,
+      "step": 31075
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.158992592592593e-05,
+      "loss": 1.78,
+      "step": 31080
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.158696296296296e-05,
+      "loss": 1.694,
+      "step": 31085
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.1584e-05,
+      "loss": 1.7249,
+      "step": 31090
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.158103703703704e-05,
+      "loss": 1.8089,
+      "step": 31095
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.157807407407408e-05,
+      "loss": 1.815,
+      "step": 31100
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.157511111111111e-05,
+      "loss": 1.7734,
+      "step": 31105
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.157214814814815e-05,
+      "loss": 1.6808,
+      "step": 31110
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.156918518518518e-05,
+      "loss": 1.7346,
+      "step": 31115
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.156622222222223e-05,
+      "loss": 1.8647,
+      "step": 31120
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.156325925925926e-05,
+      "loss": 1.6936,
+      "step": 31125
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.15602962962963e-05,
+      "loss": 1.6982,
+      "step": 31130
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.155733333333333e-05,
+      "loss": 1.8464,
+      "step": 31135
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1554370370370376e-05,
+      "loss": 1.5864,
+      "step": 31140
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.155140740740741e-05,
+      "loss": 1.7326,
+      "step": 31145
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.154844444444445e-05,
+      "loss": 1.7698,
+      "step": 31150
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.154548148148148e-05,
+      "loss": 1.7541,
+      "step": 31155
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1542518518518524e-05,
+      "loss": 1.7638,
+      "step": 31160
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1539555555555556e-05,
+      "loss": 1.8659,
+      "step": 31165
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1536592592592595e-05,
+      "loss": 1.7968,
+      "step": 31170
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.153362962962963e-05,
+      "loss": 1.7329,
+      "step": 31175
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1530666666666666e-05,
+      "loss": 1.7213,
+      "step": 31180
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1527703703703705e-05,
+      "loss": 1.7661,
+      "step": 31185
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.1524740740740744e-05,
+      "loss": 1.7155,
+      "step": 31190
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1521777777777776e-05,
+      "loss": 1.8216,
+      "step": 31195
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1518814814814815e-05,
+      "loss": 1.701,
+      "step": 31200
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1515851851851853e-05,
+      "loss": 1.8173,
+      "step": 31205
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.151288888888889e-05,
+      "loss": 1.7842,
+      "step": 31210
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1509925925925924e-05,
+      "loss": 1.6522,
+      "step": 31215
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.150696296296296e-05,
+      "loss": 1.6995,
+      "step": 31220
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.1504e-05,
+      "loss": 1.6889,
+      "step": 31225
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.150103703703704e-05,
+      "loss": 1.7295,
+      "step": 31230
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.149807407407407e-05,
+      "loss": 1.7438,
+      "step": 31235
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.149511111111111e-05,
+      "loss": 1.8172,
+      "step": 31240
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.149214814814815e-05,
+      "loss": 1.858,
+      "step": 31245
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.148918518518519e-05,
+      "loss": 1.8805,
+      "step": 31250
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.148622222222222e-05,
+      "loss": 1.773,
+      "step": 31255
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.148325925925926e-05,
+      "loss": 1.662,
+      "step": 31260
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.14802962962963e-05,
+      "loss": 1.71,
+      "step": 31265
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.147733333333334e-05,
+      "loss": 1.7215,
+      "step": 31270
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.147437037037037e-05,
+      "loss": 1.683,
+      "step": 31275
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.147140740740741e-05,
+      "loss": 1.7385,
+      "step": 31280
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.146844444444445e-05,
+      "loss": 1.8453,
+      "step": 31285
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.1465481481481486e-05,
+      "loss": 1.8511,
+      "step": 31290
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.146251851851852e-05,
+      "loss": 1.743,
+      "step": 31295
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.145955555555556e-05,
+      "loss": 1.7316,
+      "step": 31300
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1456592592592596e-05,
+      "loss": 1.7029,
+      "step": 31305
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1453629629629635e-05,
+      "loss": 1.9747,
+      "step": 31310
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.145066666666667e-05,
+      "loss": 1.7667,
+      "step": 31315
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1447703703703706e-05,
+      "loss": 1.6564,
+      "step": 31320
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1444740740740745e-05,
+      "loss": 1.7722,
+      "step": 31325
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.144177777777778e-05,
+      "loss": 1.7273,
+      "step": 31330
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1438814814814815e-05,
+      "loss": 1.7702,
+      "step": 31335
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1435851851851854e-05,
+      "loss": 1.7738,
+      "step": 31340
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1432888888888886e-05,
+      "loss": 1.8663,
+      "step": 31345
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.142992592592593e-05,
+      "loss": 1.7274,
+      "step": 31350
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.1426962962962964e-05,
+      "loss": 1.8418,
+      "step": 31355
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.1424e-05,
+      "loss": 1.8298,
+      "step": 31360
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.1421037037037035e-05,
+      "loss": 1.9037,
+      "step": 31365
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.141807407407408e-05,
+      "loss": 1.7335,
+      "step": 31370
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.141511111111111e-05,
+      "loss": 1.6693,
+      "step": 31375
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.141214814814815e-05,
+      "loss": 1.79,
+      "step": 31380
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.140918518518518e-05,
+      "loss": 1.7368,
+      "step": 31385
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.140622222222223e-05,
+      "loss": 1.7137,
+      "step": 31390
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.140325925925926e-05,
+      "loss": 1.8864,
+      "step": 31395
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.14002962962963e-05,
+      "loss": 1.7851,
+      "step": 31400
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.139733333333333e-05,
+      "loss": 1.7961,
+      "step": 31405
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.139437037037037e-05,
+      "loss": 1.7735,
+      "step": 31410
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.139140740740741e-05,
+      "loss": 1.7106,
+      "step": 31415
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.138844444444445e-05,
+      "loss": 1.7121,
+      "step": 31420
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.138548148148148e-05,
+      "loss": 1.8201,
+      "step": 31425
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.138251851851852e-05,
+      "loss": 1.7026,
+      "step": 31430
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.137955555555556e-05,
+      "loss": 1.7519,
+      "step": 31435
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.13765925925926e-05,
+      "loss": 1.7989,
+      "step": 31440
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.137362962962963e-05,
+      "loss": 1.7369,
+      "step": 31445
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.137066666666667e-05,
+      "loss": 1.7181,
+      "step": 31450
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.1367703703703706e-05,
+      "loss": 1.7591,
+      "step": 31455
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.1364740740740745e-05,
+      "loss": 1.8078,
+      "step": 31460
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.136177777777778e-05,
+      "loss": 1.6645,
+      "step": 31465
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.1358814814814816e-05,
+      "loss": 1.7519,
+      "step": 31470
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1355851851851855e-05,
+      "loss": 1.7399,
+      "step": 31475
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1352888888888894e-05,
+      "loss": 1.8531,
+      "step": 31480
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1349925925925926e-05,
+      "loss": 1.8759,
+      "step": 31485
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1346962962962965e-05,
+      "loss": 1.7434,
+      "step": 31490
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1344000000000003e-05,
+      "loss": 1.7048,
+      "step": 31495
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.134103703703704e-05,
+      "loss": 1.7197,
+      "step": 31500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.1338074074074074e-05,
+      "loss": 1.7388,
+      "step": 31505
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.133511111111111e-05,
+      "loss": 1.6485,
+      "step": 31510
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.133214814814815e-05,
+      "loss": 1.7801,
+      "step": 31515
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.132918518518519e-05,
+      "loss": 1.7048,
+      "step": 31520
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.132622222222222e-05,
+      "loss": 1.7085,
+      "step": 31525
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.132325925925926e-05,
+      "loss": 1.7465,
+      "step": 31530
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.13202962962963e-05,
+      "loss": 1.6885,
+      "step": 31535
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.131733333333334e-05,
+      "loss": 1.7921,
+      "step": 31540
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.131437037037037e-05,
+      "loss": 1.7291,
+      "step": 31545
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.131140740740741e-05,
+      "loss": 1.7648,
+      "step": 31550
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.130844444444445e-05,
+      "loss": 1.8669,
+      "step": 31555
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.130548148148149e-05,
+      "loss": 1.82,
+      "step": 31560
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.130251851851852e-05,
+      "loss": 1.8781,
+      "step": 31565
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.129955555555556e-05,
+      "loss": 1.7801,
+      "step": 31570
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.129659259259259e-05,
+      "loss": 1.8379,
+      "step": 31575
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.1293629629629636e-05,
+      "loss": 1.8033,
+      "step": 31580
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.129066666666667e-05,
+      "loss": 1.684,
+      "step": 31585
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.128770370370371e-05,
+      "loss": 1.8527,
+      "step": 31590
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.128474074074074e-05,
+      "loss": 1.8069,
+      "step": 31595
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.128177777777778e-05,
+      "loss": 1.847,
+      "step": 31600
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.127881481481482e-05,
+      "loss": 1.7497,
+      "step": 31605
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.127585185185185e-05,
+      "loss": 1.7918,
+      "step": 31610
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.127288888888889e-05,
+      "loss": 1.7614,
+      "step": 31615
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1269925925925927e-05,
+      "loss": 1.7639,
+      "step": 31620
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1266962962962965e-05,
+      "loss": 1.7072,
+      "step": 31625
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1264e-05,
+      "loss": 1.693,
+      "step": 31630
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1261037037037036e-05,
+      "loss": 1.7512,
+      "step": 31635
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 3.1258074074074075e-05,
+      "loss": 1.7095,
+      "step": 31640
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1255111111111114e-05,
+      "loss": 1.7813,
+      "step": 31645
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1252148148148146e-05,
+      "loss": 1.7179,
+      "step": 31650
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1249185185185185e-05,
+      "loss": 1.6958,
+      "step": 31655
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1246222222222224e-05,
+      "loss": 1.8054,
+      "step": 31660
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.124325925925926e-05,
+      "loss": 1.6277,
+      "step": 31665
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.1240296296296294e-05,
+      "loss": 1.8798,
+      "step": 31670
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.123733333333333e-05,
+      "loss": 1.7651,
+      "step": 31675
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.123437037037037e-05,
+      "loss": 1.8662,
+      "step": 31680
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.123140740740741e-05,
+      "loss": 1.8777,
+      "step": 31685
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.122844444444444e-05,
+      "loss": 1.8076,
+      "step": 31690
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.122548148148148e-05,
+      "loss": 1.7202,
+      "step": 31695
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.122251851851852e-05,
+      "loss": 1.6982,
+      "step": 31700
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.121955555555556e-05,
+      "loss": 1.7004,
+      "step": 31705
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.121659259259259e-05,
+      "loss": 1.9044,
+      "step": 31710
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.121362962962963e-05,
+      "loss": 1.7976,
+      "step": 31715
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.121066666666667e-05,
+      "loss": 1.844,
+      "step": 31720
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.120770370370371e-05,
+      "loss": 1.73,
+      "step": 31725
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.120474074074074e-05,
+      "loss": 1.7927,
+      "step": 31730
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.120177777777778e-05,
+      "loss": 1.7507,
+      "step": 31735
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.119881481481481e-05,
+      "loss": 1.7915,
+      "step": 31740
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.1195851851851856e-05,
+      "loss": 1.7597,
+      "step": 31745
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.119288888888889e-05,
+      "loss": 1.7875,
+      "step": 31750
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.118992592592593e-05,
+      "loss": 1.7953,
+      "step": 31755
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.118696296296296e-05,
+      "loss": 1.8576,
+      "step": 31760
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1184000000000005e-05,
+      "loss": 1.8272,
+      "step": 31765
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.118103703703704e-05,
+      "loss": 1.6859,
+      "step": 31770
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1178074074074076e-05,
+      "loss": 1.8597,
+      "step": 31775
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.117511111111111e-05,
+      "loss": 1.7329,
+      "step": 31780
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1172148148148154e-05,
+      "loss": 1.7861,
+      "step": 31785
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1169185185185186e-05,
+      "loss": 1.7981,
+      "step": 31790
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1166222222222224e-05,
+      "loss": 1.7985,
+      "step": 31795
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1163259259259256e-05,
+      "loss": 1.7294,
+      "step": 31800
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 3.1160296296296295e-05,
+      "loss": 1.7675,
+      "step": 31805
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1157333333333334e-05,
+      "loss": 1.813,
+      "step": 31810
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.115437037037037e-05,
+      "loss": 1.6723,
+      "step": 31815
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1151407407407405e-05,
+      "loss": 1.7141,
+      "step": 31820
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1148444444444444e-05,
+      "loss": 1.7696,
+      "step": 31825
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.114548148148148e-05,
+      "loss": 1.8026,
+      "step": 31830
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.114251851851852e-05,
+      "loss": 1.831,
+      "step": 31835
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.1139555555555553e-05,
+      "loss": 1.835,
+      "step": 31840
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.113659259259259e-05,
+      "loss": 1.8629,
+      "step": 31845
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.113362962962963e-05,
+      "loss": 1.7756,
+      "step": 31850
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.113066666666667e-05,
+      "loss": 1.6591,
+      "step": 31855
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.11277037037037e-05,
+      "loss": 1.827,
+      "step": 31860
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.112474074074074e-05,
+      "loss": 1.8058,
+      "step": 31865
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.112177777777778e-05,
+      "loss": 1.8309,
+      "step": 31870
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.111881481481482e-05,
+      "loss": 1.7748,
+      "step": 31875
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.111585185185185e-05,
+      "loss": 1.7768,
+      "step": 31880
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.111288888888889e-05,
+      "loss": 1.6996,
+      "step": 31885
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.110992592592593e-05,
+      "loss": 1.8126,
+      "step": 31890
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.110696296296297e-05,
+      "loss": 1.7201,
+      "step": 31895
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.1104e-05,
+      "loss": 1.8051,
+      "step": 31900
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.110103703703704e-05,
+      "loss": 1.7476,
+      "step": 31905
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.109807407407408e-05,
+      "loss": 1.6694,
+      "step": 31910
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.1095111111111115e-05,
+      "loss": 1.803,
+      "step": 31915
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 3.109214814814815e-05,
+      "loss": 1.9136,
+      "step": 31920
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1089185185185186e-05,
+      "loss": 1.8455,
+      "step": 31925
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1086222222222225e-05,
+      "loss": 1.7325,
+      "step": 31930
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1083259259259264e-05,
+      "loss": 1.7701,
+      "step": 31935
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1080296296296296e-05,
+      "loss": 1.8716,
+      "step": 31940
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1077333333333335e-05,
+      "loss": 1.8155,
+      "step": 31945
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1074370370370374e-05,
+      "loss": 1.8082,
+      "step": 31950
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.107140740740741e-05,
+      "loss": 1.6301,
+      "step": 31955
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1068444444444445e-05,
+      "loss": 1.7759,
+      "step": 31960
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.106548148148148e-05,
+      "loss": 1.768,
+      "step": 31965
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.1062518518518515e-05,
+      "loss": 1.7293,
+      "step": 31970
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.105955555555556e-05,
+      "loss": 1.8429,
+      "step": 31975
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.105659259259259e-05,
+      "loss": 1.8491,
+      "step": 31980
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.105362962962963e-05,
+      "loss": 1.7755,
+      "step": 31985
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.1050666666666664e-05,
+      "loss": 1.8813,
+      "step": 31990
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.104770370370371e-05,
+      "loss": 1.756,
+      "step": 31995
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.104474074074074e-05,
+      "loss": 1.9161,
+      "step": 32000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.104177777777778e-05,
+      "loss": 1.8715,
+      "step": 32005
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.103881481481481e-05,
+      "loss": 1.7547,
+      "step": 32010
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.103585185185186e-05,
+      "loss": 1.9679,
+      "step": 32015
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.103288888888889e-05,
+      "loss": 1.7302,
+      "step": 32020
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.102992592592593e-05,
+      "loss": 1.8798,
+      "step": 32025
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 3.102696296296296e-05,
+      "loss": 1.8049,
+      "step": 32030
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.1024e-05,
+      "loss": 1.8419,
+      "step": 32035
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.102103703703704e-05,
+      "loss": 1.6936,
+      "step": 32040
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.101807407407408e-05,
+      "loss": 1.6731,
+      "step": 32045
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.101511111111111e-05,
+      "loss": 1.7687,
+      "step": 32050
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.101214814814815e-05,
+      "loss": 1.7636,
+      "step": 32055
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.100918518518519e-05,
+      "loss": 1.8046,
+      "step": 32060
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.1006222222222226e-05,
+      "loss": 1.8165,
+      "step": 32065
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.100325925925926e-05,
+      "loss": 1.6993,
+      "step": 32070
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.10002962962963e-05,
+      "loss": 1.7324,
+      "step": 32075
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.0997333333333336e-05,
+      "loss": 1.8225,
+      "step": 32080
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.0994370370370374e-05,
+      "loss": 1.8352,
+      "step": 32085
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.0991407407407406e-05,
+      "loss": 1.7509,
+      "step": 32090
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0988444444444445e-05,
+      "loss": 1.8205,
+      "step": 32095
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0985481481481484e-05,
+      "loss": 1.7646,
+      "step": 32100
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.098251851851852e-05,
+      "loss": 1.738,
+      "step": 32105
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0979555555555555e-05,
+      "loss": 1.8839,
+      "step": 32110
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0976592592592594e-05,
+      "loss": 1.8064,
+      "step": 32115
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.097362962962963e-05,
+      "loss": 1.7439,
+      "step": 32120
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.097066666666667e-05,
+      "loss": 1.6806,
+      "step": 32125
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.0967703703703704e-05,
+      "loss": 1.8937,
+      "step": 32130
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.096474074074074e-05,
+      "loss": 1.6911,
+      "step": 32135
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.096177777777778e-05,
+      "loss": 1.835,
+      "step": 32140
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 3.095881481481482e-05,
+      "loss": 1.7977,
+      "step": 32145
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.095585185185185e-05,
+      "loss": 1.7236,
+      "step": 32150
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.095288888888889e-05,
+      "loss": 1.8154,
+      "step": 32155
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.094992592592593e-05,
+      "loss": 1.7937,
+      "step": 32160
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.094696296296297e-05,
+      "loss": 1.7209,
+      "step": 32165
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.0944e-05,
+      "loss": 1.8279,
+      "step": 32170
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.094103703703704e-05,
+      "loss": 1.6862,
+      "step": 32175
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.093807407407408e-05,
+      "loss": 1.9646,
+      "step": 32180
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.093511111111112e-05,
+      "loss": 1.8541,
+      "step": 32185
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.093214814814815e-05,
+      "loss": 1.6799,
+      "step": 32190
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.092918518518519e-05,
+      "loss": 1.6851,
+      "step": 32195
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.092622222222222e-05,
+      "loss": 1.7665,
+      "step": 32200
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0923259259259266e-05,
+      "loss": 1.8934,
+      "step": 32205
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.09202962962963e-05,
+      "loss": 1.7695,
+      "step": 32210
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0917333333333336e-05,
+      "loss": 1.7963,
+      "step": 32215
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.091437037037037e-05,
+      "loss": 1.7366,
+      "step": 32220
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0911407407407414e-05,
+      "loss": 1.7312,
+      "step": 32225
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0908444444444446e-05,
+      "loss": 1.9089,
+      "step": 32230
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0905481481481485e-05,
+      "loss": 1.7415,
+      "step": 32235
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.090251851851852e-05,
+      "loss": 1.9092,
+      "step": 32240
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.089955555555556e-05,
+      "loss": 1.7883,
+      "step": 32245
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0896592592592595e-05,
+      "loss": 1.8109,
+      "step": 32250
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.0893629629629633e-05,
+      "loss": 1.9123,
+      "step": 32255
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.0890666666666665e-05,
+      "loss": 1.8401,
+      "step": 32260
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.0887703703703704e-05,
+      "loss": 1.724,
+      "step": 32265
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.088474074074074e-05,
+      "loss": 1.8262,
+      "step": 32270
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.088177777777778e-05,
+      "loss": 1.7865,
+      "step": 32275
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.0878814814814814e-05,
+      "loss": 1.8945,
+      "step": 32280
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.087585185185185e-05,
+      "loss": 1.8649,
+      "step": 32285
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.087288888888889e-05,
+      "loss": 1.7616,
+      "step": 32290
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.086992592592593e-05,
+      "loss": 1.677,
+      "step": 32295
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.086696296296296e-05,
+      "loss": 1.758,
+      "step": 32300
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.0864e-05,
+      "loss": 1.8861,
+      "step": 32305
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.086103703703704e-05,
+      "loss": 1.7799,
+      "step": 32310
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 3.085807407407408e-05,
+      "loss": 1.7554,
+      "step": 32315
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.085511111111111e-05,
+      "loss": 1.6893,
+      "step": 32320
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.085214814814815e-05,
+      "loss": 1.8322,
+      "step": 32325
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.084918518518519e-05,
+      "loss": 1.7639,
+      "step": 32330
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.084622222222223e-05,
+      "loss": 1.7998,
+      "step": 32335
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.084325925925926e-05,
+      "loss": 1.9589,
+      "step": 32340
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.08402962962963e-05,
+      "loss": 1.6992,
+      "step": 32345
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.083733333333334e-05,
+      "loss": 1.8168,
+      "step": 32350
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.0834370370370376e-05,
+      "loss": 1.8621,
+      "step": 32355
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.083140740740741e-05,
+      "loss": 1.757,
+      "step": 32360
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.082844444444445e-05,
+      "loss": 1.8237,
+      "step": 32365
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.0825481481481486e-05,
+      "loss": 1.7844,
+      "step": 32370
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.082251851851852e-05,
+      "loss": 1.811,
+      "step": 32375
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0819555555555557e-05,
+      "loss": 1.7126,
+      "step": 32380
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.081659259259259e-05,
+      "loss": 1.8338,
+      "step": 32385
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0813629629629634e-05,
+      "loss": 1.6839,
+      "step": 32390
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0810666666666666e-05,
+      "loss": 1.8455,
+      "step": 32395
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0807703703703705e-05,
+      "loss": 1.7468,
+      "step": 32400
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.080474074074074e-05,
+      "loss": 1.6845,
+      "step": 32405
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.080177777777778e-05,
+      "loss": 1.7447,
+      "step": 32410
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0798814814814815e-05,
+      "loss": 1.7269,
+      "step": 32415
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0795851851851854e-05,
+      "loss": 1.8179,
+      "step": 32420
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 3.0792888888888886e-05,
+      "loss": 1.7824,
+      "step": 32425
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.0789925925925924e-05,
+      "loss": 1.8312,
+      "step": 32430
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.078696296296296e-05,
+      "loss": 1.7442,
+      "step": 32435
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.0784e-05,
+      "loss": 1.8164,
+      "step": 32440
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.0781037037037034e-05,
+      "loss": 1.603,
+      "step": 32445
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.077807407407407e-05,
+      "loss": 1.792,
+      "step": 32450
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.077511111111111e-05,
+      "loss": 1.6909,
+      "step": 32455
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.077214814814815e-05,
+      "loss": 1.8125,
+      "step": 32460
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.076918518518518e-05,
+      "loss": 1.8062,
+      "step": 32465
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.076622222222222e-05,
+      "loss": 1.8516,
+      "step": 32470
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.076325925925926e-05,
+      "loss": 1.8091,
+      "step": 32475
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.07602962962963e-05,
+      "loss": 1.8028,
+      "step": 32480
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.075733333333333e-05,
+      "loss": 1.7271,
+      "step": 32485
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.075437037037037e-05,
+      "loss": 1.7458,
+      "step": 32490
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.075140740740741e-05,
+      "loss": 1.7706,
+      "step": 32495
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.074844444444445e-05,
+      "loss": 1.6899,
+      "step": 32500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.074548148148148e-05,
+      "loss": 1.7182,
+      "step": 32505
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.074251851851852e-05,
+      "loss": 1.6718,
+      "step": 32510
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.073955555555556e-05,
+      "loss": 1.77,
+      "step": 32515
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.0736592592592596e-05,
+      "loss": 1.7513,
+      "step": 32520
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.073362962962963e-05,
+      "loss": 1.7807,
+      "step": 32525
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.073066666666667e-05,
+      "loss": 1.9059,
+      "step": 32530
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.0727703703703706e-05,
+      "loss": 1.8233,
+      "step": 32535
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 3.0724740740740745e-05,
+      "loss": 1.8722,
+      "step": 32540
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.072177777777778e-05,
+      "loss": 1.6277,
+      "step": 32545
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0718814814814815e-05,
+      "loss": 1.7178,
+      "step": 32550
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0715851851851854e-05,
+      "loss": 1.8676,
+      "step": 32555
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.071288888888889e-05,
+      "loss": 1.811,
+      "step": 32560
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0709925925925925e-05,
+      "loss": 1.6865,
+      "step": 32565
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0706962962962964e-05,
+      "loss": 1.7294,
+      "step": 32570
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0703999999999996e-05,
+      "loss": 1.8459,
+      "step": 32575
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.070103703703704e-05,
+      "loss": 1.7731,
+      "step": 32580
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0698074074074074e-05,
+      "loss": 1.8431,
+      "step": 32585
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.069511111111111e-05,
+      "loss": 1.7727,
+      "step": 32590
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.0692148148148145e-05,
+      "loss": 1.8884,
+      "step": 32595
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.068918518518519e-05,
+      "loss": 1.8343,
+      "step": 32600
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.068622222222222e-05,
+      "loss": 1.7352,
+      "step": 32605
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.068325925925926e-05,
+      "loss": 1.6606,
+      "step": 32610
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.068029629629629e-05,
+      "loss": 1.7679,
+      "step": 32615
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.067733333333334e-05,
+      "loss": 1.8008,
+      "step": 32620
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.067437037037037e-05,
+      "loss": 1.8578,
+      "step": 32625
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.067140740740741e-05,
+      "loss": 1.6442,
+      "step": 32630
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.066844444444444e-05,
+      "loss": 1.8161,
+      "step": 32635
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.066548148148149e-05,
+      "loss": 1.859,
+      "step": 32640
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.066251851851852e-05,
+      "loss": 1.7786,
+      "step": 32645
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 3.065955555555556e-05,
+      "loss": 1.7326,
+      "step": 32650
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.065659259259259e-05,
+      "loss": 1.8917,
+      "step": 32655
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.065362962962963e-05,
+      "loss": 1.6301,
+      "step": 32660
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.065066666666667e-05,
+      "loss": 1.7904,
+      "step": 32665
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0647703703703707e-05,
+      "loss": 1.7485,
+      "step": 32670
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.064474074074074e-05,
+      "loss": 1.6792,
+      "step": 32675
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.064177777777778e-05,
+      "loss": 1.8091,
+      "step": 32680
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0638814814814816e-05,
+      "loss": 1.843,
+      "step": 32685
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0635851851851855e-05,
+      "loss": 1.7934,
+      "step": 32690
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.063288888888889e-05,
+      "loss": 1.894,
+      "step": 32695
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0629925925925926e-05,
+      "loss": 1.6811,
+      "step": 32700
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.0626962962962965e-05,
+      "loss": 1.7364,
+      "step": 32705
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0624000000000004e-05,
+      "loss": 1.7358,
+      "step": 32710
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0621037037037036e-05,
+      "loss": 1.8058,
+      "step": 32715
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0618074074074074e-05,
+      "loss": 1.782,
+      "step": 32720
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.061511111111111e-05,
+      "loss": 1.8822,
+      "step": 32725
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.061214814814815e-05,
+      "loss": 1.8611,
+      "step": 32730
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.0609185185185184e-05,
+      "loss": 1.7027,
+      "step": 32735
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.060622222222222e-05,
+      "loss": 1.784,
+      "step": 32740
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.060325925925926e-05,
+      "loss": 1.8381,
+      "step": 32745
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.06002962962963e-05,
+      "loss": 1.9241,
+      "step": 32750
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.059733333333333e-05,
+      "loss": 1.8515,
+      "step": 32755
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.059437037037037e-05,
+      "loss": 1.9129,
+      "step": 32760
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 3.059140740740741e-05,
+      "loss": 1.764,
+      "step": 32765
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.058844444444445e-05,
+      "loss": 1.7665,
+      "step": 32770
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.058548148148148e-05,
+      "loss": 1.8665,
+      "step": 32775
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.058251851851852e-05,
+      "loss": 1.7605,
+      "step": 32780
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.057955555555556e-05,
+      "loss": 1.8172,
+      "step": 32785
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.05765925925926e-05,
+      "loss": 1.881,
+      "step": 32790
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.057362962962963e-05,
+      "loss": 1.6562,
+      "step": 32795
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.057066666666667e-05,
+      "loss": 1.835,
+      "step": 32800
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.056770370370371e-05,
+      "loss": 1.8155,
+      "step": 32805
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.0564740740740746e-05,
+      "loss": 1.7459,
+      "step": 32810
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.056177777777778e-05,
+      "loss": 1.765,
+      "step": 32815
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.055881481481482e-05,
+      "loss": 1.8043,
+      "step": 32820
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.055585185185185e-05,
+      "loss": 1.6945,
+      "step": 32825
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0552888888888895e-05,
+      "loss": 1.7981,
+      "step": 32830
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.054992592592593e-05,
+      "loss": 1.7391,
+      "step": 32835
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0546962962962966e-05,
+      "loss": 1.7702,
+      "step": 32840
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0544e-05,
+      "loss": 1.6233,
+      "step": 32845
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.054103703703704e-05,
+      "loss": 1.7269,
+      "step": 32850
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0538074074074075e-05,
+      "loss": 1.7322,
+      "step": 32855
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0535111111111114e-05,
+      "loss": 1.7417,
+      "step": 32860
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0532148148148146e-05,
+      "loss": 1.856,
+      "step": 32865
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.052918518518519e-05,
+      "loss": 1.715,
+      "step": 32870
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.0526222222222224e-05,
+      "loss": 1.6516,
+      "step": 32875
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.052325925925926e-05,
+      "loss": 1.7325,
+      "step": 32880
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0520296296296295e-05,
+      "loss": 1.8397,
+      "step": 32885
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0517333333333337e-05,
+      "loss": 1.8394,
+      "step": 32890
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0514370370370372e-05,
+      "loss": 1.8224,
+      "step": 32895
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.051140740740741e-05,
+      "loss": 1.8271,
+      "step": 32900
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0508444444444447e-05,
+      "loss": 1.8381,
+      "step": 32905
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0505481481481485e-05,
+      "loss": 1.6783,
+      "step": 32910
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.050251851851852e-05,
+      "loss": 1.8412,
+      "step": 32915
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.049955555555556e-05,
+      "loss": 1.805,
+      "step": 32920
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.049659259259259e-05,
+      "loss": 1.7668,
+      "step": 32925
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 3.0493629629629634e-05,
+      "loss": 1.8387,
+      "step": 32930
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0490666666666666e-05,
+      "loss": 1.8191,
+      "step": 32935
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0487703703703708e-05,
+      "loss": 1.7234,
+      "step": 32940
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.048474074074074e-05,
+      "loss": 1.8168,
+      "step": 32945
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0481777777777782e-05,
+      "loss": 1.7194,
+      "step": 32950
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0478814814814814e-05,
+      "loss": 1.7382,
+      "step": 32955
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0475851851851857e-05,
+      "loss": 1.9391,
+      "step": 32960
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.047288888888889e-05,
+      "loss": 1.907,
+      "step": 32965
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.046992592592593e-05,
+      "loss": 1.7447,
+      "step": 32970
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0466962962962963e-05,
+      "loss": 1.7888,
+      "step": 32975
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0464000000000005e-05,
+      "loss": 1.7229,
+      "step": 32980
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.0461037037037037e-05,
+      "loss": 1.7552,
+      "step": 32985
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.045807407407408e-05,
+      "loss": 1.706,
+      "step": 32990
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.045511111111111e-05,
+      "loss": 1.9783,
+      "step": 32995
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.045214814814815e-05,
+      "loss": 1.7567,
+      "step": 33000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0449185185185186e-05,
+      "loss": 1.8879,
+      "step": 33005
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0446222222222225e-05,
+      "loss": 1.8954,
+      "step": 33010
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.044325925925926e-05,
+      "loss": 1.7172,
+      "step": 33015
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.04402962962963e-05,
+      "loss": 1.807,
+      "step": 33020
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0437333333333334e-05,
+      "loss": 1.8217,
+      "step": 33025
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0434370370370373e-05,
+      "loss": 1.8028,
+      "step": 33030
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.043140740740741e-05,
+      "loss": 1.823,
+      "step": 33035
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0428444444444447e-05,
+      "loss": 1.8359,
+      "step": 33040
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 3.0425481481481483e-05,
+      "loss": 1.8417,
+      "step": 33045
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.042251851851852e-05,
+      "loss": 1.6888,
+      "step": 33050
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0419555555555557e-05,
+      "loss": 1.8624,
+      "step": 33055
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0416592592592596e-05,
+      "loss": 1.724,
+      "step": 33060
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.041362962962963e-05,
+      "loss": 1.7003,
+      "step": 33065
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.041066666666667e-05,
+      "loss": 1.8295,
+      "step": 33070
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0407703703703705e-05,
+      "loss": 1.8725,
+      "step": 33075
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0404740740740744e-05,
+      "loss": 1.7355,
+      "step": 33080
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.040177777777778e-05,
+      "loss": 1.8111,
+      "step": 33085
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.039881481481482e-05,
+      "loss": 1.7846,
+      "step": 33090
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0395851851851854e-05,
+      "loss": 1.7004,
+      "step": 33095
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.0392888888888893e-05,
+      "loss": 1.8205,
+      "step": 33100
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0389925925925928e-05,
+      "loss": 1.8023,
+      "step": 33105
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0386962962962967e-05,
+      "loss": 1.7341,
+      "step": 33110
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0384000000000003e-05,
+      "loss": 1.8227,
+      "step": 33115
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.038103703703704e-05,
+      "loss": 1.6902,
+      "step": 33120
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0378074074074077e-05,
+      "loss": 1.7117,
+      "step": 33125
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0375111111111116e-05,
+      "loss": 1.869,
+      "step": 33130
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.037214814814815e-05,
+      "loss": 1.7632,
+      "step": 33135
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.036918518518519e-05,
+      "loss": 1.802,
+      "step": 33140
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0366222222222225e-05,
+      "loss": 1.8446,
+      "step": 33145
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0363259259259257e-05,
+      "loss": 1.7217,
+      "step": 33150
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.0360296296296296e-05,
+      "loss": 1.8901,
+      "step": 33155
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.035733333333333e-05,
+      "loss": 1.7266,
+      "step": 33160
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.035437037037037e-05,
+      "loss": 1.7804,
+      "step": 33165
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0351407407407406e-05,
+      "loss": 1.9004,
+      "step": 33170
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0348444444444445e-05,
+      "loss": 1.7654,
+      "step": 33175
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.034548148148148e-05,
+      "loss": 1.754,
+      "step": 33180
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.034251851851852e-05,
+      "loss": 1.7461,
+      "step": 33185
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0339555555555554e-05,
+      "loss": 1.7956,
+      "step": 33190
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0336592592592593e-05,
+      "loss": 1.8341,
+      "step": 33195
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.033362962962963e-05,
+      "loss": 1.8076,
+      "step": 33200
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0330666666666667e-05,
+      "loss": 1.7301,
+      "step": 33205
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.0327703703703703e-05,
+      "loss": 1.8836,
+      "step": 33210
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.032474074074074e-05,
+      "loss": 1.8296,
+      "step": 33215
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0321777777777777e-05,
+      "loss": 1.7662,
+      "step": 33220
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0318814814814816e-05,
+      "loss": 1.8175,
+      "step": 33225
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.031585185185185e-05,
+      "loss": 1.8687,
+      "step": 33230
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.031288888888889e-05,
+      "loss": 1.8071,
+      "step": 33235
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0309925925925926e-05,
+      "loss": 1.8342,
+      "step": 33240
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0306962962962964e-05,
+      "loss": 1.7908,
+      "step": 33245
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0304e-05,
+      "loss": 1.7336,
+      "step": 33250
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.030103703703704e-05,
+      "loss": 1.8212,
+      "step": 33255
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0298074074074074e-05,
+      "loss": 1.7287,
+      "step": 33260
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.0295111111111113e-05,
+      "loss": 1.8038,
+      "step": 33265
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.029214814814815e-05,
+      "loss": 1.8705,
+      "step": 33270
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0289185185185187e-05,
+      "loss": 1.7662,
+      "step": 33275
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0286222222222223e-05,
+      "loss": 1.8536,
+      "step": 33280
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.028325925925926e-05,
+      "loss": 1.7853,
+      "step": 33285
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0280296296296297e-05,
+      "loss": 1.8236,
+      "step": 33290
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0277333333333336e-05,
+      "loss": 1.8192,
+      "step": 33295
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.027437037037037e-05,
+      "loss": 1.6988,
+      "step": 33300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.027140740740741e-05,
+      "loss": 1.7753,
+      "step": 33305
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0268444444444442e-05,
+      "loss": 1.7168,
+      "step": 33310
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0265481481481484e-05,
+      "loss": 1.9222,
+      "step": 33315
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.0262518518518516e-05,
+      "loss": 1.7062,
+      "step": 33320
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.025955555555556e-05,
+      "loss": 1.8108,
+      "step": 33325
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.025659259259259e-05,
+      "loss": 1.8518,
+      "step": 33330
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0253629629629633e-05,
+      "loss": 1.6796,
+      "step": 33335
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0250666666666665e-05,
+      "loss": 1.8585,
+      "step": 33340
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0247703703703707e-05,
+      "loss": 1.8071,
+      "step": 33345
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.024474074074074e-05,
+      "loss": 1.8277,
+      "step": 33350
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.024177777777778e-05,
+      "loss": 1.7331,
+      "step": 33355
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0238814814814813e-05,
+      "loss": 1.836,
+      "step": 33360
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0235851851851856e-05,
+      "loss": 1.7649,
+      "step": 33365
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0232888888888888e-05,
+      "loss": 1.859,
+      "step": 33370
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.022992592592593e-05,
+      "loss": 1.6977,
+      "step": 33375
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.0226962962962962e-05,
+      "loss": 1.7289,
+      "step": 33380
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0224e-05,
+      "loss": 1.7872,
+      "step": 33385
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0221037037037036e-05,
+      "loss": 1.807,
+      "step": 33390
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0218074074074075e-05,
+      "loss": 1.7508,
+      "step": 33395
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.021511111111111e-05,
+      "loss": 1.7862,
+      "step": 33400
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.021214814814815e-05,
+      "loss": 1.7932,
+      "step": 33405
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0209185185185185e-05,
+      "loss": 1.8647,
+      "step": 33410
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0206222222222223e-05,
+      "loss": 1.9436,
+      "step": 33415
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.020325925925926e-05,
+      "loss": 1.8402,
+      "step": 33420
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0200296296296298e-05,
+      "loss": 1.723,
+      "step": 33425
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0197333333333333e-05,
+      "loss": 1.7367,
+      "step": 33430
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0194370370370372e-05,
+      "loss": 1.8149,
+      "step": 33435
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 3.0191407407407407e-05,
+      "loss": 1.9072,
+      "step": 33440
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0188444444444446e-05,
+      "loss": 1.795,
+      "step": 33445
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.018548148148148e-05,
+      "loss": 1.8279,
+      "step": 33450
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.018251851851852e-05,
+      "loss": 1.7217,
+      "step": 33455
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0179555555555556e-05,
+      "loss": 1.788,
+      "step": 33460
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0176592592592595e-05,
+      "loss": 1.7992,
+      "step": 33465
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.017362962962963e-05,
+      "loss": 1.8002,
+      "step": 33470
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.017066666666667e-05,
+      "loss": 1.7818,
+      "step": 33475
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0167703703703704e-05,
+      "loss": 1.7239,
+      "step": 33480
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0164740740740743e-05,
+      "loss": 1.8778,
+      "step": 33485
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.016177777777778e-05,
+      "loss": 1.8305,
+      "step": 33490
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.0158814814814817e-05,
+      "loss": 1.7127,
+      "step": 33495
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0155851851851853e-05,
+      "loss": 1.8816,
+      "step": 33500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0152888888888892e-05,
+      "loss": 1.7817,
+      "step": 33505
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0149925925925927e-05,
+      "loss": 1.8087,
+      "step": 33510
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0146962962962966e-05,
+      "loss": 1.8684,
+      "step": 33515
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0144e-05,
+      "loss": 1.711,
+      "step": 33520
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.014103703703704e-05,
+      "loss": 1.9192,
+      "step": 33525
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0138074074074076e-05,
+      "loss": 1.7684,
+      "step": 33530
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.0135111111111114e-05,
+      "loss": 1.806,
+      "step": 33535
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.013214814814815e-05,
+      "loss": 1.7198,
+      "step": 33540
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.012918518518519e-05,
+      "loss": 1.8798,
+      "step": 33545
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 3.012622222222222e-05,
+      "loss": 1.7622,
+      "step": 33550
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0123259259259263e-05,
+      "loss": 1.9225,
+      "step": 33555
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0120296296296295e-05,
+      "loss": 1.7447,
+      "step": 33560
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0117333333333337e-05,
+      "loss": 1.7922,
+      "step": 33565
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.011437037037037e-05,
+      "loss": 1.7375,
+      "step": 33570
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.011140740740741e-05,
+      "loss": 1.84,
+      "step": 33575
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0108444444444444e-05,
+      "loss": 1.9391,
+      "step": 33580
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0105481481481486e-05,
+      "loss": 1.7395,
+      "step": 33585
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0102518518518518e-05,
+      "loss": 1.8104,
+      "step": 33590
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.009955555555556e-05,
+      "loss": 1.805,
+      "step": 33595
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0096592592592592e-05,
+      "loss": 1.8873,
+      "step": 33600
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.0093629629629634e-05,
+      "loss": 1.6849,
+      "step": 33605
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0090666666666666e-05,
+      "loss": 1.7263,
+      "step": 33610
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0087703703703705e-05,
+      "loss": 1.676,
+      "step": 33615
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.008474074074074e-05,
+      "loss": 1.8337,
+      "step": 33620
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.008177777777778e-05,
+      "loss": 1.7559,
+      "step": 33625
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0078814814814815e-05,
+      "loss": 1.8618,
+      "step": 33630
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0075851851851854e-05,
+      "loss": 1.8777,
+      "step": 33635
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.007288888888889e-05,
+      "loss": 1.7222,
+      "step": 33640
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0069925925925928e-05,
+      "loss": 1.7913,
+      "step": 33645
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0066962962962963e-05,
+      "loss": 1.8384,
+      "step": 33650
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0064000000000002e-05,
+      "loss": 1.8482,
+      "step": 33655
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0061037037037038e-05,
+      "loss": 1.8629,
+      "step": 33660
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 3.0058074074074076e-05,
+      "loss": 1.8857,
+      "step": 33665
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0055111111111112e-05,
+      "loss": 1.8183,
+      "step": 33670
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.005214814814815e-05,
+      "loss": 1.8407,
+      "step": 33675
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0049185185185186e-05,
+      "loss": 1.7283,
+      "step": 33680
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0046222222222225e-05,
+      "loss": 1.6895,
+      "step": 33685
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.004325925925926e-05,
+      "loss": 1.7861,
+      "step": 33690
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.00402962962963e-05,
+      "loss": 1.8633,
+      "step": 33695
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0037333333333335e-05,
+      "loss": 1.8043,
+      "step": 33700
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0034370370370373e-05,
+      "loss": 1.7407,
+      "step": 33705
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.003140740740741e-05,
+      "loss": 1.87,
+      "step": 33710
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0028444444444448e-05,
+      "loss": 1.8539,
+      "step": 33715
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.0025481481481483e-05,
+      "loss": 1.6793,
+      "step": 33720
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0022518518518522e-05,
+      "loss": 1.7757,
+      "step": 33725
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0019555555555557e-05,
+      "loss": 1.7561,
+      "step": 33730
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0016592592592596e-05,
+      "loss": 1.7549,
+      "step": 33735
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.001362962962963e-05,
+      "loss": 1.843,
+      "step": 33740
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.001066666666667e-05,
+      "loss": 1.8345,
+      "step": 33745
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0007703703703706e-05,
+      "loss": 1.7254,
+      "step": 33750
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.0004740740740745e-05,
+      "loss": 1.618,
+      "step": 33755
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.000177777777778e-05,
+      "loss": 1.6368,
+      "step": 33760
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.999881481481482e-05,
+      "loss": 1.6529,
+      "step": 33765
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.9995851851851854e-05,
+      "loss": 1.707,
+      "step": 33770
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.9992888888888893e-05,
+      "loss": 1.6199,
+      "step": 33775
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9989925925925925e-05,
+      "loss": 1.5544,
+      "step": 33780
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9986962962962968e-05,
+      "loss": 1.7129,
+      "step": 33785
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9984e-05,
+      "loss": 1.6938,
+      "step": 33790
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9981037037037042e-05,
+      "loss": 1.6195,
+      "step": 33795
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9978074074074074e-05,
+      "loss": 1.6191,
+      "step": 33800
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9975111111111116e-05,
+      "loss": 1.6264,
+      "step": 33805
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9972148148148148e-05,
+      "loss": 1.6907,
+      "step": 33810
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.996918518518519e-05,
+      "loss": 1.6615,
+      "step": 33815
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9966222222222222e-05,
+      "loss": 1.6945,
+      "step": 33820
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9963259259259265e-05,
+      "loss": 1.7901,
+      "step": 33825
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.9960296296296297e-05,
+      "loss": 1.7297,
+      "step": 33830
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.995733333333334e-05,
+      "loss": 1.7551,
+      "step": 33835
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.995437037037037e-05,
+      "loss": 1.6866,
+      "step": 33840
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9951407407407413e-05,
+      "loss": 1.7442,
+      "step": 33845
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9948444444444445e-05,
+      "loss": 1.7564,
+      "step": 33850
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9945481481481484e-05,
+      "loss": 1.7313,
+      "step": 33855
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.994251851851852e-05,
+      "loss": 1.7315,
+      "step": 33860
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9939555555555558e-05,
+      "loss": 1.6853,
+      "step": 33865
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9936592592592594e-05,
+      "loss": 1.7534,
+      "step": 33870
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9933629629629632e-05,
+      "loss": 1.7043,
+      "step": 33875
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9930666666666668e-05,
+      "loss": 1.6982,
+      "step": 33880
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9927703703703707e-05,
+      "loss": 1.708,
+      "step": 33885
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.9924740740740742e-05,
+      "loss": 1.7616,
+      "step": 33890
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.992177777777778e-05,
+      "loss": 1.6442,
+      "step": 33895
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9918814814814816e-05,
+      "loss": 1.6282,
+      "step": 33900
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9915851851851855e-05,
+      "loss": 1.6471,
+      "step": 33905
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.991288888888889e-05,
+      "loss": 1.5863,
+      "step": 33910
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.990992592592593e-05,
+      "loss": 1.7191,
+      "step": 33915
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9906962962962965e-05,
+      "loss": 1.6954,
+      "step": 33920
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9904e-05,
+      "loss": 1.7576,
+      "step": 33925
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.990103703703704e-05,
+      "loss": 1.6584,
+      "step": 33930
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.989807407407407e-05,
+      "loss": 1.6205,
+      "step": 33935
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9895111111111113e-05,
+      "loss": 1.6828,
+      "step": 33940
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.9892148148148145e-05,
+      "loss": 1.6621,
+      "step": 33945
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9889185185185188e-05,
+      "loss": 1.5537,
+      "step": 33950
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.988622222222222e-05,
+      "loss": 1.6692,
+      "step": 33955
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9883259259259262e-05,
+      "loss": 1.6727,
+      "step": 33960
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9880296296296294e-05,
+      "loss": 1.6318,
+      "step": 33965
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9877333333333336e-05,
+      "loss": 1.6832,
+      "step": 33970
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9874370370370368e-05,
+      "loss": 1.7019,
+      "step": 33975
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.987140740740741e-05,
+      "loss": 1.8028,
+      "step": 33980
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9868444444444442e-05,
+      "loss": 1.6411,
+      "step": 33985
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9865481481481485e-05,
+      "loss": 1.6213,
+      "step": 33990
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.9862518518518517e-05,
+      "loss": 1.578,
+      "step": 33995
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.985955555555556e-05,
+      "loss": 1.754,
+      "step": 34000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.985659259259259e-05,
+      "loss": 1.7044,
+      "step": 34005
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.985362962962963e-05,
+      "loss": 1.6541,
+      "step": 34010
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9850666666666665e-05,
+      "loss": 1.7141,
+      "step": 34015
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9847703703703704e-05,
+      "loss": 1.7351,
+      "step": 34020
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.984474074074074e-05,
+      "loss": 1.7489,
+      "step": 34025
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.984177777777778e-05,
+      "loss": 1.4647,
+      "step": 34030
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9838814814814814e-05,
+      "loss": 1.6804,
+      "step": 34035
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9835851851851853e-05,
+      "loss": 1.7143,
+      "step": 34040
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9832888888888888e-05,
+      "loss": 1.6343,
+      "step": 34045
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9829925925925927e-05,
+      "loss": 1.7678,
+      "step": 34050
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.9826962962962962e-05,
+      "loss": 1.762,
+      "step": 34055
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9824e-05,
+      "loss": 1.8423,
+      "step": 34060
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9821037037037037e-05,
+      "loss": 1.5978,
+      "step": 34065
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9818074074074075e-05,
+      "loss": 1.7009,
+      "step": 34070
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.981511111111111e-05,
+      "loss": 1.6667,
+      "step": 34075
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.981214814814815e-05,
+      "loss": 1.7031,
+      "step": 34080
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9809185185185185e-05,
+      "loss": 1.828,
+      "step": 34085
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9806222222222224e-05,
+      "loss": 1.6861,
+      "step": 34090
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.980325925925926e-05,
+      "loss": 1.6868,
+      "step": 34095
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9800296296296298e-05,
+      "loss": 1.7758,
+      "step": 34100
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9797333333333334e-05,
+      "loss": 1.6938,
+      "step": 34105
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9794370370370372e-05,
+      "loss": 1.5606,
+      "step": 34110
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.9791407407407408e-05,
+      "loss": 1.6517,
+      "step": 34115
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9788444444444447e-05,
+      "loss": 1.698,
+      "step": 34120
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9785481481481482e-05,
+      "loss": 1.6033,
+      "step": 34125
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.978251851851852e-05,
+      "loss": 1.6159,
+      "step": 34130
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9779555555555556e-05,
+      "loss": 1.8065,
+      "step": 34135
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9776592592592595e-05,
+      "loss": 1.7817,
+      "step": 34140
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.977362962962963e-05,
+      "loss": 1.6644,
+      "step": 34145
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.977066666666667e-05,
+      "loss": 1.652,
+      "step": 34150
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9767703703703705e-05,
+      "loss": 1.7358,
+      "step": 34155
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9764740740740744e-05,
+      "loss": 1.6544,
+      "step": 34160
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9761777777777776e-05,
+      "loss": 1.6763,
+      "step": 34165
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.9758814814814818e-05,
+      "loss": 1.6336,
+      "step": 34170
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.975585185185185e-05,
+      "loss": 1.665,
+      "step": 34175
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9752888888888892e-05,
+      "loss": 1.6985,
+      "step": 34180
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9749925925925924e-05,
+      "loss": 1.7731,
+      "step": 34185
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9746962962962966e-05,
+      "loss": 1.7405,
+      "step": 34190
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9744e-05,
+      "loss": 1.6614,
+      "step": 34195
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.974103703703704e-05,
+      "loss": 1.7459,
+      "step": 34200
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9738074074074073e-05,
+      "loss": 1.6534,
+      "step": 34205
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9735111111111115e-05,
+      "loss": 1.8145,
+      "step": 34210
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.9732148148148147e-05,
+      "loss": 1.6832,
+      "step": 34215
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.972918518518519e-05,
+      "loss": 1.6496,
+      "step": 34220
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.972622222222222e-05,
+      "loss": 1.7708,
+      "step": 34225
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9723259259259263e-05,
+      "loss": 1.7077,
+      "step": 34230
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9720296296296295e-05,
+      "loss": 1.6498,
+      "step": 34235
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9717333333333334e-05,
+      "loss": 1.7104,
+      "step": 34240
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.971437037037037e-05,
+      "loss": 1.7321,
+      "step": 34245
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.971140740740741e-05,
+      "loss": 1.6193,
+      "step": 34250
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9708444444444444e-05,
+      "loss": 1.6764,
+      "step": 34255
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9705481481481483e-05,
+      "loss": 1.7075,
+      "step": 34260
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9702518518518518e-05,
+      "loss": 1.6696,
+      "step": 34265
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9699555555555557e-05,
+      "loss": 1.6629,
+      "step": 34270
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.9696592592592593e-05,
+      "loss": 1.6489,
+      "step": 34275
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.969362962962963e-05,
+      "loss": 1.6474,
+      "step": 34280
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9690666666666667e-05,
+      "loss": 1.7237,
+      "step": 34285
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9687703703703706e-05,
+      "loss": 1.6771,
+      "step": 34290
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.968474074074074e-05,
+      "loss": 1.761,
+      "step": 34295
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.968177777777778e-05,
+      "loss": 1.6355,
+      "step": 34300
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9678814814814815e-05,
+      "loss": 1.6497,
+      "step": 34305
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9675851851851854e-05,
+      "loss": 1.599,
+      "step": 34310
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.967288888888889e-05,
+      "loss": 1.7796,
+      "step": 34315
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.966992592592593e-05,
+      "loss": 1.6912,
+      "step": 34320
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9666962962962964e-05,
+      "loss": 1.8121,
+      "step": 34325
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9664000000000003e-05,
+      "loss": 1.827,
+      "step": 34330
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9661037037037038e-05,
+      "loss": 1.7152,
+      "step": 34335
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.9658074074074077e-05,
+      "loss": 1.7404,
+      "step": 34340
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9655111111111112e-05,
+      "loss": 1.7682,
+      "step": 34345
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.965214814814815e-05,
+      "loss": 1.724,
+      "step": 34350
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9649185185185187e-05,
+      "loss": 1.5152,
+      "step": 34355
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9646222222222225e-05,
+      "loss": 1.6644,
+      "step": 34360
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.964325925925926e-05,
+      "loss": 1.7016,
+      "step": 34365
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.96402962962963e-05,
+      "loss": 1.7095,
+      "step": 34370
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9637333333333335e-05,
+      "loss": 1.7522,
+      "step": 34375
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9634370370370374e-05,
+      "loss": 1.7325,
+      "step": 34380
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.963140740740741e-05,
+      "loss": 1.8635,
+      "step": 34385
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.9628444444444448e-05,
+      "loss": 1.6929,
+      "step": 34390
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.962548148148148e-05,
+      "loss": 1.6304,
+      "step": 34395
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9622518518518522e-05,
+      "loss": 1.7542,
+      "step": 34400
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9619555555555554e-05,
+      "loss": 1.4836,
+      "step": 34405
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9616592592592597e-05,
+      "loss": 1.5855,
+      "step": 34410
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.961362962962963e-05,
+      "loss": 1.7014,
+      "step": 34415
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.961066666666667e-05,
+      "loss": 1.5937,
+      "step": 34420
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9607703703703703e-05,
+      "loss": 1.7605,
+      "step": 34425
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9604740740740745e-05,
+      "loss": 1.6697,
+      "step": 34430
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9601777777777777e-05,
+      "loss": 1.7434,
+      "step": 34435
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.959881481481482e-05,
+      "loss": 1.575,
+      "step": 34440
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.959585185185185e-05,
+      "loss": 1.7406,
+      "step": 34445
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.9592888888888894e-05,
+      "loss": 1.6422,
+      "step": 34450
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9589925925925926e-05,
+      "loss": 1.6747,
+      "step": 34455
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9586962962962968e-05,
+      "loss": 1.7242,
+      "step": 34460
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9584e-05,
+      "loss": 1.7692,
+      "step": 34465
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.958103703703704e-05,
+      "loss": 1.7235,
+      "step": 34470
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9578074074074074e-05,
+      "loss": 1.7879,
+      "step": 34475
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9575111111111113e-05,
+      "loss": 1.6452,
+      "step": 34480
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.957214814814815e-05,
+      "loss": 1.5948,
+      "step": 34485
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9569185185185187e-05,
+      "loss": 1.7005,
+      "step": 34490
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9566222222222223e-05,
+      "loss": 1.6388,
+      "step": 34495
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.956325925925926e-05,
+      "loss": 1.7898,
+      "step": 34500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.9560296296296297e-05,
+      "loss": 1.6613,
+      "step": 34505
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9557333333333336e-05,
+      "loss": 1.5934,
+      "step": 34510
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.955437037037037e-05,
+      "loss": 1.6482,
+      "step": 34515
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.955140740740741e-05,
+      "loss": 1.6685,
+      "step": 34520
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9548444444444446e-05,
+      "loss": 1.7316,
+      "step": 34525
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9545481481481484e-05,
+      "loss": 1.7118,
+      "step": 34530
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.954251851851852e-05,
+      "loss": 1.7731,
+      "step": 34535
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.953955555555556e-05,
+      "loss": 1.6958,
+      "step": 34540
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9536592592592594e-05,
+      "loss": 1.514,
+      "step": 34545
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9533629629629633e-05,
+      "loss": 1.6441,
+      "step": 34550
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.953066666666667e-05,
+      "loss": 1.7139,
+      "step": 34555
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9527703703703707e-05,
+      "loss": 1.6581,
+      "step": 34560
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.9524740740740743e-05,
+      "loss": 1.8076,
+      "step": 34565
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.952177777777778e-05,
+      "loss": 1.7009,
+      "step": 34570
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9518814814814817e-05,
+      "loss": 1.7005,
+      "step": 34575
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9515851851851856e-05,
+      "loss": 1.7571,
+      "step": 34580
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.951288888888889e-05,
+      "loss": 1.7,
+      "step": 34585
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.950992592592593e-05,
+      "loss": 1.7021,
+      "step": 34590
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9506962962962965e-05,
+      "loss": 1.8133,
+      "step": 34595
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9504000000000004e-05,
+      "loss": 1.7433,
+      "step": 34600
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.950103703703704e-05,
+      "loss": 1.6422,
+      "step": 34605
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.949807407407408e-05,
+      "loss": 1.6432,
+      "step": 34610
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9495111111111114e-05,
+      "loss": 1.7034,
+      "step": 34615
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.9492148148148153e-05,
+      "loss": 1.805,
+      "step": 34620
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9489185185185185e-05,
+      "loss": 1.7426,
+      "step": 34625
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9486222222222227e-05,
+      "loss": 1.6444,
+      "step": 34630
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.948325925925926e-05,
+      "loss": 1.6825,
+      "step": 34635
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.94802962962963e-05,
+      "loss": 1.6265,
+      "step": 34640
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9477333333333333e-05,
+      "loss": 1.743,
+      "step": 34645
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9474370370370375e-05,
+      "loss": 1.7748,
+      "step": 34650
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9471407407407407e-05,
+      "loss": 1.7518,
+      "step": 34655
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.946844444444445e-05,
+      "loss": 1.7464,
+      "step": 34660
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9465481481481482e-05,
+      "loss": 1.8027,
+      "step": 34665
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9462518518518524e-05,
+      "loss": 1.6877,
+      "step": 34670
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.9459555555555556e-05,
+      "loss": 1.6869,
+      "step": 34675
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9456592592592598e-05,
+      "loss": 1.6942,
+      "step": 34680
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.945362962962963e-05,
+      "loss": 1.6864,
+      "step": 34685
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9450666666666672e-05,
+      "loss": 1.7328,
+      "step": 34690
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9447703703703705e-05,
+      "loss": 1.5921,
+      "step": 34695
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.944474074074074e-05,
+      "loss": 1.6978,
+      "step": 34700
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.944177777777778e-05,
+      "loss": 1.7213,
+      "step": 34705
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9438814814814814e-05,
+      "loss": 1.7641,
+      "step": 34710
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9435851851851853e-05,
+      "loss": 1.6318,
+      "step": 34715
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.943288888888889e-05,
+      "loss": 1.6806,
+      "step": 34720
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9429925925925927e-05,
+      "loss": 1.6974,
+      "step": 34725
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.9426962962962963e-05,
+      "loss": 1.5681,
+      "step": 34730
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9424e-05,
+      "loss": 1.7455,
+      "step": 34735
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9421037037037037e-05,
+      "loss": 1.8332,
+      "step": 34740
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9418074074074076e-05,
+      "loss": 1.6696,
+      "step": 34745
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.941511111111111e-05,
+      "loss": 1.8294,
+      "step": 34750
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.941214814814815e-05,
+      "loss": 1.7163,
+      "step": 34755
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9409185185185185e-05,
+      "loss": 1.7248,
+      "step": 34760
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9406222222222224e-05,
+      "loss": 1.5798,
+      "step": 34765
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.940325925925926e-05,
+      "loss": 1.6716,
+      "step": 34770
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.94002962962963e-05,
+      "loss": 1.7085,
+      "step": 34775
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.939733333333333e-05,
+      "loss": 1.5901,
+      "step": 34780
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9394370370370373e-05,
+      "loss": 1.6369,
+      "step": 34785
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.9391407407407405e-05,
+      "loss": 1.6093,
+      "step": 34790
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9388444444444447e-05,
+      "loss": 1.7818,
+      "step": 34795
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.938548148148148e-05,
+      "loss": 1.5222,
+      "step": 34800
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.938251851851852e-05,
+      "loss": 1.7544,
+      "step": 34805
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9379555555555553e-05,
+      "loss": 1.6635,
+      "step": 34810
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9376592592592596e-05,
+      "loss": 1.7475,
+      "step": 34815
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9373629629629628e-05,
+      "loss": 1.7178,
+      "step": 34820
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.937066666666667e-05,
+      "loss": 1.6827,
+      "step": 34825
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9367703703703702e-05,
+      "loss": 1.6979,
+      "step": 34830
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9364740740740744e-05,
+      "loss": 1.5967,
+      "step": 34835
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.9361777777777776e-05,
+      "loss": 1.7171,
+      "step": 34840
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.935881481481482e-05,
+      "loss": 1.7448,
+      "step": 34845
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.935585185185185e-05,
+      "loss": 1.7429,
+      "step": 34850
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.935288888888889e-05,
+      "loss": 1.6259,
+      "step": 34855
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9349925925925925e-05,
+      "loss": 1.5977,
+      "step": 34860
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9346962962962963e-05,
+      "loss": 1.5715,
+      "step": 34865
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9344e-05,
+      "loss": 1.7037,
+      "step": 34870
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9341037037037038e-05,
+      "loss": 1.697,
+      "step": 34875
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9338074074074073e-05,
+      "loss": 1.7249,
+      "step": 34880
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9335111111111112e-05,
+      "loss": 1.6468,
+      "step": 34885
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9332148148148147e-05,
+      "loss": 1.6014,
+      "step": 34890
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.9329185185185186e-05,
+      "loss": 1.6939,
+      "step": 34895
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.932622222222222e-05,
+      "loss": 1.6192,
+      "step": 34900
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.932325925925926e-05,
+      "loss": 1.5189,
+      "step": 34905
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9320296296296296e-05,
+      "loss": 1.5842,
+      "step": 34910
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9317333333333335e-05,
+      "loss": 1.7684,
+      "step": 34915
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.931437037037037e-05,
+      "loss": 1.8412,
+      "step": 34920
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.931140740740741e-05,
+      "loss": 1.5999,
+      "step": 34925
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9308444444444444e-05,
+      "loss": 1.7073,
+      "step": 34930
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9305481481481483e-05,
+      "loss": 1.5543,
+      "step": 34935
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.930251851851852e-05,
+      "loss": 1.6312,
+      "step": 34940
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9299555555555558e-05,
+      "loss": 1.7119,
+      "step": 34945
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9296592592592593e-05,
+      "loss": 1.6556,
+      "step": 34950
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.9293629629629632e-05,
+      "loss": 1.7503,
+      "step": 34955
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9290666666666667e-05,
+      "loss": 1.6169,
+      "step": 34960
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9287703703703706e-05,
+      "loss": 1.6711,
+      "step": 34965
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.928474074074074e-05,
+      "loss": 1.6329,
+      "step": 34970
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.928177777777778e-05,
+      "loss": 1.6745,
+      "step": 34975
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9278814814814816e-05,
+      "loss": 1.7142,
+      "step": 34980
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9275851851851855e-05,
+      "loss": 1.7648,
+      "step": 34985
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.927288888888889e-05,
+      "loss": 1.7745,
+      "step": 34990
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.926992592592593e-05,
+      "loss": 1.6848,
+      "step": 34995
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9266962962962964e-05,
+      "loss": 1.6403,
+      "step": 35000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9264000000000003e-05,
+      "loss": 1.6901,
+      "step": 35005
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.926103703703704e-05,
+      "loss": 1.6245,
+      "step": 35010
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.9258074074074077e-05,
+      "loss": 1.7866,
+      "step": 35015
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.925511111111111e-05,
+      "loss": 1.6307,
+      "step": 35020
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.925214814814815e-05,
+      "loss": 1.675,
+      "step": 35025
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9249185185185184e-05,
+      "loss": 1.6596,
+      "step": 35030
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9246222222222226e-05,
+      "loss": 1.745,
+      "step": 35035
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9243259259259258e-05,
+      "loss": 1.8042,
+      "step": 35040
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.92402962962963e-05,
+      "loss": 1.6436,
+      "step": 35045
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9237333333333332e-05,
+      "loss": 1.6487,
+      "step": 35050
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9234370370370374e-05,
+      "loss": 1.7189,
+      "step": 35055
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.9231407407407406e-05,
+      "loss": 1.8007,
+      "step": 35060
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.922844444444445e-05,
+      "loss": 1.6904,
+      "step": 35065
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.922548148148148e-05,
+      "loss": 1.6839,
+      "step": 35070
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9222518518518523e-05,
+      "loss": 1.7326,
+      "step": 35075
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9219555555555555e-05,
+      "loss": 1.7593,
+      "step": 35080
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9216592592592594e-05,
+      "loss": 1.8014,
+      "step": 35085
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.921362962962963e-05,
+      "loss": 1.8002,
+      "step": 35090
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9210666666666668e-05,
+      "loss": 1.6133,
+      "step": 35095
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9207703703703703e-05,
+      "loss": 1.6689,
+      "step": 35100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9204740740740742e-05,
+      "loss": 1.7544,
+      "step": 35105
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9201777777777778e-05,
+      "loss": 1.7256,
+      "step": 35110
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9198814814814816e-05,
+      "loss": 1.6477,
+      "step": 35115
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.9195851851851852e-05,
+      "loss": 1.7478,
+      "step": 35120
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.919288888888889e-05,
+      "loss": 1.731,
+      "step": 35125
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9189925925925926e-05,
+      "loss": 1.7869,
+      "step": 35130
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9186962962962965e-05,
+      "loss": 1.6297,
+      "step": 35135
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9184e-05,
+      "loss": 1.6673,
+      "step": 35140
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.918103703703704e-05,
+      "loss": 1.5686,
+      "step": 35145
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9178074074074075e-05,
+      "loss": 1.6817,
+      "step": 35150
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9175111111111114e-05,
+      "loss": 1.7309,
+      "step": 35155
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.917214814814815e-05,
+      "loss": 1.6829,
+      "step": 35160
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9169185185185188e-05,
+      "loss": 1.7266,
+      "step": 35165
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9166222222222223e-05,
+      "loss": 1.7686,
+      "step": 35170
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9163259259259262e-05,
+      "loss": 1.6987,
+      "step": 35175
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.9160296296296297e-05,
+      "loss": 1.5816,
+      "step": 35180
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9157333333333336e-05,
+      "loss": 1.782,
+      "step": 35185
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9154370370370372e-05,
+      "loss": 1.9132,
+      "step": 35190
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.915140740740741e-05,
+      "loss": 1.846,
+      "step": 35195
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9148444444444446e-05,
+      "loss": 1.5972,
+      "step": 35200
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9145481481481485e-05,
+      "loss": 1.6839,
+      "step": 35205
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.914251851851852e-05,
+      "loss": 1.6688,
+      "step": 35210
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.913955555555556e-05,
+      "loss": 1.6094,
+      "step": 35215
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9136592592592594e-05,
+      "loss": 1.6683,
+      "step": 35220
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9133629629629633e-05,
+      "loss": 1.6496,
+      "step": 35225
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.913066666666667e-05,
+      "loss": 1.6207,
+      "step": 35230
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9127703703703708e-05,
+      "loss": 1.6322,
+      "step": 35235
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.9124740740740743e-05,
+      "loss": 1.7996,
+      "step": 35240
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9121777777777782e-05,
+      "loss": 1.6854,
+      "step": 35245
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9118814814814814e-05,
+      "loss": 1.6989,
+      "step": 35250
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9115851851851856e-05,
+      "loss": 1.6679,
+      "step": 35255
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9112888888888888e-05,
+      "loss": 1.5994,
+      "step": 35260
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.910992592592593e-05,
+      "loss": 1.7172,
+      "step": 35265
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9106962962962962e-05,
+      "loss": 1.7505,
+      "step": 35270
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9104000000000005e-05,
+      "loss": 1.6467,
+      "step": 35275
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9101037037037037e-05,
+      "loss": 1.8069,
+      "step": 35280
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.909807407407408e-05,
+      "loss": 1.7493,
+      "step": 35285
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.909511111111111e-05,
+      "loss": 1.628,
+      "step": 35290
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.9092148148148153e-05,
+      "loss": 1.6491,
+      "step": 35295
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9089185185185185e-05,
+      "loss": 1.5969,
+      "step": 35300
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9086222222222227e-05,
+      "loss": 1.7183,
+      "step": 35305
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.908325925925926e-05,
+      "loss": 1.6179,
+      "step": 35310
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9080296296296298e-05,
+      "loss": 1.7476,
+      "step": 35315
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9077333333333334e-05,
+      "loss": 1.6681,
+      "step": 35320
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9074370370370372e-05,
+      "loss": 1.602,
+      "step": 35325
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9071407407407408e-05,
+      "loss": 1.6958,
+      "step": 35330
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9068444444444447e-05,
+      "loss": 1.6426,
+      "step": 35335
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9065481481481482e-05,
+      "loss": 1.8645,
+      "step": 35340
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.906251851851852e-05,
+      "loss": 1.6872,
+      "step": 35345
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.9059555555555556e-05,
+      "loss": 1.7233,
+      "step": 35350
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9056592592592595e-05,
+      "loss": 1.704,
+      "step": 35355
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.905362962962963e-05,
+      "loss": 1.5488,
+      "step": 35360
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.905066666666667e-05,
+      "loss": 1.7239,
+      "step": 35365
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9047703703703705e-05,
+      "loss": 1.7043,
+      "step": 35370
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9044740740740744e-05,
+      "loss": 1.8472,
+      "step": 35375
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.904177777777778e-05,
+      "loss": 1.7252,
+      "step": 35380
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9038814814814818e-05,
+      "loss": 1.6132,
+      "step": 35385
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9035851851851853e-05,
+      "loss": 1.6126,
+      "step": 35390
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9032888888888892e-05,
+      "loss": 1.7249,
+      "step": 35395
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9029925925925928e-05,
+      "loss": 1.7575,
+      "step": 35400
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.9026962962962967e-05,
+      "loss": 1.699,
+      "step": 35405
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9024000000000002e-05,
+      "loss": 1.5789,
+      "step": 35410
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.902103703703704e-05,
+      "loss": 1.7811,
+      "step": 35415
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9018074074074076e-05,
+      "loss": 1.7265,
+      "step": 35420
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9015111111111115e-05,
+      "loss": 1.6848,
+      "step": 35425
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.901214814814815e-05,
+      "loss": 1.7577,
+      "step": 35430
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.900918518518519e-05,
+      "loss": 1.7382,
+      "step": 35435
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9006222222222225e-05,
+      "loss": 1.6024,
+      "step": 35440
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.9003259259259264e-05,
+      "loss": 1.7424,
+      "step": 35445
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.90002962962963e-05,
+      "loss": 1.6956,
+      "step": 35450
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.8997333333333338e-05,
+      "loss": 1.6871,
+      "step": 35455
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.8994370370370373e-05,
+      "loss": 1.6488,
+      "step": 35460
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.8991407407407412e-05,
+      "loss": 1.6925,
+      "step": 35465
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8988444444444448e-05,
+      "loss": 1.6803,
+      "step": 35470
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.898548148148148e-05,
+      "loss": 1.6696,
+      "step": 35475
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.898251851851852e-05,
+      "loss": 1.6735,
+      "step": 35480
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8979555555555554e-05,
+      "loss": 1.6817,
+      "step": 35485
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8976592592592593e-05,
+      "loss": 1.6523,
+      "step": 35490
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8973629629629628e-05,
+      "loss": 1.5696,
+      "step": 35495
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8970666666666667e-05,
+      "loss": 1.7812,
+      "step": 35500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8967703703703702e-05,
+      "loss": 1.623,
+      "step": 35505
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.896474074074074e-05,
+      "loss": 1.5621,
+      "step": 35510
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8961777777777777e-05,
+      "loss": 1.8169,
+      "step": 35515
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.8958814814814815e-05,
+      "loss": 1.7048,
+      "step": 35520
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.895585185185185e-05,
+      "loss": 1.6495,
+      "step": 35525
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.895288888888889e-05,
+      "loss": 1.6285,
+      "step": 35530
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8949925925925925e-05,
+      "loss": 1.7909,
+      "step": 35535
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8946962962962964e-05,
+      "loss": 1.8344,
+      "step": 35540
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8944e-05,
+      "loss": 1.6961,
+      "step": 35545
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8941037037037038e-05,
+      "loss": 1.6589,
+      "step": 35550
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8938074074074074e-05,
+      "loss": 1.7257,
+      "step": 35555
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8935111111111112e-05,
+      "loss": 1.6998,
+      "step": 35560
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8932148148148148e-05,
+      "loss": 1.6839,
+      "step": 35565
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8929185185185187e-05,
+      "loss": 1.6863,
+      "step": 35570
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.8926222222222222e-05,
+      "loss": 1.7956,
+      "step": 35575
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.892325925925926e-05,
+      "loss": 1.7074,
+      "step": 35580
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8920296296296296e-05,
+      "loss": 1.61,
+      "step": 35585
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8917333333333335e-05,
+      "loss": 1.6168,
+      "step": 35590
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.891437037037037e-05,
+      "loss": 1.6629,
+      "step": 35595
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.891140740740741e-05,
+      "loss": 1.615,
+      "step": 35600
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8908444444444445e-05,
+      "loss": 1.6262,
+      "step": 35605
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8905481481481484e-05,
+      "loss": 1.6818,
+      "step": 35610
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.890251851851852e-05,
+      "loss": 1.6881,
+      "step": 35615
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8899555555555558e-05,
+      "loss": 1.7645,
+      "step": 35620
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8896592592592593e-05,
+      "loss": 1.6151,
+      "step": 35625
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.8893629629629632e-05,
+      "loss": 1.7182,
+      "step": 35630
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8890666666666664e-05,
+      "loss": 1.7784,
+      "step": 35635
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8887703703703706e-05,
+      "loss": 1.7886,
+      "step": 35640
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.888474074074074e-05,
+      "loss": 1.7255,
+      "step": 35645
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.888177777777778e-05,
+      "loss": 1.7832,
+      "step": 35650
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8878814814814813e-05,
+      "loss": 1.7318,
+      "step": 35655
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8875851851851855e-05,
+      "loss": 1.7444,
+      "step": 35660
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8872888888888887e-05,
+      "loss": 1.681,
+      "step": 35665
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.886992592592593e-05,
+      "loss": 1.715,
+      "step": 35670
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.886696296296296e-05,
+      "loss": 1.6661,
+      "step": 35675
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8864000000000004e-05,
+      "loss": 1.6536,
+      "step": 35680
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8861037037037036e-05,
+      "loss": 1.6566,
+      "step": 35685
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.8858074074074078e-05,
+      "loss": 1.6786,
+      "step": 35690
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.885511111111111e-05,
+      "loss": 1.8193,
+      "step": 35695
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8852148148148152e-05,
+      "loss": 1.7458,
+      "step": 35700
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8849185185185184e-05,
+      "loss": 1.7614,
+      "step": 35705
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8846222222222223e-05,
+      "loss": 1.7476,
+      "step": 35710
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.884325925925926e-05,
+      "loss": 1.7565,
+      "step": 35715
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8840296296296297e-05,
+      "loss": 1.5851,
+      "step": 35720
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8837333333333333e-05,
+      "loss": 1.619,
+      "step": 35725
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.883437037037037e-05,
+      "loss": 1.7964,
+      "step": 35730
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8831407407407407e-05,
+      "loss": 1.7471,
+      "step": 35735
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.8828444444444446e-05,
+      "loss": 1.6224,
+      "step": 35740
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.882548148148148e-05,
+      "loss": 1.6818,
+      "step": 35745
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.882251851851852e-05,
+      "loss": 1.6923,
+      "step": 35750
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8819555555555555e-05,
+      "loss": 1.6103,
+      "step": 35755
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8816592592592594e-05,
+      "loss": 1.6712,
+      "step": 35760
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.881362962962963e-05,
+      "loss": 1.7183,
+      "step": 35765
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.881066666666667e-05,
+      "loss": 1.7547,
+      "step": 35770
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8807703703703704e-05,
+      "loss": 1.6302,
+      "step": 35775
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8804740740740743e-05,
+      "loss": 1.6706,
+      "step": 35780
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8801777777777778e-05,
+      "loss": 1.5911,
+      "step": 35785
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8798814814814817e-05,
+      "loss": 1.6637,
+      "step": 35790
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.8795851851851852e-05,
+      "loss": 1.7942,
+      "step": 35795
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.879288888888889e-05,
+      "loss": 1.5871,
+      "step": 35800
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8789925925925927e-05,
+      "loss": 1.7172,
+      "step": 35805
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8786962962962965e-05,
+      "loss": 1.7684,
+      "step": 35810
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8784e-05,
+      "loss": 1.7756,
+      "step": 35815
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.878103703703704e-05,
+      "loss": 1.6873,
+      "step": 35820
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8778074074074075e-05,
+      "loss": 1.7449,
+      "step": 35825
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8775111111111114e-05,
+      "loss": 1.7066,
+      "step": 35830
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.877214814814815e-05,
+      "loss": 1.5802,
+      "step": 35835
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8769185185185188e-05,
+      "loss": 1.6759,
+      "step": 35840
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8766222222222224e-05,
+      "loss": 1.7996,
+      "step": 35845
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8763259259259262e-05,
+      "loss": 1.633,
+      "step": 35850
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.8760296296296298e-05,
+      "loss": 1.6693,
+      "step": 35855
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8757333333333337e-05,
+      "loss": 1.7648,
+      "step": 35860
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.875437037037037e-05,
+      "loss": 1.7205,
+      "step": 35865
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.875140740740741e-05,
+      "loss": 1.6858,
+      "step": 35870
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8748444444444443e-05,
+      "loss": 1.5705,
+      "step": 35875
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8745481481481485e-05,
+      "loss": 1.6418,
+      "step": 35880
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8742518518518517e-05,
+      "loss": 1.8294,
+      "step": 35885
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.873955555555556e-05,
+      "loss": 1.6743,
+      "step": 35890
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.873659259259259e-05,
+      "loss": 1.6596,
+      "step": 35895
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8733629629629634e-05,
+      "loss": 1.7387,
+      "step": 35900
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8730666666666666e-05,
+      "loss": 1.6302,
+      "step": 35905
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.8727703703703708e-05,
+      "loss": 1.6678,
+      "step": 35910
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.872474074074074e-05,
+      "loss": 1.7268,
+      "step": 35915
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8721777777777782e-05,
+      "loss": 1.6373,
+      "step": 35920
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8718814814814814e-05,
+      "loss": 1.6894,
+      "step": 35925
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8715851851851857e-05,
+      "loss": 1.6514,
+      "step": 35930
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.871288888888889e-05,
+      "loss": 1.745,
+      "step": 35935
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8709925925925927e-05,
+      "loss": 1.8065,
+      "step": 35940
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8706962962962963e-05,
+      "loss": 1.8386,
+      "step": 35945
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8704e-05,
+      "loss": 1.798,
+      "step": 35950
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8701037037037037e-05,
+      "loss": 1.6165,
+      "step": 35955
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.8698074074074076e-05,
+      "loss": 1.6324,
+      "step": 35960
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.869511111111111e-05,
+      "loss": 1.5644,
+      "step": 35965
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.869214814814815e-05,
+      "loss": 1.6807,
+      "step": 35970
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8689185185185186e-05,
+      "loss": 1.6495,
+      "step": 35975
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8686222222222224e-05,
+      "loss": 1.8092,
+      "step": 35980
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.868325925925926e-05,
+      "loss": 1.6521,
+      "step": 35985
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.86802962962963e-05,
+      "loss": 1.6443,
+      "step": 35990
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8677333333333334e-05,
+      "loss": 1.7089,
+      "step": 35995
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8674370370370373e-05,
+      "loss": 1.7755,
+      "step": 36000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.867140740740741e-05,
+      "loss": 1.7712,
+      "step": 36005
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8668444444444447e-05,
+      "loss": 1.6435,
+      "step": 36010
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8665481481481483e-05,
+      "loss": 1.7075,
+      "step": 36015
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.866251851851852e-05,
+      "loss": 1.7878,
+      "step": 36020
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.8659555555555557e-05,
+      "loss": 1.6686,
+      "step": 36025
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8656592592592596e-05,
+      "loss": 1.716,
+      "step": 36030
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.865362962962963e-05,
+      "loss": 1.6834,
+      "step": 36035
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.865066666666667e-05,
+      "loss": 1.7007,
+      "step": 36040
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8647703703703705e-05,
+      "loss": 1.6985,
+      "step": 36045
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8644740740740744e-05,
+      "loss": 1.62,
+      "step": 36050
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.864177777777778e-05,
+      "loss": 1.6819,
+      "step": 36055
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.863881481481482e-05,
+      "loss": 1.6782,
+      "step": 36060
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8635851851851854e-05,
+      "loss": 1.6066,
+      "step": 36065
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8632888888888893e-05,
+      "loss": 1.5761,
+      "step": 36070
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8629925925925928e-05,
+      "loss": 1.6755,
+      "step": 36075
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.8626962962962967e-05,
+      "loss": 1.6864,
+      "step": 36080
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8624000000000002e-05,
+      "loss": 1.6985,
+      "step": 36085
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.862103703703704e-05,
+      "loss": 1.7328,
+      "step": 36090
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8618074074074073e-05,
+      "loss": 1.7107,
+      "step": 36095
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8615111111111115e-05,
+      "loss": 1.7662,
+      "step": 36100
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8612148148148148e-05,
+      "loss": 1.7049,
+      "step": 36105
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.860918518518519e-05,
+      "loss": 1.6431,
+      "step": 36110
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8606222222222222e-05,
+      "loss": 1.6793,
+      "step": 36115
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8603259259259264e-05,
+      "loss": 1.7822,
+      "step": 36120
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8600296296296296e-05,
+      "loss": 1.7163,
+      "step": 36125
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8597333333333338e-05,
+      "loss": 1.764,
+      "step": 36130
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.859437037037037e-05,
+      "loss": 1.767,
+      "step": 36135
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 2.8591407407407413e-05,
+      "loss": 1.7765,
+      "step": 36140
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8588444444444445e-05,
+      "loss": 1.6454,
+      "step": 36145
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8585481481481487e-05,
+      "loss": 1.7626,
+      "step": 36150
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.858251851851852e-05,
+      "loss": 1.7795,
+      "step": 36155
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.857955555555556e-05,
+      "loss": 1.6693,
+      "step": 36160
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8576592592592593e-05,
+      "loss": 1.74,
+      "step": 36165
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8573629629629632e-05,
+      "loss": 1.7159,
+      "step": 36170
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8570666666666667e-05,
+      "loss": 1.651,
+      "step": 36175
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8567703703703706e-05,
+      "loss": 1.7222,
+      "step": 36180
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.856474074074074e-05,
+      "loss": 1.7134,
+      "step": 36185
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.856177777777778e-05,
+      "loss": 1.7167,
+      "step": 36190
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 2.8558814814814816e-05,
+      "loss": 1.6719,
+      "step": 36195
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8555851851851855e-05,
+      "loss": 1.6683,
+      "step": 36200
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.855288888888889e-05,
+      "loss": 1.8396,
+      "step": 36205
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.854992592592593e-05,
+      "loss": 1.6848,
+      "step": 36210
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8546962962962964e-05,
+      "loss": 1.6915,
+      "step": 36215
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8544000000000003e-05,
+      "loss": 1.6469,
+      "step": 36220
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.854103703703704e-05,
+      "loss": 1.7248,
+      "step": 36225
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8538074074074077e-05,
+      "loss": 1.6567,
+      "step": 36230
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8535111111111113e-05,
+      "loss": 1.5925,
+      "step": 36235
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.853214814814815e-05,
+      "loss": 1.6462,
+      "step": 36240
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.8529185185185187e-05,
+      "loss": 1.6808,
+      "step": 36245
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 2.852622222222222e-05,
+      "loss": 1.62,
+      "step": 36250
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.852325925925926e-05,
+      "loss": 1.8099,
+      "step": 36255
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8520296296296293e-05,
+      "loss": 1.7827,
+      "step": 36260
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8517333333333336e-05,
+      "loss": 1.7174,
+      "step": 36265
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8514370370370368e-05,
+      "loss": 1.768,
+      "step": 36270
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.851140740740741e-05,
+      "loss": 1.6862,
+      "step": 36275
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8508444444444442e-05,
+      "loss": 1.6689,
+      "step": 36280
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8505481481481484e-05,
+      "loss": 1.743,
+      "step": 36285
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8502518518518516e-05,
+      "loss": 1.7272,
+      "step": 36290
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.849955555555556e-05,
+      "loss": 1.8075,
+      "step": 36295
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.849659259259259e-05,
+      "loss": 1.6846,
+      "step": 36300
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 2.8493629629629633e-05,
+      "loss": 1.7026,
+      "step": 36305
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8490666666666665e-05,
+      "loss": 1.7219,
+      "step": 36310
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8487703703703707e-05,
+      "loss": 1.7353,
+      "step": 36315
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.848474074074074e-05,
+      "loss": 1.6455,
+      "step": 36320
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8481777777777778e-05,
+      "loss": 1.6074,
+      "step": 36325
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8478814814814813e-05,
+      "loss": 1.744,
+      "step": 36330
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8475851851851852e-05,
+      "loss": 1.6917,
+      "step": 36335
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8472888888888887e-05,
+      "loss": 1.6971,
+      "step": 36340
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8469925925925926e-05,
+      "loss": 1.5893,
+      "step": 36345
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8466962962962962e-05,
+      "loss": 1.5295,
+      "step": 36350
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8464e-05,
+      "loss": 1.6715,
+      "step": 36355
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8461037037037036e-05,
+      "loss": 1.6306,
+      "step": 36360
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 2.8458074074074075e-05,
+      "loss": 1.7374,
+      "step": 36365
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.845511111111111e-05,
+      "loss": 1.6939,
+      "step": 36370
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.845214814814815e-05,
+      "loss": 1.6596,
+      "step": 36375
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8449185185185185e-05,
+      "loss": 1.7467,
+      "step": 36380
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8446222222222223e-05,
+      "loss": 1.6553,
+      "step": 36385
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.844325925925926e-05,
+      "loss": 1.7137,
+      "step": 36390
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8440296296296298e-05,
+      "loss": 1.6375,
+      "step": 36395
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8437333333333333e-05,
+      "loss": 1.5801,
+      "step": 36400
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8434370370370372e-05,
+      "loss": 1.7203,
+      "step": 36405
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8431407407407407e-05,
+      "loss": 1.7435,
+      "step": 36410
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.8428444444444446e-05,
+      "loss": 1.8046,
+      "step": 36415
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 2.842548148148148e-05,
+      "loss": 1.7054,
+      "step": 36420
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.842251851851852e-05,
+      "loss": 1.6894,
+      "step": 36425
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8419555555555556e-05,
+      "loss": 1.7202,
+      "step": 36430
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8416592592592595e-05,
+      "loss": 1.7252,
+      "step": 36435
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.841362962962963e-05,
+      "loss": 1.7016,
+      "step": 36440
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.841066666666667e-05,
+      "loss": 1.5628,
+      "step": 36445
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8407703703703704e-05,
+      "loss": 1.6319,
+      "step": 36450
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8404740740740743e-05,
+      "loss": 1.7028,
+      "step": 36455
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.840177777777778e-05,
+      "loss": 1.6473,
+      "step": 36460
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8398814814814817e-05,
+      "loss": 1.6792,
+      "step": 36465
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.8395851851851853e-05,
+      "loss": 1.7081,
+      "step": 36470
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 2.839288888888889e-05,
+      "loss": 1.6674,
+      "step": 36475
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8389925925925924e-05,
+      "loss": 1.777,
+      "step": 36480
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8386962962962966e-05,
+      "loss": 1.6902,
+      "step": 36485
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8383999999999998e-05,
+      "loss": 1.6702,
+      "step": 36490
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.838103703703704e-05,
+      "loss": 1.7862,
+      "step": 36495
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8378074074074072e-05,
+      "loss": 1.6143,
+      "step": 36500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8375111111111114e-05,
+      "loss": 1.7763,
+      "step": 36505
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8372148148148146e-05,
+      "loss": 1.5738,
+      "step": 36510
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.836918518518519e-05,
+      "loss": 1.763,
+      "step": 36515
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.836622222222222e-05,
+      "loss": 1.8592,
+      "step": 36520
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8363259259259263e-05,
+      "loss": 1.6954,
+      "step": 36525
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 2.8360296296296295e-05,
+      "loss": 1.811,
+      "step": 36530
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8357333333333337e-05,
+      "loss": 1.7427,
+      "step": 36535
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.835437037037037e-05,
+      "loss": 1.7426,
+      "step": 36540
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.835140740740741e-05,
+      "loss": 1.7703,
+      "step": 36545
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8348444444444443e-05,
+      "loss": 1.7319,
+      "step": 36550
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8345481481481482e-05,
+      "loss": 1.7613,
+      "step": 36555
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8342518518518518e-05,
+      "loss": 1.7009,
+      "step": 36560
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8339555555555557e-05,
+      "loss": 1.7612,
+      "step": 36565
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8336592592592592e-05,
+      "loss": 1.6782,
+      "step": 36570
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.833362962962963e-05,
+      "loss": 1.7726,
+      "step": 36575
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8330666666666666e-05,
+      "loss": 1.6242,
+      "step": 36580
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.8327703703703705e-05,
+      "loss": 1.7328,
+      "step": 36585
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 2.832474074074074e-05,
+      "loss": 1.6987,
+      "step": 36590
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.832177777777778e-05,
+      "loss": 1.7388,
+      "step": 36595
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8318814814814815e-05,
+      "loss": 1.6625,
+      "step": 36600
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8315851851851854e-05,
+      "loss": 1.6675,
+      "step": 36605
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.831288888888889e-05,
+      "loss": 1.6722,
+      "step": 36610
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8309925925925928e-05,
+      "loss": 1.576,
+      "step": 36615
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8306962962962963e-05,
+      "loss": 1.6292,
+      "step": 36620
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8304000000000002e-05,
+      "loss": 1.6363,
+      "step": 36625
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8301037037037038e-05,
+      "loss": 1.6855,
+      "step": 36630
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8298074074074076e-05,
+      "loss": 1.6527,
+      "step": 36635
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.8295111111111112e-05,
+      "loss": 1.6172,
+      "step": 36640
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 2.829214814814815e-05,
+      "loss": 1.7243,
+      "step": 36645
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8289185185185186e-05,
+      "loss": 1.6595,
+      "step": 36650
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8286222222222225e-05,
+      "loss": 1.617,
+      "step": 36655
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.828325925925926e-05,
+      "loss": 1.6275,
+      "step": 36660
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.82802962962963e-05,
+      "loss": 1.6912,
+      "step": 36665
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8277333333333335e-05,
+      "loss": 1.5998,
+      "step": 36670
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8274370370370373e-05,
+      "loss": 1.7649,
+      "step": 36675
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.827140740740741e-05,
+      "loss": 1.6509,
+      "step": 36680
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8268444444444448e-05,
+      "loss": 1.7309,
+      "step": 36685
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8265481481481483e-05,
+      "loss": 1.7929,
+      "step": 36690
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8262518518518522e-05,
+      "loss": 1.6679,
+      "step": 36695
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 2.8259555555555557e-05,
+      "loss": 1.6754,
+      "step": 36700
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8256592592592596e-05,
+      "loss": 1.6957,
+      "step": 36705
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.825362962962963e-05,
+      "loss": 1.794,
+      "step": 36710
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.825066666666667e-05,
+      "loss": 1.744,
+      "step": 36715
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8247703703703702e-05,
+      "loss": 1.7253,
+      "step": 36720
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8244740740740745e-05,
+      "loss": 1.6742,
+      "step": 36725
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8241777777777777e-05,
+      "loss": 1.662,
+      "step": 36730
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.823881481481482e-05,
+      "loss": 1.7324,
+      "step": 36735
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.823585185185185e-05,
+      "loss": 1.687,
+      "step": 36740
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8232888888888893e-05,
+      "loss": 1.7163,
+      "step": 36745
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8229925925925925e-05,
+      "loss": 1.7221,
+      "step": 36750
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 2.8226962962962967e-05,
+      "loss": 1.7284,
+      "step": 36755
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8224e-05,
+      "loss": 1.7503,
+      "step": 36760
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.822103703703704e-05,
+      "loss": 1.6126,
+      "step": 36765
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8218074074074074e-05,
+      "loss": 1.7407,
+      "step": 36770
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8215111111111116e-05,
+      "loss": 1.5999,
+      "step": 36775
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8212148148148148e-05,
+      "loss": 1.7796,
+      "step": 36780
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8209185185185187e-05,
+      "loss": 1.7596,
+      "step": 36785
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8206222222222222e-05,
+      "loss": 1.608,
+      "step": 36790
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.820325925925926e-05,
+      "loss": 1.7843,
+      "step": 36795
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8200296296296296e-05,
+      "loss": 1.7297,
+      "step": 36800
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.8197333333333335e-05,
+      "loss": 1.6764,
+      "step": 36805
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.819437037037037e-05,
+      "loss": 1.6951,
+      "step": 36810
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 2.819140740740741e-05,
+      "loss": 1.5253,
+      "step": 36815
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8188444444444445e-05,
+      "loss": 1.6556,
+      "step": 36820
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8185481481481484e-05,
+      "loss": 1.7757,
+      "step": 36825
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.818251851851852e-05,
+      "loss": 1.638,
+      "step": 36830
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8179555555555558e-05,
+      "loss": 1.6229,
+      "step": 36835
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8176592592592594e-05,
+      "loss": 1.7166,
+      "step": 36840
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8173629629629632e-05,
+      "loss": 1.7254,
+      "step": 36845
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8170666666666668e-05,
+      "loss": 1.7394,
+      "step": 36850
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8167703703703707e-05,
+      "loss": 1.7183,
+      "step": 36855
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8164740740740742e-05,
+      "loss": 1.6788,
+      "step": 36860
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.816177777777778e-05,
+      "loss": 1.5688,
+      "step": 36865
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 2.8158814814814816e-05,
+      "loss": 1.7802,
+      "step": 36870
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8155851851851855e-05,
+      "loss": 1.7025,
+      "step": 36875
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.815288888888889e-05,
+      "loss": 1.6699,
+      "step": 36880
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.814992592592593e-05,
+      "loss": 1.6333,
+      "step": 36885
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8146962962962965e-05,
+      "loss": 1.6357,
+      "step": 36890
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8144000000000004e-05,
+      "loss": 1.7468,
+      "step": 36895
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.814103703703704e-05,
+      "loss": 1.746,
+      "step": 36900
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8138074074074078e-05,
+      "loss": 1.726,
+      "step": 36905
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8135111111111113e-05,
+      "loss": 1.7173,
+      "step": 36910
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8132148148148152e-05,
+      "loss": 1.6483,
+      "step": 36915
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8129185185185188e-05,
+      "loss": 1.6667,
+      "step": 36920
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 2.8126222222222226e-05,
+      "loss": 1.5941,
+      "step": 36925
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8123259259259262e-05,
+      "loss": 1.6653,
+      "step": 36930
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.81202962962963e-05,
+      "loss": 1.7748,
+      "step": 36935
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8117333333333336e-05,
+      "loss": 1.7226,
+      "step": 36940
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8114370370370375e-05,
+      "loss": 1.7881,
+      "step": 36945
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8111407407407407e-05,
+      "loss": 1.6415,
+      "step": 36950
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.810844444444445e-05,
+      "loss": 1.5883,
+      "step": 36955
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.810548148148148e-05,
+      "loss": 1.6338,
+      "step": 36960
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8102518518518523e-05,
+      "loss": 1.6433,
+      "step": 36965
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8099555555555555e-05,
+      "loss": 1.6955,
+      "step": 36970
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.8096592592592598e-05,
+      "loss": 1.701,
+      "step": 36975
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 2.809362962962963e-05,
+      "loss": 1.6956,
+      "step": 36980
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8090666666666672e-05,
+      "loss": 1.6669,
+      "step": 36985
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8087703703703704e-05,
+      "loss": 1.6975,
+      "step": 36990
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8084740740740746e-05,
+      "loss": 1.5879,
+      "step": 36995
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8081777777777778e-05,
+      "loss": 1.7765,
+      "step": 37000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.807881481481482e-05,
+      "loss": 1.7033,
+      "step": 37005
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8075851851851852e-05,
+      "loss": 1.667,
+      "step": 37010
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8072888888888888e-05,
+      "loss": 1.7115,
+      "step": 37015
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8069925925925927e-05,
+      "loss": 1.6972,
+      "step": 37020
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8066962962962962e-05,
+      "loss": 1.6282,
+      "step": 37025
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8064e-05,
+      "loss": 1.7011,
+      "step": 37030
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8061037037037036e-05,
+      "loss": 1.6973,
+      "step": 37035
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 2.8058074074074075e-05,
+      "loss": 1.6613,
+      "step": 37040
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.805511111111111e-05,
+      "loss": 1.5986,
+      "step": 37045
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.805214814814815e-05,
+      "loss": 1.68,
+      "step": 37050
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8049185185185185e-05,
+      "loss": 1.8451,
+      "step": 37055
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8046222222222224e-05,
+      "loss": 1.7597,
+      "step": 37060
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.804325925925926e-05,
+      "loss": 1.7371,
+      "step": 37065
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8040296296296298e-05,
+      "loss": 1.7445,
+      "step": 37070
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8037333333333333e-05,
+      "loss": 1.7925,
+      "step": 37075
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8034370370370372e-05,
+      "loss": 1.7462,
+      "step": 37080
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8031407407407408e-05,
+      "loss": 1.7,
+      "step": 37085
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8028444444444447e-05,
+      "loss": 1.7781,
+      "step": 37090
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 2.8025481481481482e-05,
+      "loss": 1.7328,
+      "step": 37095
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.802251851851852e-05,
+      "loss": 1.6862,
+      "step": 37100
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8019555555555553e-05,
+      "loss": 1.6607,
+      "step": 37105
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8016592592592595e-05,
+      "loss": 1.6926,
+      "step": 37110
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8013629629629627e-05,
+      "loss": 1.6385,
+      "step": 37115
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.801066666666667e-05,
+      "loss": 1.8231,
+      "step": 37120
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.80077037037037e-05,
+      "loss": 1.6058,
+      "step": 37125
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8004740740740744e-05,
+      "loss": 1.6244,
+      "step": 37130
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.8001777777777776e-05,
+      "loss": 1.8306,
+      "step": 37135
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.7998814814814818e-05,
+      "loss": 1.722,
+      "step": 37140
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.799585185185185e-05,
+      "loss": 1.7487,
+      "step": 37145
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 2.7992888888888892e-05,
+      "loss": 1.7595,
+      "step": 37150
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7989925925925924e-05,
+      "loss": 1.7614,
+      "step": 37155
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7986962962962966e-05,
+      "loss": 1.6809,
+      "step": 37160
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7984e-05,
+      "loss": 1.695,
+      "step": 37165
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.798103703703704e-05,
+      "loss": 1.6628,
+      "step": 37170
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7978074074074073e-05,
+      "loss": 1.6343,
+      "step": 37175
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.797511111111111e-05,
+      "loss": 1.7839,
+      "step": 37180
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7972148148148147e-05,
+      "loss": 1.7136,
+      "step": 37185
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7969185185185186e-05,
+      "loss": 1.7572,
+      "step": 37190
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.796622222222222e-05,
+      "loss": 1.7662,
+      "step": 37195
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.796325925925926e-05,
+      "loss": 1.5771,
+      "step": 37200
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 2.7960296296296295e-05,
+      "loss": 1.6898,
+      "step": 37205
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7957333333333334e-05,
+      "loss": 1.7676,
+      "step": 37210
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.795437037037037e-05,
+      "loss": 1.6991,
+      "step": 37215
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.795140740740741e-05,
+      "loss": 1.5731,
+      "step": 37220
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7948444444444444e-05,
+      "loss": 1.6064,
+      "step": 37225
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7945481481481483e-05,
+      "loss": 1.6628,
+      "step": 37230
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7942518518518518e-05,
+      "loss": 1.6861,
+      "step": 37235
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7939555555555557e-05,
+      "loss": 1.7704,
+      "step": 37240
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7936592592592592e-05,
+      "loss": 1.6042,
+      "step": 37245
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.793362962962963e-05,
+      "loss": 1.7919,
+      "step": 37250
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7930666666666667e-05,
+      "loss": 1.7071,
+      "step": 37255
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.7927703703703706e-05,
+      "loss": 1.8354,
+      "step": 37260
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.792474074074074e-05,
+      "loss": 1.678,
+      "step": 37265
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.792177777777778e-05,
+      "loss": 1.8012,
+      "step": 37270
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7918814814814815e-05,
+      "loss": 1.67,
+      "step": 37275
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7915851851851854e-05,
+      "loss": 1.7499,
+      "step": 37280
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.791288888888889e-05,
+      "loss": 1.6338,
+      "step": 37285
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7909925925925928e-05,
+      "loss": 1.6439,
+      "step": 37290
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7906962962962964e-05,
+      "loss": 1.8399,
+      "step": 37295
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7904000000000003e-05,
+      "loss": 1.7093,
+      "step": 37300
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7901037037037038e-05,
+      "loss": 1.759,
+      "step": 37305
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7898074074074077e-05,
+      "loss": 1.7399,
+      "step": 37310
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.7895111111111112e-05,
+      "loss": 1.6921,
+      "step": 37315
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.789214814814815e-05,
+      "loss": 1.6583,
+      "step": 37320
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7889185185185186e-05,
+      "loss": 1.6882,
+      "step": 37325
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7886222222222225e-05,
+      "loss": 1.7021,
+      "step": 37330
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7883259259259257e-05,
+      "loss": 1.6507,
+      "step": 37335
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.78802962962963e-05,
+      "loss": 1.7626,
+      "step": 37340
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.787792592592593e-05,
+      "loss": 1.6454,
+      "step": 37345
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7874962962962965e-05,
+      "loss": 1.8968,
+      "step": 37350
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7872000000000004e-05,
+      "loss": 1.7958,
+      "step": 37355
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.786903703703704e-05,
+      "loss": 1.636,
+      "step": 37360
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7866074074074078e-05,
+      "loss": 1.7187,
+      "step": 37365
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7863111111111113e-05,
+      "loss": 1.68,
+      "step": 37370
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.7860148148148152e-05,
+      "loss": 1.6245,
+      "step": 37375
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7857185185185187e-05,
+      "loss": 1.7621,
+      "step": 37380
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7854222222222226e-05,
+      "loss": 1.6942,
+      "step": 37385
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7851259259259262e-05,
+      "loss": 1.6967,
+      "step": 37390
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.78482962962963e-05,
+      "loss": 1.6773,
+      "step": 37395
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7845333333333336e-05,
+      "loss": 1.6733,
+      "step": 37400
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7842370370370375e-05,
+      "loss": 1.6554,
+      "step": 37405
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7839407407407407e-05,
+      "loss": 1.6194,
+      "step": 37410
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.783644444444445e-05,
+      "loss": 1.8113,
+      "step": 37415
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.783348148148148e-05,
+      "loss": 1.7262,
+      "step": 37420
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7830518518518523e-05,
+      "loss": 1.9016,
+      "step": 37425
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7827555555555555e-05,
+      "loss": 1.7025,
+      "step": 37430
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7824592592592598e-05,
+      "loss": 1.6478,
+      "step": 37435
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.782162962962963e-05,
+      "loss": 1.752,
+      "step": 37440
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7818666666666672e-05,
+      "loss": 1.8082,
+      "step": 37445
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7815703703703704e-05,
+      "loss": 1.6672,
+      "step": 37450
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7812740740740746e-05,
+      "loss": 1.6883,
+      "step": 37455
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7809777777777778e-05,
+      "loss": 1.7377,
+      "step": 37460
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.780681481481482e-05,
+      "loss": 1.6921,
+      "step": 37465
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7803851851851852e-05,
+      "loss": 1.7795,
+      "step": 37470
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7800888888888895e-05,
+      "loss": 1.5579,
+      "step": 37475
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7797925925925927e-05,
+      "loss": 1.6036,
+      "step": 37480
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7794962962962962e-05,
+      "loss": 1.6752,
+      "step": 37485
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.7792e-05,
+      "loss": 1.6855,
+      "step": 37490
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7789037037037036e-05,
+      "loss": 1.6936,
+      "step": 37495
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7786074074074075e-05,
+      "loss": 1.7692,
+      "step": 37500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.778311111111111e-05,
+      "loss": 1.7005,
+      "step": 37505
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.778014814814815e-05,
+      "loss": 1.5449,
+      "step": 37510
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7777185185185185e-05,
+      "loss": 1.5893,
+      "step": 37515
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7774222222222224e-05,
+      "loss": 1.5849,
+      "step": 37520
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.777125925925926e-05,
+      "loss": 1.787,
+      "step": 37525
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7768296296296298e-05,
+      "loss": 1.6223,
+      "step": 37530
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7765333333333333e-05,
+      "loss": 1.6336,
+      "step": 37535
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7762370370370372e-05,
+      "loss": 1.7443,
+      "step": 37540
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.7759407407407408e-05,
+      "loss": 1.6701,
+      "step": 37545
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7756444444444446e-05,
+      "loss": 1.8236,
+      "step": 37550
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7753481481481482e-05,
+      "loss": 1.687,
+      "step": 37555
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.775051851851852e-05,
+      "loss": 1.7662,
+      "step": 37560
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7747555555555553e-05,
+      "loss": 1.7042,
+      "step": 37565
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7744592592592595e-05,
+      "loss": 1.6591,
+      "step": 37570
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7741629629629627e-05,
+      "loss": 1.738,
+      "step": 37575
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.773866666666667e-05,
+      "loss": 1.6957,
+      "step": 37580
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.77357037037037e-05,
+      "loss": 1.7032,
+      "step": 37585
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7732740740740743e-05,
+      "loss": 1.795,
+      "step": 37590
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7729777777777776e-05,
+      "loss": 1.86,
+      "step": 37595
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.7726814814814818e-05,
+      "loss": 1.644,
+      "step": 37600
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.772385185185185e-05,
+      "loss": 1.7699,
+      "step": 37605
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7720888888888892e-05,
+      "loss": 1.6672,
+      "step": 37610
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7717925925925924e-05,
+      "loss": 1.6136,
+      "step": 37615
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7714962962962966e-05,
+      "loss": 1.6395,
+      "step": 37620
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7711999999999998e-05,
+      "loss": 1.6436,
+      "step": 37625
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.770903703703704e-05,
+      "loss": 1.6875,
+      "step": 37630
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7706074074074073e-05,
+      "loss": 1.8121,
+      "step": 37635
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.770311111111111e-05,
+      "loss": 1.7108,
+      "step": 37640
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7700148148148147e-05,
+      "loss": 1.7058,
+      "step": 37645
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.7697185185185186e-05,
+      "loss": 1.68,
+      "step": 37650
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.769422222222222e-05,
+      "loss": 1.7114,
+      "step": 37655
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.769125925925926e-05,
+      "loss": 1.6908,
+      "step": 37660
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7688296296296295e-05,
+      "loss": 1.663,
+      "step": 37665
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7685333333333334e-05,
+      "loss": 1.7002,
+      "step": 37670
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.768237037037037e-05,
+      "loss": 1.8325,
+      "step": 37675
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.767940740740741e-05,
+      "loss": 1.6941,
+      "step": 37680
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7676444444444444e-05,
+      "loss": 1.768,
+      "step": 37685
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7673481481481483e-05,
+      "loss": 1.6701,
+      "step": 37690
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7670518518518518e-05,
+      "loss": 1.6397,
+      "step": 37695
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7667555555555557e-05,
+      "loss": 1.5641,
+      "step": 37700
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7664592592592592e-05,
+      "loss": 1.7797,
+      "step": 37705
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.766162962962963e-05,
+      "loss": 1.6815,
+      "step": 37710
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.7658666666666667e-05,
+      "loss": 1.7425,
+      "step": 37715
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7655703703703705e-05,
+      "loss": 1.6911,
+      "step": 37720
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.765274074074074e-05,
+      "loss": 1.8086,
+      "step": 37725
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.764977777777778e-05,
+      "loss": 1.7508,
+      "step": 37730
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7646814814814815e-05,
+      "loss": 1.6767,
+      "step": 37735
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7643851851851854e-05,
+      "loss": 1.5837,
+      "step": 37740
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.764088888888889e-05,
+      "loss": 1.6061,
+      "step": 37745
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7637925925925928e-05,
+      "loss": 1.6955,
+      "step": 37750
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7634962962962964e-05,
+      "loss": 1.7186,
+      "step": 37755
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7632000000000002e-05,
+      "loss": 1.6586,
+      "step": 37760
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7629037037037038e-05,
+      "loss": 1.61,
+      "step": 37765
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.7626074074074077e-05,
+      "loss": 1.6162,
+      "step": 37770
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7623111111111112e-05,
+      "loss": 1.7439,
+      "step": 37775
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.762014814814815e-05,
+      "loss": 1.8286,
+      "step": 37780
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7617185185185186e-05,
+      "loss": 1.777,
+      "step": 37785
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7614222222222225e-05,
+      "loss": 1.8202,
+      "step": 37790
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.761125925925926e-05,
+      "loss": 1.7619,
+      "step": 37795
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.76082962962963e-05,
+      "loss": 1.6695,
+      "step": 37800
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.760533333333333e-05,
+      "loss": 1.6611,
+      "step": 37805
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7602370370370374e-05,
+      "loss": 1.695,
+      "step": 37810
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7599407407407406e-05,
+      "loss": 1.6386,
+      "step": 37815
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.7596444444444448e-05,
+      "loss": 1.7514,
+      "step": 37820
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.759348148148148e-05,
+      "loss": 1.688,
+      "step": 37825
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7590518518518522e-05,
+      "loss": 1.6023,
+      "step": 37830
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7587555555555554e-05,
+      "loss": 1.6761,
+      "step": 37835
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7584592592592596e-05,
+      "loss": 1.6187,
+      "step": 37840
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.758162962962963e-05,
+      "loss": 1.7075,
+      "step": 37845
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.757866666666667e-05,
+      "loss": 1.7353,
+      "step": 37850
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7575703703703703e-05,
+      "loss": 1.7485,
+      "step": 37855
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7572740740740745e-05,
+      "loss": 1.6554,
+      "step": 37860
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7569777777777777e-05,
+      "loss": 1.7078,
+      "step": 37865
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.7566814814814816e-05,
+      "loss": 1.6511,
+      "step": 37870
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.756385185185185e-05,
+      "loss": 1.7261,
+      "step": 37875
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.756088888888889e-05,
+      "loss": 1.6764,
+      "step": 37880
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7557925925925926e-05,
+      "loss": 1.6939,
+      "step": 37885
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7554962962962964e-05,
+      "loss": 1.709,
+      "step": 37890
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7552e-05,
+      "loss": 1.7082,
+      "step": 37895
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.754903703703704e-05,
+      "loss": 1.7022,
+      "step": 37900
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7546074074074074e-05,
+      "loss": 1.7175,
+      "step": 37905
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7543111111111113e-05,
+      "loss": 1.7595,
+      "step": 37910
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.754014814814815e-05,
+      "loss": 1.7246,
+      "step": 37915
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7537185185185187e-05,
+      "loss": 1.7699,
+      "step": 37920
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7534222222222223e-05,
+      "loss": 1.7495,
+      "step": 37925
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.753125925925926e-05,
+      "loss": 1.6983,
+      "step": 37930
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7528296296296297e-05,
+      "loss": 1.6816,
+      "step": 37935
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.7525333333333336e-05,
+      "loss": 1.6985,
+      "step": 37940
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.752237037037037e-05,
+      "loss": 1.7094,
+      "step": 37945
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.751940740740741e-05,
+      "loss": 1.5645,
+      "step": 37950
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7516444444444445e-05,
+      "loss": 1.6342,
+      "step": 37955
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7513481481481484e-05,
+      "loss": 1.6337,
+      "step": 37960
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.751051851851852e-05,
+      "loss": 1.7071,
+      "step": 37965
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.750755555555556e-05,
+      "loss": 1.6709,
+      "step": 37970
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7504592592592594e-05,
+      "loss": 1.7292,
+      "step": 37975
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7501629629629633e-05,
+      "loss": 1.7846,
+      "step": 37980
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7498666666666668e-05,
+      "loss": 1.6359,
+      "step": 37985
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7495703703703707e-05,
+      "loss": 1.69,
+      "step": 37990
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 2.7492740740740742e-05,
+      "loss": 1.7307,
+      "step": 37995
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.748977777777778e-05,
+      "loss": 1.7098,
+      "step": 38000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7486814814814817e-05,
+      "loss": 1.7211,
+      "step": 38005
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7483851851851855e-05,
+      "loss": 1.625,
+      "step": 38010
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.748088888888889e-05,
+      "loss": 1.7492,
+      "step": 38015
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.747792592592593e-05,
+      "loss": 1.7915,
+      "step": 38020
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7474962962962965e-05,
+      "loss": 1.6924,
+      "step": 38025
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7472000000000004e-05,
+      "loss": 1.7519,
+      "step": 38030
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7469037037037036e-05,
+      "loss": 1.7871,
+      "step": 38035
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7466074074074078e-05,
+      "loss": 1.7404,
+      "step": 38040
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.746311111111111e-05,
+      "loss": 1.6605,
+      "step": 38045
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 2.7460148148148152e-05,
+      "loss": 1.7049,
+      "step": 38050
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7457185185185185e-05,
+      "loss": 1.5744,
+      "step": 38055
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7454222222222227e-05,
+      "loss": 1.6262,
+      "step": 38060
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.745125925925926e-05,
+      "loss": 1.7362,
+      "step": 38065
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.74482962962963e-05,
+      "loss": 1.7156,
+      "step": 38070
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7445333333333333e-05,
+      "loss": 1.7149,
+      "step": 38075
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7442370370370375e-05,
+      "loss": 1.6334,
+      "step": 38080
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7439407407407407e-05,
+      "loss": 1.6634,
+      "step": 38085
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.743644444444445e-05,
+      "loss": 1.7249,
+      "step": 38090
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.743348148148148e-05,
+      "loss": 1.7338,
+      "step": 38095
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.743051851851852e-05,
+      "loss": 1.6821,
+      "step": 38100
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 2.7427555555555556e-05,
+      "loss": 1.7063,
+      "step": 38105
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7424592592592595e-05,
+      "loss": 1.6882,
+      "step": 38110
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.742162962962963e-05,
+      "loss": 1.6067,
+      "step": 38115
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.741866666666667e-05,
+      "loss": 1.7357,
+      "step": 38120
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7415703703703704e-05,
+      "loss": 1.8009,
+      "step": 38125
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7412740740740743e-05,
+      "loss": 1.7765,
+      "step": 38130
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.740977777777778e-05,
+      "loss": 1.6655,
+      "step": 38135
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7406814814814817e-05,
+      "loss": 1.6118,
+      "step": 38140
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7403851851851853e-05,
+      "loss": 1.7657,
+      "step": 38145
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.740088888888889e-05,
+      "loss": 1.7294,
+      "step": 38150
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7397925925925927e-05,
+      "loss": 1.6436,
+      "step": 38155
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7394962962962966e-05,
+      "loss": 1.7629,
+      "step": 38160
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 2.7392e-05,
+      "loss": 1.6935,
+      "step": 38165
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.738903703703704e-05,
+      "loss": 1.6477,
+      "step": 38170
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7386074074074076e-05,
+      "loss": 1.7179,
+      "step": 38175
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7383111111111114e-05,
+      "loss": 1.7263,
+      "step": 38180
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.738014814814815e-05,
+      "loss": 1.7099,
+      "step": 38185
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.737718518518519e-05,
+      "loss": 1.7855,
+      "step": 38190
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7374222222222224e-05,
+      "loss": 1.7591,
+      "step": 38195
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7371259259259263e-05,
+      "loss": 1.707,
+      "step": 38200
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.73682962962963e-05,
+      "loss": 1.7097,
+      "step": 38205
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7365333333333337e-05,
+      "loss": 1.7096,
+      "step": 38210
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.7362370370370373e-05,
+      "loss": 1.7861,
+      "step": 38215
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 2.735940740740741e-05,
+      "loss": 1.7729,
+      "step": 38220
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7356444444444447e-05,
+      "loss": 1.709,
+      "step": 38225
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7353481481481486e-05,
+      "loss": 1.6498,
+      "step": 38230
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.735051851851852e-05,
+      "loss": 1.7757,
+      "step": 38235
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.734755555555556e-05,
+      "loss": 1.7951,
+      "step": 38240
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7344592592592595e-05,
+      "loss": 1.6588,
+      "step": 38245
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7341629629629634e-05,
+      "loss": 1.6522,
+      "step": 38250
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.733866666666667e-05,
+      "loss": 1.621,
+      "step": 38255
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.73357037037037e-05,
+      "loss": 1.6179,
+      "step": 38260
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.733274074074074e-05,
+      "loss": 1.7425,
+      "step": 38265
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7329777777777776e-05,
+      "loss": 1.8092,
+      "step": 38270
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 2.7326814814814815e-05,
+      "loss": 1.6904,
+      "step": 38275
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.732385185185185e-05,
+      "loss": 1.7383,
+      "step": 38280
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.732088888888889e-05,
+      "loss": 1.7665,
+      "step": 38285
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7317925925925924e-05,
+      "loss": 1.7098,
+      "step": 38290
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7314962962962963e-05,
+      "loss": 1.7437,
+      "step": 38295
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7312e-05,
+      "loss": 1.6721,
+      "step": 38300
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7309037037037038e-05,
+      "loss": 1.644,
+      "step": 38305
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7306074074074073e-05,
+      "loss": 1.7014,
+      "step": 38310
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7303111111111112e-05,
+      "loss": 1.7397,
+      "step": 38315
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7300148148148147e-05,
+      "loss": 1.6223,
+      "step": 38320
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.7297185185185186e-05,
+      "loss": 1.6928,
+      "step": 38325
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 2.729422222222222e-05,
+      "loss": 1.6738,
+      "step": 38330
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.729125925925926e-05,
+      "loss": 1.7409,
+      "step": 38335
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7288296296296296e-05,
+      "loss": 1.656,
+      "step": 38340
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7285333333333335e-05,
+      "loss": 1.7484,
+      "step": 38345
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.728237037037037e-05,
+      "loss": 1.8178,
+      "step": 38350
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.727940740740741e-05,
+      "loss": 1.6635,
+      "step": 38355
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7276444444444444e-05,
+      "loss": 1.7901,
+      "step": 38360
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7273481481481483e-05,
+      "loss": 1.6988,
+      "step": 38365
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.727051851851852e-05,
+      "loss": 1.8211,
+      "step": 38370
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7267555555555557e-05,
+      "loss": 1.7936,
+      "step": 38375
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7265185185185187e-05,
+      "loss": 1.7099,
+      "step": 38380
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.7262222222222222e-05,
+      "loss": 1.795,
+      "step": 38385
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 2.725925925925926e-05,
+      "loss": 1.7067,
+      "step": 38390
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7256296296296297e-05,
+      "loss": 1.7444,
+      "step": 38395
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7253333333333336e-05,
+      "loss": 1.7236,
+      "step": 38400
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.725037037037037e-05,
+      "loss": 1.7856,
+      "step": 38405
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.724740740740741e-05,
+      "loss": 1.863,
+      "step": 38410
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7244444444444445e-05,
+      "loss": 1.6582,
+      "step": 38415
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7241481481481484e-05,
+      "loss": 1.7438,
+      "step": 38420
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.723851851851852e-05,
+      "loss": 1.8004,
+      "step": 38425
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.723555555555556e-05,
+      "loss": 1.8069,
+      "step": 38430
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7232592592592594e-05,
+      "loss": 1.6842,
+      "step": 38435
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7229629629629633e-05,
+      "loss": 1.8119,
+      "step": 38440
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 2.7226666666666668e-05,
+      "loss": 1.813,
+      "step": 38445
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7223703703703707e-05,
+      "loss": 1.7099,
+      "step": 38450
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7220740740740742e-05,
+      "loss": 1.6226,
+      "step": 38455
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.721777777777778e-05,
+      "loss": 1.6923,
+      "step": 38460
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7214814814814817e-05,
+      "loss": 1.6306,
+      "step": 38465
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7211851851851855e-05,
+      "loss": 1.6701,
+      "step": 38470
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.720888888888889e-05,
+      "loss": 1.7723,
+      "step": 38475
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.720592592592593e-05,
+      "loss": 1.755,
+      "step": 38480
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7202962962962965e-05,
+      "loss": 1.6728,
+      "step": 38485
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 1.8165,
+      "step": 38490
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7197037037037036e-05,
+      "loss": 1.6725,
+      "step": 38495
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 2.7194074074074078e-05,
+      "loss": 1.7182,
+      "step": 38500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.719111111111111e-05,
+      "loss": 1.6642,
+      "step": 38505
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7188148148148152e-05,
+      "loss": 1.7075,
+      "step": 38510
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7185185185185184e-05,
+      "loss": 1.6003,
+      "step": 38515
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7182222222222227e-05,
+      "loss": 1.8028,
+      "step": 38520
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.717925925925926e-05,
+      "loss": 1.8123,
+      "step": 38525
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.71762962962963e-05,
+      "loss": 1.719,
+      "step": 38530
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7173333333333333e-05,
+      "loss": 1.6868,
+      "step": 38535
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7170370370370375e-05,
+      "loss": 1.6494,
+      "step": 38540
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.7167407407407407e-05,
+      "loss": 1.7459,
+      "step": 38545
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.716444444444445e-05,
+      "loss": 1.677,
+      "step": 38550
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 2.716148148148148e-05,
+      "loss": 1.7114,
+      "step": 38555
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7158518518518524e-05,
+      "loss": 1.749,
+      "step": 38560
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7155555555555556e-05,
+      "loss": 1.7212,
+      "step": 38565
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7152592592592595e-05,
+      "loss": 1.614,
+      "step": 38570
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.714962962962963e-05,
+      "loss": 1.7823,
+      "step": 38575
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.714666666666667e-05,
+      "loss": 1.735,
+      "step": 38580
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7143703703703704e-05,
+      "loss": 1.6902,
+      "step": 38585
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7140740740740743e-05,
+      "loss": 1.5945,
+      "step": 38590
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.713777777777778e-05,
+      "loss": 1.7352,
+      "step": 38595
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7134814814814817e-05,
+      "loss": 1.6341,
+      "step": 38600
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7131851851851853e-05,
+      "loss": 1.7438,
+      "step": 38605
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.712888888888889e-05,
+      "loss": 1.7386,
+      "step": 38610
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 2.7125925925925927e-05,
+      "loss": 1.6602,
+      "step": 38615
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7122962962962966e-05,
+      "loss": 1.7025,
+      "step": 38620
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.712e-05,
+      "loss": 1.7348,
+      "step": 38625
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.711703703703704e-05,
+      "loss": 1.6417,
+      "step": 38630
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7114074074074075e-05,
+      "loss": 1.7161,
+      "step": 38635
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7111111111111114e-05,
+      "loss": 1.637,
+      "step": 38640
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.710814814814815e-05,
+      "loss": 1.7299,
+      "step": 38645
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.710518518518519e-05,
+      "loss": 1.7832,
+      "step": 38650
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7102222222222224e-05,
+      "loss": 1.6933,
+      "step": 38655
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7099259259259263e-05,
+      "loss": 1.677,
+      "step": 38660
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7096296296296298e-05,
+      "loss": 1.7022,
+      "step": 38665
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 2.7093333333333337e-05,
+      "loss": 1.8013,
+      "step": 38670
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7090370370370373e-05,
+      "loss": 1.5793,
+      "step": 38675
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.708740740740741e-05,
+      "loss": 1.7315,
+      "step": 38680
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7084444444444447e-05,
+      "loss": 1.8213,
+      "step": 38685
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7081481481481486e-05,
+      "loss": 1.7617,
+      "step": 38690
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.707851851851852e-05,
+      "loss": 1.6321,
+      "step": 38695
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.707555555555556e-05,
+      "loss": 1.7548,
+      "step": 38700
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7072592592592595e-05,
+      "loss": 1.7607,
+      "step": 38705
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.7069629629629634e-05,
+      "loss": 1.7558,
+      "step": 38710
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.706666666666667e-05,
+      "loss": 1.7335,
+      "step": 38715
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.70637037037037e-05,
+      "loss": 1.7815,
+      "step": 38720
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 2.706074074074074e-05,
+      "loss": 1.7328,
+      "step": 38725
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7057777777777776e-05,
+      "loss": 1.8017,
+      "step": 38730
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7054814814814815e-05,
+      "loss": 1.7455,
+      "step": 38735
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.705185185185185e-05,
+      "loss": 1.7456,
+      "step": 38740
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.704888888888889e-05,
+      "loss": 1.5899,
+      "step": 38745
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7045925925925924e-05,
+      "loss": 1.737,
+      "step": 38750
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7042962962962963e-05,
+      "loss": 1.5926,
+      "step": 38755
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.704e-05,
+      "loss": 1.7656,
+      "step": 38760
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7037037037037037e-05,
+      "loss": 1.6769,
+      "step": 38765
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7034074074074073e-05,
+      "loss": 1.7178,
+      "step": 38770
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.703111111111111e-05,
+      "loss": 1.7439,
+      "step": 38775
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 2.7028148148148147e-05,
+      "loss": 1.743,
+      "step": 38780
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7025185185185186e-05,
+      "loss": 1.8063,
+      "step": 38785
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.702222222222222e-05,
+      "loss": 1.6693,
+      "step": 38790
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.701925925925926e-05,
+      "loss": 1.7541,
+      "step": 38795
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7016296296296296e-05,
+      "loss": 1.5864,
+      "step": 38800
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7013333333333334e-05,
+      "loss": 1.749,
+      "step": 38805
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.701037037037037e-05,
+      "loss": 1.7416,
+      "step": 38810
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.700740740740741e-05,
+      "loss": 1.6326,
+      "step": 38815
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7004444444444444e-05,
+      "loss": 1.6893,
+      "step": 38820
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.7001481481481483e-05,
+      "loss": 1.5128,
+      "step": 38825
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.699851851851852e-05,
+      "loss": 1.6727,
+      "step": 38830
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.6995555555555557e-05,
+      "loss": 1.5707,
+      "step": 38835
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 2.6992592592592593e-05,
+      "loss": 1.737,
+      "step": 38840
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.698962962962963e-05,
+      "loss": 1.5429,
+      "step": 38845
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6986666666666667e-05,
+      "loss": 1.6957,
+      "step": 38850
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6983703703703706e-05,
+      "loss": 1.8308,
+      "step": 38855
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.698074074074074e-05,
+      "loss": 1.7329,
+      "step": 38860
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.697777777777778e-05,
+      "loss": 1.6496,
+      "step": 38865
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6974814814814815e-05,
+      "loss": 1.831,
+      "step": 38870
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6971851851851854e-05,
+      "loss": 1.6553,
+      "step": 38875
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6968888888888886e-05,
+      "loss": 1.4619,
+      "step": 38880
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.696592592592593e-05,
+      "loss": 1.7402,
+      "step": 38885
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.696296296296296e-05,
+      "loss": 1.6119,
+      "step": 38890
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 2.6960000000000003e-05,
+      "loss": 1.7387,
+      "step": 38895
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6957037037037035e-05,
+      "loss": 1.9125,
+      "step": 38900
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6954074074074077e-05,
+      "loss": 1.6853,
+      "step": 38905
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.695111111111111e-05,
+      "loss": 1.7305,
+      "step": 38910
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.694814814814815e-05,
+      "loss": 1.7969,
+      "step": 38915
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6945185185185183e-05,
+      "loss": 1.5919,
+      "step": 38920
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6942222222222226e-05,
+      "loss": 1.8155,
+      "step": 38925
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6939259259259258e-05,
+      "loss": 1.7932,
+      "step": 38930
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.69362962962963e-05,
+      "loss": 1.7487,
+      "step": 38935
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6933333333333332e-05,
+      "loss": 1.6724,
+      "step": 38940
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6930370370370374e-05,
+      "loss": 1.7682,
+      "step": 38945
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 2.6927407407407406e-05,
+      "loss": 1.7245,
+      "step": 38950
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6924444444444445e-05,
+      "loss": 1.6018,
+      "step": 38955
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.692148148148148e-05,
+      "loss": 1.6359,
+      "step": 38960
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.691851851851852e-05,
+      "loss": 1.8212,
+      "step": 38965
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6915555555555555e-05,
+      "loss": 1.7602,
+      "step": 38970
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6912592592592593e-05,
+      "loss": 1.6906,
+      "step": 38975
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.690962962962963e-05,
+      "loss": 1.651,
+      "step": 38980
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6906666666666668e-05,
+      "loss": 1.8139,
+      "step": 38985
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6903703703703703e-05,
+      "loss": 1.7974,
+      "step": 38990
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6900740740740742e-05,
+      "loss": 1.7303,
+      "step": 38995
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6897777777777777e-05,
+      "loss": 1.58,
+      "step": 39000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 2.6894814814814816e-05,
+      "loss": 1.7261,
+      "step": 39005
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.689185185185185e-05,
+      "loss": 1.6666,
+      "step": 39010
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.688888888888889e-05,
+      "loss": 1.6752,
+      "step": 39015
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6885925925925926e-05,
+      "loss": 1.7705,
+      "step": 39020
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6882962962962965e-05,
+      "loss": 1.5543,
+      "step": 39025
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.688e-05,
+      "loss": 1.8508,
+      "step": 39030
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.687703703703704e-05,
+      "loss": 1.6733,
+      "step": 39035
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6874074074074074e-05,
+      "loss": 1.7489,
+      "step": 39040
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6871111111111113e-05,
+      "loss": 1.7324,
+      "step": 39045
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.686814814814815e-05,
+      "loss": 1.704,
+      "step": 39050
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6865185185185187e-05,
+      "loss": 1.7032,
+      "step": 39055
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6862222222222223e-05,
+      "loss": 1.732,
+      "step": 39060
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 2.6859259259259262e-05,
+      "loss": 1.6605,
+      "step": 39065
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6856296296296297e-05,
+      "loss": 1.674,
+      "step": 39070
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6853333333333336e-05,
+      "loss": 1.7638,
+      "step": 39075
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.685037037037037e-05,
+      "loss": 1.5656,
+      "step": 39080
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.684740740740741e-05,
+      "loss": 1.6959,
+      "step": 39085
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6844444444444446e-05,
+      "loss": 1.6939,
+      "step": 39090
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6841481481481484e-05,
+      "loss": 1.7364,
+      "step": 39095
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.683851851851852e-05,
+      "loss": 1.6231,
+      "step": 39100
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.683555555555556e-05,
+      "loss": 1.8087,
+      "step": 39105
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6832592592592594e-05,
+      "loss": 1.6183,
+      "step": 39110
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6829629629629633e-05,
+      "loss": 1.6509,
+      "step": 39115
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 2.6826666666666665e-05,
+      "loss": 1.6792,
+      "step": 39120
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6823703703703707e-05,
+      "loss": 1.8754,
+      "step": 39125
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.682074074074074e-05,
+      "loss": 1.6914,
+      "step": 39130
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.681777777777778e-05,
+      "loss": 1.7282,
+      "step": 39135
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6814814814814814e-05,
+      "loss": 1.7117,
+      "step": 39140
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6811851851851856e-05,
+      "loss": 1.7073,
+      "step": 39145
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6808888888888888e-05,
+      "loss": 1.8195,
+      "step": 39150
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.680592592592593e-05,
+      "loss": 1.6012,
+      "step": 39155
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6802962962962962e-05,
+      "loss": 1.6131,
+      "step": 39160
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 1.7422,
+      "step": 39165
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.6797037037037036e-05,
+      "loss": 1.8097,
+      "step": 39170
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.679407407407408e-05,
+      "loss": 1.7416,
+      "step": 39175
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.679111111111111e-05,
+      "loss": 1.6002,
+      "step": 39180
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.678814814814815e-05,
+      "loss": 1.8867,
+      "step": 39185
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6785185185185185e-05,
+      "loss": 1.7756,
+      "step": 39190
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6782222222222224e-05,
+      "loss": 1.8117,
+      "step": 39195
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.677925925925926e-05,
+      "loss": 1.8052,
+      "step": 39200
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6776296296296298e-05,
+      "loss": 1.7525,
+      "step": 39205
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6773333333333333e-05,
+      "loss": 1.6986,
+      "step": 39210
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6770370370370372e-05,
+      "loss": 1.6348,
+      "step": 39215
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6767407407407408e-05,
+      "loss": 1.7393,
+      "step": 39220
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6764444444444446e-05,
+      "loss": 1.5454,
+      "step": 39225
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.6761481481481482e-05,
+      "loss": 1.7708,
+      "step": 39230
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.675851851851852e-05,
+      "loss": 1.8216,
+      "step": 39235
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6755555555555556e-05,
+      "loss": 1.7061,
+      "step": 39240
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6752592592592595e-05,
+      "loss": 1.6456,
+      "step": 39245
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.674962962962963e-05,
+      "loss": 1.7191,
+      "step": 39250
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.674666666666667e-05,
+      "loss": 1.8653,
+      "step": 39255
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6743703703703705e-05,
+      "loss": 1.7782,
+      "step": 39260
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6740740740740743e-05,
+      "loss": 1.6248,
+      "step": 39265
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.673777777777778e-05,
+      "loss": 1.8535,
+      "step": 39270
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6734814814814818e-05,
+      "loss": 1.6592,
+      "step": 39275
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6731851851851853e-05,
+      "loss": 1.7687,
+      "step": 39280
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6728888888888892e-05,
+      "loss": 1.76,
+      "step": 39285
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 2.6725925925925927e-05,
+      "loss": 1.5495,
+      "step": 39290
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6722962962962966e-05,
+      "loss": 1.7736,
+      "step": 39295
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.672e-05,
+      "loss": 1.7244,
+      "step": 39300
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.671703703703704e-05,
+      "loss": 1.8789,
+      "step": 39305
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6714074074074076e-05,
+      "loss": 1.7802,
+      "step": 39310
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6711111111111115e-05,
+      "loss": 1.6449,
+      "step": 39315
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.670814814814815e-05,
+      "loss": 1.76,
+      "step": 39320
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.670518518518519e-05,
+      "loss": 1.8295,
+      "step": 39325
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6702222222222224e-05,
+      "loss": 1.6972,
+      "step": 39330
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6699259259259263e-05,
+      "loss": 1.7639,
+      "step": 39335
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.66962962962963e-05,
+      "loss": 1.704,
+      "step": 39340
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 2.6693333333333338e-05,
+      "loss": 1.6268,
+      "step": 39345
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.669037037037037e-05,
+      "loss": 1.6957,
+      "step": 39350
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6687407407407412e-05,
+      "loss": 1.6491,
+      "step": 39355
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6684444444444444e-05,
+      "loss": 1.7614,
+      "step": 39360
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6681481481481486e-05,
+      "loss": 1.7203,
+      "step": 39365
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6678518518518518e-05,
+      "loss": 1.7488,
+      "step": 39370
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.667555555555556e-05,
+      "loss": 1.7527,
+      "step": 39375
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6672592592592592e-05,
+      "loss": 1.5614,
+      "step": 39380
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6669629629629635e-05,
+      "loss": 1.6325,
+      "step": 39385
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 1.5437,
+      "step": 39390
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.666370370370371e-05,
+      "loss": 1.6232,
+      "step": 39395
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 2.666074074074074e-05,
+      "loss": 1.662,
+      "step": 39400
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6657777777777783e-05,
+      "loss": 1.6137,
+      "step": 39405
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6654814814814815e-05,
+      "loss": 1.544,
+      "step": 39410
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6651851851851857e-05,
+      "loss": 1.5205,
+      "step": 39415
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.664888888888889e-05,
+      "loss": 1.5082,
+      "step": 39420
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6645925925925928e-05,
+      "loss": 1.5779,
+      "step": 39425
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6642962962962964e-05,
+      "loss": 1.6413,
+      "step": 39430
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 1.7245,
+      "step": 39435
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6637037037037038e-05,
+      "loss": 1.6349,
+      "step": 39440
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6634074074074077e-05,
+      "loss": 1.6914,
+      "step": 39445
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.6631111111111112e-05,
+      "loss": 1.6252,
+      "step": 39450
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 2.662814814814815e-05,
+      "loss": 1.6854,
+      "step": 39455
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6625185185185186e-05,
+      "loss": 1.6044,
+      "step": 39460
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6622222222222225e-05,
+      "loss": 1.6407,
+      "step": 39465
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.661925925925926e-05,
+      "loss": 1.4787,
+      "step": 39470
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.66162962962963e-05,
+      "loss": 1.6658,
+      "step": 39475
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6613333333333335e-05,
+      "loss": 1.6632,
+      "step": 39480
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6610370370370374e-05,
+      "loss": 1.5911,
+      "step": 39485
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.660740740740741e-05,
+      "loss": 1.5671,
+      "step": 39490
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6604444444444445e-05,
+      "loss": 1.7011,
+      "step": 39495
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6601481481481483e-05,
+      "loss": 1.6292,
+      "step": 39500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6598518518518515e-05,
+      "loss": 1.6105,
+      "step": 39505
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.6595555555555558e-05,
+      "loss": 1.6266,
+      "step": 39510
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 2.659259259259259e-05,
+      "loss": 1.5613,
+      "step": 39515
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6589629629629632e-05,
+      "loss": 1.636,
+      "step": 39520
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6586666666666664e-05,
+      "loss": 1.5723,
+      "step": 39525
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6583703703703706e-05,
+      "loss": 1.7308,
+      "step": 39530
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6580740740740738e-05,
+      "loss": 1.6096,
+      "step": 39535
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.657777777777778e-05,
+      "loss": 1.6317,
+      "step": 39540
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6574814814814812e-05,
+      "loss": 1.4433,
+      "step": 39545
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6571851851851855e-05,
+      "loss": 1.6213,
+      "step": 39550
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6568888888888887e-05,
+      "loss": 1.6322,
+      "step": 39555
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.656592592592593e-05,
+      "loss": 1.5699,
+      "step": 39560
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.656296296296296e-05,
+      "loss": 1.5164,
+      "step": 39565
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 1.58,
+      "step": 39570
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6557037037037035e-05,
+      "loss": 1.6357,
+      "step": 39575
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6554074074074074e-05,
+      "loss": 1.6281,
+      "step": 39580
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.655111111111111e-05,
+      "loss": 1.6633,
+      "step": 39585
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.654814814814815e-05,
+      "loss": 1.7541,
+      "step": 39590
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6545185185185184e-05,
+      "loss": 1.7321,
+      "step": 39595
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6542222222222223e-05,
+      "loss": 1.6777,
+      "step": 39600
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6539259259259258e-05,
+      "loss": 1.5609,
+      "step": 39605
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6536296296296297e-05,
+      "loss": 1.5672,
+      "step": 39610
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6533333333333332e-05,
+      "loss": 1.574,
+      "step": 39615
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.653037037037037e-05,
+      "loss": 1.5763,
+      "step": 39620
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 2.6527407407407407e-05,
+      "loss": 1.63,
+      "step": 39625
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6524444444444445e-05,
+      "loss": 1.6669,
+      "step": 39630
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.652148148148148e-05,
+      "loss": 1.6175,
+      "step": 39635
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.651851851851852e-05,
+      "loss": 1.6702,
+      "step": 39640
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6515555555555555e-05,
+      "loss": 1.5274,
+      "step": 39645
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6512592592592594e-05,
+      "loss": 1.6793,
+      "step": 39650
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.650962962962963e-05,
+      "loss": 1.6243,
+      "step": 39655
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6506666666666668e-05,
+      "loss": 1.5327,
+      "step": 39660
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6503703703703704e-05,
+      "loss": 1.6488,
+      "step": 39665
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6500740740740742e-05,
+      "loss": 1.549,
+      "step": 39670
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6497777777777778e-05,
+      "loss": 1.6504,
+      "step": 39675
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 2.6494814814814817e-05,
+      "loss": 1.4957,
+      "step": 39680
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6491851851851852e-05,
+      "loss": 1.5378,
+      "step": 39685
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.648888888888889e-05,
+      "loss": 1.625,
+      "step": 39690
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6485925925925926e-05,
+      "loss": 1.7486,
+      "step": 39695
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6482962962962965e-05,
+      "loss": 1.5924,
+      "step": 39700
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.648e-05,
+      "loss": 1.5741,
+      "step": 39705
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.647703703703704e-05,
+      "loss": 1.5325,
+      "step": 39710
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6474074074074075e-05,
+      "loss": 1.6584,
+      "step": 39715
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6471111111111114e-05,
+      "loss": 1.5532,
+      "step": 39720
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.646814814814815e-05,
+      "loss": 1.6679,
+      "step": 39725
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6465185185185188e-05,
+      "loss": 1.5596,
+      "step": 39730
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.646222222222222e-05,
+      "loss": 1.6577,
+      "step": 39735
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 2.6459259259259262e-05,
+      "loss": 1.5754,
+      "step": 39740
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6456296296296294e-05,
+      "loss": 1.4986,
+      "step": 39745
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6453333333333336e-05,
+      "loss": 1.6387,
+      "step": 39750
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.645037037037037e-05,
+      "loss": 1.6718,
+      "step": 39755
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.644740740740741e-05,
+      "loss": 1.5717,
+      "step": 39760
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6444444444444443e-05,
+      "loss": 1.5841,
+      "step": 39765
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6441481481481485e-05,
+      "loss": 1.7476,
+      "step": 39770
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6438518518518517e-05,
+      "loss": 1.545,
+      "step": 39775
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.643555555555556e-05,
+      "loss": 1.6577,
+      "step": 39780
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.643259259259259e-05,
+      "loss": 1.6188,
+      "step": 39785
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6429629629629633e-05,
+      "loss": 1.6151,
+      "step": 39790
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 2.6426666666666665e-05,
+      "loss": 1.5332,
+      "step": 39795
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6423703703703708e-05,
+      "loss": 1.5699,
+      "step": 39800
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.642074074074074e-05,
+      "loss": 1.6497,
+      "step": 39805
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.641777777777778e-05,
+      "loss": 1.582,
+      "step": 39810
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6414814814814814e-05,
+      "loss": 1.6528,
+      "step": 39815
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6411851851851853e-05,
+      "loss": 1.6658,
+      "step": 39820
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6408888888888888e-05,
+      "loss": 1.6631,
+      "step": 39825
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6405925925925927e-05,
+      "loss": 1.6084,
+      "step": 39830
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6402962962962963e-05,
+      "loss": 1.6147,
+      "step": 39835
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.64e-05,
+      "loss": 1.555,
+      "step": 39840
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6397037037037037e-05,
+      "loss": 1.5685,
+      "step": 39845
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 2.6394074074074076e-05,
+      "loss": 1.6103,
+      "step": 39850
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.639111111111111e-05,
+      "loss": 1.5714,
+      "step": 39855
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.638814814814815e-05,
+      "loss": 1.725,
+      "step": 39860
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6385185185185185e-05,
+      "loss": 1.5902,
+      "step": 39865
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6382222222222224e-05,
+      "loss": 1.6609,
+      "step": 39870
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.637925925925926e-05,
+      "loss": 1.6478,
+      "step": 39875
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.63762962962963e-05,
+      "loss": 1.6393,
+      "step": 39880
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6373333333333334e-05,
+      "loss": 1.5522,
+      "step": 39885
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6370370370370373e-05,
+      "loss": 1.471,
+      "step": 39890
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6367407407407408e-05,
+      "loss": 1.6275,
+      "step": 39895
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6364444444444447e-05,
+      "loss": 1.6529,
+      "step": 39900
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 2.6361481481481482e-05,
+      "loss": 1.6992,
+      "step": 39905
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.635851851851852e-05,
+      "loss": 1.6153,
+      "step": 39910
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6355555555555557e-05,
+      "loss": 1.5969,
+      "step": 39915
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6352592592592595e-05,
+      "loss": 1.6598,
+      "step": 39920
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.634962962962963e-05,
+      "loss": 1.5874,
+      "step": 39925
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.634666666666667e-05,
+      "loss": 1.5792,
+      "step": 39930
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6343703703703705e-05,
+      "loss": 1.5977,
+      "step": 39935
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6340740740740744e-05,
+      "loss": 1.6166,
+      "step": 39940
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.633777777777778e-05,
+      "loss": 1.6723,
+      "step": 39945
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6334814814814818e-05,
+      "loss": 1.72,
+      "step": 39950
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6331851851851854e-05,
+      "loss": 1.6055,
+      "step": 39955
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6328888888888892e-05,
+      "loss": 1.4739,
+      "step": 39960
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 2.6325925925925924e-05,
+      "loss": 1.6605,
+      "step": 39965
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6322962962962967e-05,
+      "loss": 1.5691,
+      "step": 39970
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.632e-05,
+      "loss": 1.6001,
+      "step": 39975
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.631703703703704e-05,
+      "loss": 1.5721,
+      "step": 39980
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6314074074074073e-05,
+      "loss": 1.6088,
+      "step": 39985
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6311111111111115e-05,
+      "loss": 1.7683,
+      "step": 39990
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6308148148148147e-05,
+      "loss": 1.7085,
+      "step": 39995
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.630518518518519e-05,
+      "loss": 1.7289,
+      "step": 40000
+    },
+    {
+      "epoch": 7.11,
+      "eval_loss": 1.5409154891967773,
+      "eval_rouge2_fmeasure": 0.195,
+      "eval_rouge2_precision": 0.2256,
+      "eval_rouge2_recall": 0.1809,
+      "eval_runtime": 38351.4186,
+      "eval_samples_per_second": 0.13,
+      "eval_steps_per_second": 0.065,
+      "step": 40000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.630222222222222e-05,
+      "loss": 1.6161,
+      "step": 40005
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6299259259259264e-05,
+      "loss": 1.5507,
+      "step": 40010
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6296296296296296e-05,
+      "loss": 1.7311,
+      "step": 40015
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 2.6293333333333338e-05,
+      "loss": 1.5518,
+      "step": 40020
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.629037037037037e-05,
+      "loss": 1.507,
+      "step": 40025
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6287407407407412e-05,
+      "loss": 1.5933,
+      "step": 40030
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6284444444444444e-05,
+      "loss": 1.5868,
+      "step": 40035
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6281481481481483e-05,
+      "loss": 1.5985,
+      "step": 40040
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.627851851851852e-05,
+      "loss": 1.6734,
+      "step": 40045
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6275555555555557e-05,
+      "loss": 1.494,
+      "step": 40050
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6272592592592593e-05,
+      "loss": 1.573,
+      "step": 40055
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.626962962962963e-05,
+      "loss": 1.5441,
+      "step": 40060
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6266666666666667e-05,
+      "loss": 1.6247,
+      "step": 40065
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.6263703703703706e-05,
+      "loss": 1.6404,
+      "step": 40070
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 2.626074074074074e-05,
+      "loss": 1.7445,
+      "step": 40075
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.625777777777778e-05,
+      "loss": 1.5704,
+      "step": 40080
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6254814814814816e-05,
+      "loss": 1.6002,
+      "step": 40085
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6251851851851854e-05,
+      "loss": 1.6565,
+      "step": 40090
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.624888888888889e-05,
+      "loss": 1.6243,
+      "step": 40095
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.624592592592593e-05,
+      "loss": 1.6308,
+      "step": 40100
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6242962962962964e-05,
+      "loss": 1.7161,
+      "step": 40105
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 1.5762,
+      "step": 40110
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.623703703703704e-05,
+      "loss": 1.6354,
+      "step": 40115
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6234074074074077e-05,
+      "loss": 1.7506,
+      "step": 40120
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.6231111111111113e-05,
+      "loss": 1.5823,
+      "step": 40125
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 2.622814814814815e-05,
+      "loss": 1.5956,
+      "step": 40130
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6225185185185187e-05,
+      "loss": 1.5283,
+      "step": 40135
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6222222222222226e-05,
+      "loss": 1.7062,
+      "step": 40140
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.621925925925926e-05,
+      "loss": 1.5892,
+      "step": 40145
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.62162962962963e-05,
+      "loss": 1.5382,
+      "step": 40150
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6213333333333335e-05,
+      "loss": 1.5796,
+      "step": 40155
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6210370370370374e-05,
+      "loss": 1.6191,
+      "step": 40160
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.620740740740741e-05,
+      "loss": 1.5465,
+      "step": 40165
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.620444444444445e-05,
+      "loss": 1.5481,
+      "step": 40170
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6201481481481484e-05,
+      "loss": 1.5265,
+      "step": 40175
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6198518518518523e-05,
+      "loss": 1.6412,
+      "step": 40180
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6195555555555558e-05,
+      "loss": 1.5585,
+      "step": 40185
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 2.6192592592592597e-05,
+      "loss": 1.5731,
+      "step": 40190
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.618962962962963e-05,
+      "loss": 1.6809,
+      "step": 40195
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.618666666666667e-05,
+      "loss": 1.5661,
+      "step": 40200
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6183703703703703e-05,
+      "loss": 1.6546,
+      "step": 40205
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6180740740740745e-05,
+      "loss": 1.579,
+      "step": 40210
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6177777777777777e-05,
+      "loss": 1.6666,
+      "step": 40215
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.617481481481482e-05,
+      "loss": 1.5772,
+      "step": 40220
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6171851851851852e-05,
+      "loss": 1.7114,
+      "step": 40225
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6168888888888894e-05,
+      "loss": 1.5842,
+      "step": 40230
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6165925925925926e-05,
+      "loss": 1.5407,
+      "step": 40235
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.6162962962962968e-05,
+      "loss": 1.778,
+      "step": 40240
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 2.616e-05,
+      "loss": 1.5054,
+      "step": 40245
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6157037037037042e-05,
+      "loss": 1.6951,
+      "step": 40250
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6154074074074075e-05,
+      "loss": 1.6027,
+      "step": 40255
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6151111111111117e-05,
+      "loss": 1.6308,
+      "step": 40260
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.614814814814815e-05,
+      "loss": 1.7071,
+      "step": 40265
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6145185185185184e-05,
+      "loss": 1.5611,
+      "step": 40270
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6142222222222223e-05,
+      "loss": 1.5572,
+      "step": 40275
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.613925925925926e-05,
+      "loss": 1.6683,
+      "step": 40280
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6136296296296297e-05,
+      "loss": 1.5243,
+      "step": 40285
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6133333333333333e-05,
+      "loss": 1.6808,
+      "step": 40290
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.613037037037037e-05,
+      "loss": 1.7563,
+      "step": 40295
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 2.6127407407407407e-05,
+      "loss": 1.5166,
+      "step": 40300
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6124444444444446e-05,
+      "loss": 1.5997,
+      "step": 40305
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.612148148148148e-05,
+      "loss": 1.541,
+      "step": 40310
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.611851851851852e-05,
+      "loss": 1.5923,
+      "step": 40315
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6115555555555555e-05,
+      "loss": 1.6752,
+      "step": 40320
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6112592592592594e-05,
+      "loss": 1.5918,
+      "step": 40325
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.610962962962963e-05,
+      "loss": 1.5369,
+      "step": 40330
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.610666666666667e-05,
+      "loss": 1.7484,
+      "step": 40335
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6103703703703704e-05,
+      "loss": 1.6152,
+      "step": 40340
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6100740740740743e-05,
+      "loss": 1.5989,
+      "step": 40345
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6097777777777775e-05,
+      "loss": 1.5734,
+      "step": 40350
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 2.6094814814814817e-05,
+      "loss": 1.6829,
+      "step": 40355
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.609185185185185e-05,
+      "loss": 1.6296,
+      "step": 40360
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.608888888888889e-05,
+      "loss": 1.6255,
+      "step": 40365
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6085925925925923e-05,
+      "loss": 1.8139,
+      "step": 40370
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6082962962962966e-05,
+      "loss": 1.5585,
+      "step": 40375
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6079999999999998e-05,
+      "loss": 1.5572,
+      "step": 40380
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.607703703703704e-05,
+      "loss": 1.553,
+      "step": 40385
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6074074074074072e-05,
+      "loss": 1.5755,
+      "step": 40390
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6071111111111114e-05,
+      "loss": 1.6307,
+      "step": 40395
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6068148148148146e-05,
+      "loss": 1.5193,
+      "step": 40400
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.606518518518519e-05,
+      "loss": 1.6372,
+      "step": 40405
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.606222222222222e-05,
+      "loss": 1.5971,
+      "step": 40410
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 2.6059259259259263e-05,
+      "loss": 1.7176,
+      "step": 40415
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6056296296296295e-05,
+      "loss": 1.564,
+      "step": 40420
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6053333333333333e-05,
+      "loss": 1.6633,
+      "step": 40425
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6050962962962967e-05,
+      "loss": 1.6445,
+      "step": 40430
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6048e-05,
+      "loss": 1.6395,
+      "step": 40435
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.604503703703704e-05,
+      "loss": 1.4316,
+      "step": 40440
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6042074074074073e-05,
+      "loss": 1.6773,
+      "step": 40445
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6039111111111115e-05,
+      "loss": 1.7636,
+      "step": 40450
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6036148148148147e-05,
+      "loss": 1.6464,
+      "step": 40455
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.603318518518519e-05,
+      "loss": 1.6541,
+      "step": 40460
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.603022222222222e-05,
+      "loss": 1.6546,
+      "step": 40465
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 2.6027259259259264e-05,
+      "loss": 1.6637,
+      "step": 40470
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6024296296296296e-05,
+      "loss": 1.6088,
+      "step": 40475
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6021333333333338e-05,
+      "loss": 1.5192,
+      "step": 40480
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.601837037037037e-05,
+      "loss": 1.6664,
+      "step": 40485
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6016000000000003e-05,
+      "loss": 1.5942,
+      "step": 40490
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6013037037037035e-05,
+      "loss": 1.5652,
+      "step": 40495
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6010074074074077e-05,
+      "loss": 1.5669,
+      "step": 40500
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.600711111111111e-05,
+      "loss": 1.6862,
+      "step": 40505
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6004148148148148e-05,
+      "loss": 1.6841,
+      "step": 40510
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.6001185185185184e-05,
+      "loss": 1.6149,
+      "step": 40515
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.5998222222222222e-05,
+      "loss": 1.7018,
+      "step": 40520
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 2.5995259259259258e-05,
+      "loss": 1.6876,
+      "step": 40525
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5992296296296297e-05,
+      "loss": 1.717,
+      "step": 40530
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5989333333333332e-05,
+      "loss": 1.6344,
+      "step": 40535
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.598637037037037e-05,
+      "loss": 1.7217,
+      "step": 40540
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5983407407407406e-05,
+      "loss": 1.6277,
+      "step": 40545
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5980444444444445e-05,
+      "loss": 1.6149,
+      "step": 40550
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.597748148148148e-05,
+      "loss": 1.6715,
+      "step": 40555
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.597451851851852e-05,
+      "loss": 1.6932,
+      "step": 40560
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5971555555555555e-05,
+      "loss": 1.6322,
+      "step": 40565
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5968592592592594e-05,
+      "loss": 1.5955,
+      "step": 40570
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.596562962962963e-05,
+      "loss": 1.6561,
+      "step": 40575
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 2.5962666666666668e-05,
+      "loss": 1.5022,
+      "step": 40580
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5959703703703703e-05,
+      "loss": 1.5576,
+      "step": 40585
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5956740740740742e-05,
+      "loss": 1.6379,
+      "step": 40590
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5953777777777778e-05,
+      "loss": 1.7231,
+      "step": 40595
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5950814814814816e-05,
+      "loss": 1.4801,
+      "step": 40600
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5947851851851852e-05,
+      "loss": 1.601,
+      "step": 40605
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.594488888888889e-05,
+      "loss": 1.6112,
+      "step": 40610
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5941925925925926e-05,
+      "loss": 1.6433,
+      "step": 40615
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5938962962962965e-05,
+      "loss": 1.6634,
+      "step": 40620
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5936e-05,
+      "loss": 1.639,
+      "step": 40625
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.593303703703704e-05,
+      "loss": 1.5013,
+      "step": 40630
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5930074074074075e-05,
+      "loss": 1.6138,
+      "step": 40635
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 2.5927111111111113e-05,
+      "loss": 1.7167,
+      "step": 40640
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.592414814814815e-05,
+      "loss": 1.7774,
+      "step": 40645
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5921185185185188e-05,
+      "loss": 1.5659,
+      "step": 40650
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5918222222222223e-05,
+      "loss": 1.7521,
+      "step": 40655
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5915259259259262e-05,
+      "loss": 1.6002,
+      "step": 40660
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5912296296296294e-05,
+      "loss": 1.7577,
+      "step": 40665
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5909333333333336e-05,
+      "loss": 1.5114,
+      "step": 40670
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5906370370370368e-05,
+      "loss": 1.6489,
+      "step": 40675
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.590340740740741e-05,
+      "loss": 1.6646,
+      "step": 40680
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5900444444444443e-05,
+      "loss": 1.6403,
+      "step": 40685
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5897481481481485e-05,
+      "loss": 1.5136,
+      "step": 40690
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 2.5894518518518517e-05,
+      "loss": 1.6413,
+      "step": 40695
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.589155555555556e-05,
+      "loss": 1.5207,
+      "step": 40700
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.588859259259259e-05,
+      "loss": 1.6235,
+      "step": 40705
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5885629629629633e-05,
+      "loss": 1.6365,
+      "step": 40710
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5882666666666665e-05,
+      "loss": 1.536,
+      "step": 40715
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5879703703703707e-05,
+      "loss": 1.6377,
+      "step": 40720
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.587674074074074e-05,
+      "loss": 1.5948,
+      "step": 40725
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5873777777777782e-05,
+      "loss": 1.6323,
+      "step": 40730
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5870814814814814e-05,
+      "loss": 1.6256,
+      "step": 40735
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5867851851851853e-05,
+      "loss": 1.5577,
+      "step": 40740
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5864888888888888e-05,
+      "loss": 1.7172,
+      "step": 40745
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 2.5861925925925927e-05,
+      "loss": 1.5852,
+      "step": 40750
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5858962962962962e-05,
+      "loss": 1.5515,
+      "step": 40755
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5856e-05,
+      "loss": 1.6314,
+      "step": 40760
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5853037037037037e-05,
+      "loss": 1.5578,
+      "step": 40765
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5850074074074075e-05,
+      "loss": 1.5651,
+      "step": 40770
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.584711111111111e-05,
+      "loss": 1.6884,
+      "step": 40775
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.584414814814815e-05,
+      "loss": 1.6532,
+      "step": 40780
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5841185185185185e-05,
+      "loss": 1.6346,
+      "step": 40785
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5838222222222224e-05,
+      "loss": 1.6402,
+      "step": 40790
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.583525925925926e-05,
+      "loss": 1.7316,
+      "step": 40795
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5832296296296298e-05,
+      "loss": 1.6149,
+      "step": 40800
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 2.5829333333333334e-05,
+      "loss": 1.6287,
+      "step": 40805
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5826370370370372e-05,
+      "loss": 1.651,
+      "step": 40810
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5823407407407408e-05,
+      "loss": 1.5266,
+      "step": 40815
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5820444444444447e-05,
+      "loss": 1.5881,
+      "step": 40820
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5817481481481482e-05,
+      "loss": 1.6744,
+      "step": 40825
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.581451851851852e-05,
+      "loss": 1.6872,
+      "step": 40830
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5811555555555556e-05,
+      "loss": 1.5689,
+      "step": 40835
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5808592592592595e-05,
+      "loss": 1.5866,
+      "step": 40840
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.580562962962963e-05,
+      "loss": 1.6319,
+      "step": 40845
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.580266666666667e-05,
+      "loss": 1.6249,
+      "step": 40850
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5799703703703705e-05,
+      "loss": 1.5755,
+      "step": 40855
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.5796740740740744e-05,
+      "loss": 1.5097,
+      "step": 40860
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 2.579377777777778e-05,
+      "loss": 1.6023,
+      "step": 40865
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5790814814814818e-05,
+      "loss": 1.6305,
+      "step": 40870
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5787851851851853e-05,
+      "loss": 1.6617,
+      "step": 40875
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5784888888888892e-05,
+      "loss": 1.5318,
+      "step": 40880
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5781925925925928e-05,
+      "loss": 1.703,
+      "step": 40885
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5778962962962966e-05,
+      "loss": 1.5126,
+      "step": 40890
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5776e-05,
+      "loss": 1.7034,
+      "step": 40895
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.577303703703704e-05,
+      "loss": 1.6794,
+      "step": 40900
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5770074074074073e-05,
+      "loss": 1.709,
+      "step": 40905
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5767111111111115e-05,
+      "loss": 1.5878,
+      "step": 40910
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.5764148148148147e-05,
+      "loss": 1.6028,
+      "step": 40915
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 2.576118518518519e-05,
+      "loss": 1.5957,
+      "step": 40920
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.575822222222222e-05,
+      "loss": 1.6464,
+      "step": 40925
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5755259259259263e-05,
+      "loss": 1.6435,
+      "step": 40930
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5752296296296296e-05,
+      "loss": 1.5976,
+      "step": 40935
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5749333333333338e-05,
+      "loss": 1.633,
+      "step": 40940
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.574637037037037e-05,
+      "loss": 1.5551,
+      "step": 40945
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5743407407407412e-05,
+      "loss": 1.5906,
+      "step": 40950
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5740444444444444e-05,
+      "loss": 1.6156,
+      "step": 40955
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5737481481481486e-05,
+      "loss": 1.7116,
+      "step": 40960
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5734518518518518e-05,
+      "loss": 1.686,
+      "step": 40965
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5731555555555557e-05,
+      "loss": 1.7061,
+      "step": 40970
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 2.5728592592592593e-05,
+      "loss": 1.6666,
+      "step": 40975
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.572562962962963e-05,
+      "loss": 1.7211,
+      "step": 40980
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5722666666666667e-05,
+      "loss": 1.6433,
+      "step": 40985
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5719703703703706e-05,
+      "loss": 1.6192,
+      "step": 40990
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.571674074074074e-05,
+      "loss": 1.667,
+      "step": 40995
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.571377777777778e-05,
+      "loss": 1.6456,
+      "step": 41000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5710814814814815e-05,
+      "loss": 1.5619,
+      "step": 41005
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5707851851851854e-05,
+      "loss": 1.5684,
+      "step": 41010
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.570488888888889e-05,
+      "loss": 1.5446,
+      "step": 41015
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.570192592592593e-05,
+      "loss": 1.5835,
+      "step": 41020
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5698962962962964e-05,
+      "loss": 1.5784,
+      "step": 41025
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 2.5696000000000003e-05,
+      "loss": 1.6169,
+      "step": 41030
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5693037037037038e-05,
+      "loss": 1.6434,
+      "step": 41035
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5690074074074077e-05,
+      "loss": 1.6005,
+      "step": 41040
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5687111111111112e-05,
+      "loss": 1.5837,
+      "step": 41045
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.568414814814815e-05,
+      "loss": 1.4372,
+      "step": 41050
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5681185185185187e-05,
+      "loss": 1.6525,
+      "step": 41055
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5678222222222225e-05,
+      "loss": 1.5805,
+      "step": 41060
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.567525925925926e-05,
+      "loss": 1.5815,
+      "step": 41065
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.56722962962963e-05,
+      "loss": 1.6489,
+      "step": 41070
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5669333333333335e-05,
+      "loss": 1.6114,
+      "step": 41075
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5666370370370374e-05,
+      "loss": 1.5593,
+      "step": 41080
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.566340740740741e-05,
+      "loss": 1.5721,
+      "step": 41085
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 2.5660444444444448e-05,
+      "loss": 1.6531,
+      "step": 41090
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5657481481481484e-05,
+      "loss": 1.5929,
+      "step": 41095
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5654518518518522e-05,
+      "loss": 1.5989,
+      "step": 41100
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5651555555555558e-05,
+      "loss": 1.5592,
+      "step": 41105
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5648592592592597e-05,
+      "loss": 1.5599,
+      "step": 41110
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5645629629629632e-05,
+      "loss": 1.6955,
+      "step": 41115
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.564266666666667e-05,
+      "loss": 1.641,
+      "step": 41120
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5639703703703703e-05,
+      "loss": 1.6073,
+      "step": 41125
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5636740740740745e-05,
+      "loss": 1.6717,
+      "step": 41130
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.5633777777777777e-05,
+      "loss": 1.727,
+      "step": 41135
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.563081481481482e-05,
+      "loss": 1.7254,
+      "step": 41140
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 2.562785185185185e-05,
+      "loss": 1.5932,
+      "step": 41145
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5624888888888894e-05,
+      "loss": 1.5431,
+      "step": 41150
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5621925925925926e-05,
+      "loss": 1.6437,
+      "step": 41155
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5618962962962968e-05,
+      "loss": 1.6051,
+      "step": 41160
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5616e-05,
+      "loss": 1.5959,
+      "step": 41165
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5613037037037042e-05,
+      "loss": 1.6433,
+      "step": 41170
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5610074074074074e-05,
+      "loss": 1.4625,
+      "step": 41175
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5607111111111117e-05,
+      "loss": 1.5762,
+      "step": 41180
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.560414814814815e-05,
+      "loss": 1.6527,
+      "step": 41185
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.560118518518519e-05,
+      "loss": 1.5029,
+      "step": 41190
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5598222222222223e-05,
+      "loss": 1.6771,
+      "step": 41195
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 2.5595259259259258e-05,
+      "loss": 1.6128,
+      "step": 41200
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5592296296296297e-05,
+      "loss": 1.6624,
+      "step": 41205
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5589333333333333e-05,
+      "loss": 1.5417,
+      "step": 41210
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.558637037037037e-05,
+      "loss": 1.5949,
+      "step": 41215
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5583407407407407e-05,
+      "loss": 1.7662,
+      "step": 41220
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5580444444444446e-05,
+      "loss": 1.6175,
+      "step": 41225
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.557748148148148e-05,
+      "loss": 1.7357,
+      "step": 41230
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.557451851851852e-05,
+      "loss": 1.654,
+      "step": 41235
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5571555555555555e-05,
+      "loss": 1.6801,
+      "step": 41240
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.5568592592592594e-05,
+      "loss": 1.5849,
+      "step": 41245
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.556562962962963e-05,
+      "loss": 1.6278,
+      "step": 41250
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 2.556266666666667e-05,
+      "loss": 1.649,
+      "step": 41255
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5559703703703704e-05,
+      "loss": 1.6587,
+      "step": 41260
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5556740740740743e-05,
+      "loss": 1.7225,
+      "step": 41265
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5553777777777778e-05,
+      "loss": 1.5738,
+      "step": 41270
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5550814814814817e-05,
+      "loss": 1.552,
+      "step": 41275
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.554785185185185e-05,
+      "loss": 1.6288,
+      "step": 41280
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.554488888888889e-05,
+      "loss": 1.6773,
+      "step": 41285
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5541925925925923e-05,
+      "loss": 1.7602,
+      "step": 41290
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5538962962962965e-05,
+      "loss": 1.6795,
+      "step": 41295
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5535999999999997e-05,
+      "loss": 1.6347,
+      "step": 41300
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.553303703703704e-05,
+      "loss": 1.6056,
+      "step": 41305
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.553007407407407e-05,
+      "loss": 1.7138,
+      "step": 41310
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 2.5527111111111114e-05,
+      "loss": 1.7164,
+      "step": 41315
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5524148148148146e-05,
+      "loss": 1.6878,
+      "step": 41320
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5521185185185188e-05,
+      "loss": 1.6425,
+      "step": 41325
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.551822222222222e-05,
+      "loss": 1.5943,
+      "step": 41330
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5515259259259262e-05,
+      "loss": 1.5752,
+      "step": 41335
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5512296296296294e-05,
+      "loss": 1.6251,
+      "step": 41340
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5509333333333337e-05,
+      "loss": 1.6019,
+      "step": 41345
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.550637037037037e-05,
+      "loss": 1.7045,
+      "step": 41350
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5503407407407408e-05,
+      "loss": 1.6335,
+      "step": 41355
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5500444444444443e-05,
+      "loss": 1.6608,
+      "step": 41360
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5497481481481482e-05,
+      "loss": 1.5172,
+      "step": 41365
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 2.5494518518518517e-05,
+      "loss": 1.6686,
+      "step": 41370
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5491555555555556e-05,
+      "loss": 1.6731,
+      "step": 41375
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.548859259259259e-05,
+      "loss": 1.7093,
+      "step": 41380
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.548562962962963e-05,
+      "loss": 1.7359,
+      "step": 41385
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5482666666666666e-05,
+      "loss": 1.7284,
+      "step": 41390
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5479703703703705e-05,
+      "loss": 1.6845,
+      "step": 41395
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.547674074074074e-05,
+      "loss": 1.5892,
+      "step": 41400
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.547377777777778e-05,
+      "loss": 1.6972,
+      "step": 41405
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5470814814814814e-05,
+      "loss": 1.5473,
+      "step": 41410
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5467851851851853e-05,
+      "loss": 1.7014,
+      "step": 41415
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.546488888888889e-05,
+      "loss": 1.6495,
+      "step": 41420
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 2.5461925925925927e-05,
+      "loss": 1.6484,
+      "step": 41425
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5458962962962963e-05,
+      "loss": 1.5989,
+      "step": 41430
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5456e-05,
+      "loss": 1.6658,
+      "step": 41435
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5453037037037037e-05,
+      "loss": 1.6028,
+      "step": 41440
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5450074074074076e-05,
+      "loss": 1.7127,
+      "step": 41445
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.544711111111111e-05,
+      "loss": 1.5209,
+      "step": 41450
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.544414814814815e-05,
+      "loss": 1.6499,
+      "step": 41455
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5441185185185186e-05,
+      "loss": 1.6512,
+      "step": 41460
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5438222222222224e-05,
+      "loss": 1.5478,
+      "step": 41465
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.543525925925926e-05,
+      "loss": 1.6361,
+      "step": 41470
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.54322962962963e-05,
+      "loss": 1.5953,
+      "step": 41475
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 2.5429333333333334e-05,
+      "loss": 1.5439,
+      "step": 41480
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5426370370370373e-05,
+      "loss": 1.4655,
+      "step": 41485
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5423407407407408e-05,
+      "loss": 1.6708,
+      "step": 41490
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5420444444444447e-05,
+      "loss": 1.6819,
+      "step": 41495
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5417481481481483e-05,
+      "loss": 1.5493,
+      "step": 41500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.541451851851852e-05,
+      "loss": 1.6189,
+      "step": 41505
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5411555555555557e-05,
+      "loss": 1.7419,
+      "step": 41510
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5408592592592596e-05,
+      "loss": 1.6409,
+      "step": 41515
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5405629629629628e-05,
+      "loss": 1.5769,
+      "step": 41520
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.540266666666667e-05,
+      "loss": 1.6083,
+      "step": 41525
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5399703703703702e-05,
+      "loss": 1.6913,
+      "step": 41530
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5396740740740744e-05,
+      "loss": 1.6569,
+      "step": 41535
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 2.5393777777777776e-05,
+      "loss": 1.5657,
+      "step": 41540
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.539081481481482e-05,
+      "loss": 1.6118,
+      "step": 41545
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.538785185185185e-05,
+      "loss": 1.5427,
+      "step": 41550
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5384888888888893e-05,
+      "loss": 1.6034,
+      "step": 41555
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5381925925925925e-05,
+      "loss": 1.6307,
+      "step": 41560
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5378962962962967e-05,
+      "loss": 1.6683,
+      "step": 41565
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5376e-05,
+      "loss": 1.6206,
+      "step": 41570
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.537303703703704e-05,
+      "loss": 1.7441,
+      "step": 41575
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5370074074074073e-05,
+      "loss": 1.6082,
+      "step": 41580
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5367111111111112e-05,
+      "loss": 1.6098,
+      "step": 41585
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5364148148148147e-05,
+      "loss": 1.6529,
+      "step": 41590
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 2.5361185185185186e-05,
+      "loss": 1.545,
+      "step": 41595
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5358222222222222e-05,
+      "loss": 1.5832,
+      "step": 41600
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.535525925925926e-05,
+      "loss": 1.6464,
+      "step": 41605
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5352296296296296e-05,
+      "loss": 1.5574,
+      "step": 41610
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5349333333333335e-05,
+      "loss": 1.6297,
+      "step": 41615
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.534637037037037e-05,
+      "loss": 1.5489,
+      "step": 41620
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.534340740740741e-05,
+      "loss": 1.5527,
+      "step": 41625
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5340444444444444e-05,
+      "loss": 1.5542,
+      "step": 41630
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5337481481481483e-05,
+      "loss": 1.8144,
+      "step": 41635
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.533451851851852e-05,
+      "loss": 1.6329,
+      "step": 41640
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5331555555555558e-05,
+      "loss": 1.6166,
+      "step": 41645
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 2.5328592592592593e-05,
+      "loss": 1.6536,
+      "step": 41650
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5325629629629632e-05,
+      "loss": 1.7672,
+      "step": 41655
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5322666666666667e-05,
+      "loss": 1.4989,
+      "step": 41660
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5319703703703706e-05,
+      "loss": 1.6485,
+      "step": 41665
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.531674074074074e-05,
+      "loss": 1.6296,
+      "step": 41670
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.531377777777778e-05,
+      "loss": 1.6301,
+      "step": 41675
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5311407407407407e-05,
+      "loss": 1.5545,
+      "step": 41680
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5308444444444445e-05,
+      "loss": 1.6356,
+      "step": 41685
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.530548148148148e-05,
+      "loss": 1.6627,
+      "step": 41690
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.530251851851852e-05,
+      "loss": 1.6412,
+      "step": 41695
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5299555555555555e-05,
+      "loss": 1.6722,
+      "step": 41700
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 2.5296592592592594e-05,
+      "loss": 1.6125,
+      "step": 41705
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.529362962962963e-05,
+      "loss": 1.6116,
+      "step": 41710
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5290666666666668e-05,
+      "loss": 1.7176,
+      "step": 41715
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5287703703703704e-05,
+      "loss": 1.6946,
+      "step": 41720
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5284740740740743e-05,
+      "loss": 1.5997,
+      "step": 41725
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5281777777777778e-05,
+      "loss": 1.6476,
+      "step": 41730
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5278814814814817e-05,
+      "loss": 1.6579,
+      "step": 41735
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5275851851851852e-05,
+      "loss": 1.8409,
+      "step": 41740
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.527288888888889e-05,
+      "loss": 1.6001,
+      "step": 41745
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5269925925925923e-05,
+      "loss": 1.7044,
+      "step": 41750
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5266962962962965e-05,
+      "loss": 1.5899,
+      "step": 41755
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.5263999999999997e-05,
+      "loss": 1.677,
+      "step": 41760
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 2.526103703703704e-05,
+      "loss": 1.6315,
+      "step": 41765
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.525807407407407e-05,
+      "loss": 1.6409,
+      "step": 41770
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5255111111111114e-05,
+      "loss": 1.6421,
+      "step": 41775
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5252148148148146e-05,
+      "loss": 1.5427,
+      "step": 41780
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5249185185185188e-05,
+      "loss": 1.6634,
+      "step": 41785
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.524622222222222e-05,
+      "loss": 1.7061,
+      "step": 41790
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5243259259259262e-05,
+      "loss": 1.5799,
+      "step": 41795
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5240296296296294e-05,
+      "loss": 1.6489,
+      "step": 41800
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5237333333333337e-05,
+      "loss": 1.5817,
+      "step": 41805
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.523437037037037e-05,
+      "loss": 1.6374,
+      "step": 41810
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.523140740740741e-05,
+      "loss": 1.7403,
+      "step": 41815
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 2.5228444444444443e-05,
+      "loss": 1.6932,
+      "step": 41820
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.522548148148148e-05,
+      "loss": 1.6759,
+      "step": 41825
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5222518518518517e-05,
+      "loss": 1.7883,
+      "step": 41830
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5219555555555556e-05,
+      "loss": 1.5962,
+      "step": 41835
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.521659259259259e-05,
+      "loss": 1.6014,
+      "step": 41840
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.521362962962963e-05,
+      "loss": 1.5833,
+      "step": 41845
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5210666666666666e-05,
+      "loss": 1.5206,
+      "step": 41850
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5207703703703704e-05,
+      "loss": 1.6621,
+      "step": 41855
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.520474074074074e-05,
+      "loss": 1.5107,
+      "step": 41860
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.520177777777778e-05,
+      "loss": 1.5731,
+      "step": 41865
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5198814814814814e-05,
+      "loss": 1.6366,
+      "step": 41870
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 2.5195851851851853e-05,
+      "loss": 1.5646,
+      "step": 41875
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.519288888888889e-05,
+      "loss": 1.7499,
+      "step": 41880
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5189925925925927e-05,
+      "loss": 1.5806,
+      "step": 41885
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5186962962962963e-05,
+      "loss": 1.6774,
+      "step": 41890
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5184e-05,
+      "loss": 1.6753,
+      "step": 41895
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5181037037037037e-05,
+      "loss": 1.7546,
+      "step": 41900
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5178074074074076e-05,
+      "loss": 1.5589,
+      "step": 41905
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.517511111111111e-05,
+      "loss": 1.5687,
+      "step": 41910
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.517214814814815e-05,
+      "loss": 1.6259,
+      "step": 41915
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5169185185185185e-05,
+      "loss": 1.6281,
+      "step": 41920
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.5166222222222224e-05,
+      "loss": 1.6865,
+      "step": 41925
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 2.516325925925926e-05,
+      "loss": 1.6292,
+      "step": 41930
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.51602962962963e-05,
+      "loss": 1.5822,
+      "step": 41935
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5157333333333334e-05,
+      "loss": 1.6015,
+      "step": 41940
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5154370370370373e-05,
+      "loss": 1.5819,
+      "step": 41945
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5151407407407408e-05,
+      "loss": 1.614,
+      "step": 41950
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5148444444444447e-05,
+      "loss": 1.4966,
+      "step": 41955
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5145481481481482e-05,
+      "loss": 1.6192,
+      "step": 41960
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.514251851851852e-05,
+      "loss": 1.757,
+      "step": 41965
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5139555555555557e-05,
+      "loss": 1.5685,
+      "step": 41970
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5136592592592596e-05,
+      "loss": 1.7316,
+      "step": 41975
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5133629629629628e-05,
+      "loss": 1.5773,
+      "step": 41980
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.513066666666667e-05,
+      "loss": 1.6452,
+      "step": 41985
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 2.5127703703703702e-05,
+      "loss": 1.6205,
+      "step": 41990
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5124740740740744e-05,
+      "loss": 1.5643,
+      "step": 41995
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5121777777777776e-05,
+      "loss": 1.7198,
+      "step": 42000
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5118814814814818e-05,
+      "loss": 1.6528,
+      "step": 42005
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.511585185185185e-05,
+      "loss": 1.6434,
+      "step": 42010
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5112888888888893e-05,
+      "loss": 1.7483,
+      "step": 42015
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5109925925925925e-05,
+      "loss": 1.694,
+      "step": 42020
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5106962962962967e-05,
+      "loss": 1.6142,
+      "step": 42025
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5104e-05,
+      "loss": 1.6599,
+      "step": 42030
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.510103703703704e-05,
+      "loss": 1.7811,
+      "step": 42035
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5098074074074073e-05,
+      "loss": 1.6326,
+      "step": 42040
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 2.5095111111111115e-05,
+      "loss": 1.531,
+      "step": 42045
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5092148148148147e-05,
+      "loss": 1.5703,
+      "step": 42050
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5089185185185186e-05,
+      "loss": 1.5535,
+      "step": 42055
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.508622222222222e-05,
+      "loss": 1.6417,
+      "step": 42060
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.508325925925926e-05,
+      "loss": 1.5833,
+      "step": 42065
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5080296296296296e-05,
+      "loss": 1.6644,
+      "step": 42070
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5077333333333335e-05,
+      "loss": 1.6165,
+      "step": 42075
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.507437037037037e-05,
+      "loss": 1.6205,
+      "step": 42080
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.507140740740741e-05,
+      "loss": 1.5739,
+      "step": 42085
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5068444444444444e-05,
+      "loss": 1.526,
+      "step": 42090
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.5065481481481483e-05,
+      "loss": 1.7597,
+      "step": 42095
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 2.506251851851852e-05,
+      "loss": 1.4853,
+      "step": 42100
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5059555555555557e-05,
+      "loss": 1.6709,
+      "step": 42105
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5056592592592593e-05,
+      "loss": 1.6051,
+      "step": 42110
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5053629629629632e-05,
+      "loss": 1.6251,
+      "step": 42115
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5050666666666667e-05,
+      "loss": 1.65,
+      "step": 42120
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5047703703703706e-05,
+      "loss": 1.542,
+      "step": 42125
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.504474074074074e-05,
+      "loss": 1.6226,
+      "step": 42130
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.504177777777778e-05,
+      "loss": 1.5605,
+      "step": 42135
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5038814814814816e-05,
+      "loss": 1.6663,
+      "step": 42140
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.5035851851851854e-05,
+      "loss": 1.6892,
+      "step": 42145
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.503288888888889e-05,
+      "loss": 1.6128,
+      "step": 42150
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 2.502992592592593e-05,
+      "loss": 1.7222,
+      "step": 42155
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5026962962962964e-05,
+      "loss": 1.5738,
+      "step": 42160
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5024000000000003e-05,
+      "loss": 1.6176,
+      "step": 42165
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.502103703703704e-05,
+      "loss": 1.7673,
+      "step": 42170
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5018074074074077e-05,
+      "loss": 1.7285,
+      "step": 42175
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5015111111111113e-05,
+      "loss": 1.6898,
+      "step": 42180
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.501214814814815e-05,
+      "loss": 1.6084,
+      "step": 42185
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5009185185185187e-05,
+      "loss": 1.5221,
+      "step": 42190
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.5006222222222226e-05,
+      "loss": 1.6874,
+      "step": 42195
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.500325925925926e-05,
+      "loss": 1.5295,
+      "step": 42200
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.50002962962963e-05,
+      "loss": 1.6089,
+      "step": 42205
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.4997333333333332e-05,
+      "loss": 1.6896,
+      "step": 42210
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 2.499437037037037e-05,
+      "loss": 1.6879,
+      "step": 42215
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4991407407407406e-05,
+      "loss": 1.6323,
+      "step": 42220
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4988444444444445e-05,
+      "loss": 1.7671,
+      "step": 42225
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.498548148148148e-05,
+      "loss": 1.6656,
+      "step": 42230
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.498251851851852e-05,
+      "loss": 1.6066,
+      "step": 42235
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4979555555555555e-05,
+      "loss": 1.6886,
+      "step": 42240
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4976592592592594e-05,
+      "loss": 1.5731,
+      "step": 42245
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.497362962962963e-05,
+      "loss": 1.6841,
+      "step": 42250
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4970666666666668e-05,
+      "loss": 1.6698,
+      "step": 42255
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4967703703703703e-05,
+      "loss": 1.7037,
+      "step": 42260
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4964740740740742e-05,
+      "loss": 1.5582,
+      "step": 42265
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 2.4961777777777778e-05,
+      "loss": 1.7294,
+      "step": 42270
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4958814814814816e-05,
+      "loss": 1.6091,
+      "step": 42275
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4955851851851852e-05,
+      "loss": 1.6461,
+      "step": 42280
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.495288888888889e-05,
+      "loss": 1.634,
+      "step": 42285
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4949925925925926e-05,
+      "loss": 1.6421,
+      "step": 42290
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4946962962962965e-05,
+      "loss": 1.8087,
+      "step": 42295
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4944e-05,
+      "loss": 1.5589,
+      "step": 42300
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.494103703703704e-05,
+      "loss": 1.4519,
+      "step": 42305
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4938074074074075e-05,
+      "loss": 1.543,
+      "step": 42310
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4935111111111113e-05,
+      "loss": 1.6649,
+      "step": 42315
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.493214814814815e-05,
+      "loss": 1.7435,
+      "step": 42320
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 2.4929185185185184e-05,
+      "loss": 1.6823,
+      "step": 42325
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4926222222222223e-05,
+      "loss": 1.5719,
+      "step": 42330
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.492325925925926e-05,
+      "loss": 1.592,
+      "step": 42335
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4920296296296297e-05,
+      "loss": 1.6043,
+      "step": 42340
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4917333333333333e-05,
+      "loss": 1.6624,
+      "step": 42345
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.491437037037037e-05,
+      "loss": 1.5336,
+      "step": 42350
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4911407407407407e-05,
+      "loss": 1.6267,
+      "step": 42355
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4908444444444446e-05,
+      "loss": 1.609,
+      "step": 42360
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.490548148148148e-05,
+      "loss": 1.6932,
+      "step": 42365
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.490251851851852e-05,
+      "loss": 1.6819,
+      "step": 42370
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4899555555555556e-05,
+      "loss": 1.7832,
+      "step": 42375
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 2.4896592592592594e-05,
+      "loss": 1.5035,
+      "step": 42380
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.489362962962963e-05,
+      "loss": 1.5465,
+      "step": 42385
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.489066666666667e-05,
+      "loss": 1.6317,
+      "step": 42390
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4887703703703704e-05,
+      "loss": 1.5068,
+      "step": 42395
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4884740740740743e-05,
+      "loss": 1.7252,
+      "step": 42400
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.488177777777778e-05,
+      "loss": 1.7102,
+      "step": 42405
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4878814814814817e-05,
+      "loss": 1.7168,
+      "step": 42410
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4875851851851853e-05,
+      "loss": 1.5764,
+      "step": 42415
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.487288888888889e-05,
+      "loss": 1.7415,
+      "step": 42420
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4869925925925927e-05,
+      "loss": 1.6487,
+      "step": 42425
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4866962962962966e-05,
+      "loss": 1.6504,
+      "step": 42430
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4864e-05,
+      "loss": 1.7856,
+      "step": 42435
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 2.4861037037037037e-05,
+      "loss": 1.5304,
+      "step": 42440
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4858074074074075e-05,
+      "loss": 1.6206,
+      "step": 42445
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.485511111111111e-05,
+      "loss": 1.7216,
+      "step": 42450
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.485214814814815e-05,
+      "loss": 1.6094,
+      "step": 42455
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4849185185185185e-05,
+      "loss": 1.569,
+      "step": 42460
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4846222222222224e-05,
+      "loss": 1.6995,
+      "step": 42465
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.484325925925926e-05,
+      "loss": 1.6641,
+      "step": 42470
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4840296296296298e-05,
+      "loss": 1.744,
+      "step": 42475
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4837333333333334e-05,
+      "loss": 1.4976,
+      "step": 42480
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4834370370370372e-05,
+      "loss": 1.5236,
+      "step": 42485
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4831407407407408e-05,
+      "loss": 1.6698,
+      "step": 42490
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.4828444444444447e-05,
+      "loss": 1.5981,
+      "step": 42495
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4825481481481482e-05,
+      "loss": 1.5869,
+      "step": 42500
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.482251851851852e-05,
+      "loss": 1.5718,
+      "step": 42505
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4819555555555556e-05,
+      "loss": 1.6357,
+      "step": 42510
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4816592592592595e-05,
+      "loss": 1.702,
+      "step": 42515
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.481362962962963e-05,
+      "loss": 1.6627,
+      "step": 42520
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.481066666666667e-05,
+      "loss": 1.6371,
+      "step": 42525
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4807703703703705e-05,
+      "loss": 1.5821,
+      "step": 42530
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4804740740740744e-05,
+      "loss": 1.6404,
+      "step": 42535
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.480177777777778e-05,
+      "loss": 1.6249,
+      "step": 42540
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4798814814814818e-05,
+      "loss": 1.5984,
+      "step": 42545
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.4795851851851853e-05,
+      "loss": 1.5236,
+      "step": 42550
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.479288888888889e-05,
+      "loss": 1.6313,
+      "step": 42555
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4789925925925928e-05,
+      "loss": 1.5679,
+      "step": 42560
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4786962962962963e-05,
+      "loss": 1.6113,
+      "step": 42565
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4784000000000002e-05,
+      "loss": 1.6254,
+      "step": 42570
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4781037037037037e-05,
+      "loss": 1.8004,
+      "step": 42575
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4778074074074076e-05,
+      "loss": 1.6595,
+      "step": 42580
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.477511111111111e-05,
+      "loss": 1.6381,
+      "step": 42585
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.477214814814815e-05,
+      "loss": 1.6644,
+      "step": 42590
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4769185185185186e-05,
+      "loss": 1.6372,
+      "step": 42595
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.4766222222222225e-05,
+      "loss": 1.6561,
+      "step": 42600
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.476325925925926e-05,
+      "loss": 1.6087,
+      "step": 42605
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.47602962962963e-05,
+      "loss": 1.7179,
+      "step": 42610
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4757333333333334e-05,
+      "loss": 1.6885,
+      "step": 42615
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4754370370370373e-05,
+      "loss": 1.5692,
+      "step": 42620
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.475140740740741e-05,
+      "loss": 1.486,
+      "step": 42625
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4748444444444447e-05,
+      "loss": 1.5632,
+      "step": 42630
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4745481481481483e-05,
+      "loss": 1.6604,
+      "step": 42635
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4742518518518522e-05,
+      "loss": 1.5488,
+      "step": 42640
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4739555555555557e-05,
+      "loss": 1.7109,
+      "step": 42645
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4736592592592596e-05,
+      "loss": 1.6458,
+      "step": 42650
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.473362962962963e-05,
+      "loss": 1.6154,
+      "step": 42655
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.473066666666667e-05,
+      "loss": 1.5993,
+      "step": 42660
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.4727703703703706e-05,
+      "loss": 1.6742,
+      "step": 42665
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.472474074074074e-05,
+      "loss": 1.6274,
+      "step": 42670
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.472177777777778e-05,
+      "loss": 1.5852,
+      "step": 42675
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4718814814814815e-05,
+      "loss": 1.6499,
+      "step": 42680
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4715851851851854e-05,
+      "loss": 1.6443,
+      "step": 42685
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.471288888888889e-05,
+      "loss": 1.735,
+      "step": 42690
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.470992592592593e-05,
+      "loss": 1.7045,
+      "step": 42695
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4706962962962964e-05,
+      "loss": 1.5261,
+      "step": 42700
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4704000000000003e-05,
+      "loss": 1.5027,
+      "step": 42705
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4701037037037038e-05,
+      "loss": 1.615,
+      "step": 42710
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4698074074074077e-05,
+      "loss": 1.7258,
+      "step": 42715
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.4695111111111112e-05,
+      "loss": 1.7026,
+      "step": 42720
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.469214814814815e-05,
+      "loss": 1.5972,
+      "step": 42725
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4689185185185187e-05,
+      "loss": 1.5827,
+      "step": 42730
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4686222222222225e-05,
+      "loss": 1.7622,
+      "step": 42735
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.468325925925926e-05,
+      "loss": 1.6694,
+      "step": 42740
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.46802962962963e-05,
+      "loss": 1.7835,
+      "step": 42745
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4677333333333335e-05,
+      "loss": 1.6116,
+      "step": 42750
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4674370370370374e-05,
+      "loss": 1.6849,
+      "step": 42755
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.467140740740741e-05,
+      "loss": 1.6807,
+      "step": 42760
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4668444444444448e-05,
+      "loss": 1.7528,
+      "step": 42765
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4665481481481484e-05,
+      "loss": 1.579,
+      "step": 42770
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.4662518518518522e-05,
+      "loss": 1.6419,
+      "step": 42775
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4659555555555558e-05,
+      "loss": 1.6443,
+      "step": 42780
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4656592592592593e-05,
+      "loss": 1.6024,
+      "step": 42785
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4653629629629632e-05,
+      "loss": 1.5986,
+      "step": 42790
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4650666666666668e-05,
+      "loss": 1.6702,
+      "step": 42795
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4647703703703706e-05,
+      "loss": 1.7322,
+      "step": 42800
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4644740740740742e-05,
+      "loss": 1.7453,
+      "step": 42805
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.464177777777778e-05,
+      "loss": 1.6644,
+      "step": 42810
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4638814814814816e-05,
+      "loss": 1.5886,
+      "step": 42815
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4635851851851855e-05,
+      "loss": 1.6832,
+      "step": 42820
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.463288888888889e-05,
+      "loss": 1.6364,
+      "step": 42825
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.4629925925925926e-05,
+      "loss": 1.6866,
+      "step": 42830
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.462696296296296e-05,
+      "loss": 1.6079,
+      "step": 42835
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4624e-05,
+      "loss": 1.7441,
+      "step": 42840
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4621037037037035e-05,
+      "loss": 1.6391,
+      "step": 42845
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4618074074074074e-05,
+      "loss": 1.69,
+      "step": 42850
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.461511111111111e-05,
+      "loss": 1.6441,
+      "step": 42855
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.461214814814815e-05,
+      "loss": 1.6265,
+      "step": 42860
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4609185185185184e-05,
+      "loss": 1.6849,
+      "step": 42865
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4606222222222223e-05,
+      "loss": 1.7985,
+      "step": 42870
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4603259259259258e-05,
+      "loss": 1.6619,
+      "step": 42875
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4600296296296297e-05,
+      "loss": 1.5778,
+      "step": 42880
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.4597333333333333e-05,
+      "loss": 1.5081,
+      "step": 42885
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.459437037037037e-05,
+      "loss": 1.5907,
+      "step": 42890
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4591407407407407e-05,
+      "loss": 1.6738,
+      "step": 42895
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4588444444444446e-05,
+      "loss": 1.6922,
+      "step": 42900
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.458548148148148e-05,
+      "loss": 1.5943,
+      "step": 42905
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.458251851851852e-05,
+      "loss": 1.6382,
+      "step": 42910
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4579555555555555e-05,
+      "loss": 1.634,
+      "step": 42915
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4576592592592594e-05,
+      "loss": 1.7045,
+      "step": 42920
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.457362962962963e-05,
+      "loss": 1.6522,
+      "step": 42925
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.457066666666667e-05,
+      "loss": 1.6777,
+      "step": 42930
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4567703703703704e-05,
+      "loss": 1.5629,
+      "step": 42935
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4564740740740743e-05,
+      "loss": 1.7201,
+      "step": 42940
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.4561777777777778e-05,
+      "loss": 1.7041,
+      "step": 42945
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4558814814814813e-05,
+      "loss": 1.6801,
+      "step": 42950
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4555851851851852e-05,
+      "loss": 1.604,
+      "step": 42955
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4552888888888888e-05,
+      "loss": 1.5547,
+      "step": 42960
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4549925925925927e-05,
+      "loss": 1.7886,
+      "step": 42965
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4546962962962962e-05,
+      "loss": 1.5996,
+      "step": 42970
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4544e-05,
+      "loss": 1.6517,
+      "step": 42975
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4541037037037036e-05,
+      "loss": 1.6069,
+      "step": 42980
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4538074074074075e-05,
+      "loss": 1.5577,
+      "step": 42985
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.453511111111111e-05,
+      "loss": 1.6836,
+      "step": 42990
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.453214814814815e-05,
+      "loss": 1.6156,
+      "step": 42995
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.4529185185185185e-05,
+      "loss": 1.5328,
+      "step": 43000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4526222222222224e-05,
+      "loss": 1.6149,
+      "step": 43005
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.452325925925926e-05,
+      "loss": 1.6449,
+      "step": 43010
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4520296296296298e-05,
+      "loss": 1.5699,
+      "step": 43015
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4517333333333333e-05,
+      "loss": 1.5564,
+      "step": 43020
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4514370370370372e-05,
+      "loss": 1.7178,
+      "step": 43025
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4511407407407408e-05,
+      "loss": 1.6843,
+      "step": 43030
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4508444444444446e-05,
+      "loss": 1.6051,
+      "step": 43035
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4505481481481482e-05,
+      "loss": 1.624,
+      "step": 43040
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.450251851851852e-05,
+      "loss": 1.6851,
+      "step": 43045
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4499555555555556e-05,
+      "loss": 1.5136,
+      "step": 43050
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.4496592592592595e-05,
+      "loss": 1.5978,
+      "step": 43055
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.449362962962963e-05,
+      "loss": 1.715,
+      "step": 43060
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4490666666666666e-05,
+      "loss": 1.6392,
+      "step": 43065
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4487703703703705e-05,
+      "loss": 1.718,
+      "step": 43070
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.448474074074074e-05,
+      "loss": 1.682,
+      "step": 43075
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.448177777777778e-05,
+      "loss": 1.6026,
+      "step": 43080
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4478814814814814e-05,
+      "loss": 1.6445,
+      "step": 43085
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4475851851851853e-05,
+      "loss": 1.488,
+      "step": 43090
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.447288888888889e-05,
+      "loss": 1.6129,
+      "step": 43095
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4469925925925927e-05,
+      "loss": 1.7074,
+      "step": 43100
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4466962962962963e-05,
+      "loss": 1.6178,
+      "step": 43105
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4464e-05,
+      "loss": 1.7074,
+      "step": 43110
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.4461037037037037e-05,
+      "loss": 1.6178,
+      "step": 43115
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4458074074074076e-05,
+      "loss": 1.5888,
+      "step": 43120
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.445511111111111e-05,
+      "loss": 1.7443,
+      "step": 43125
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.445214814814815e-05,
+      "loss": 1.6867,
+      "step": 43130
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4449185185185186e-05,
+      "loss": 1.5555,
+      "step": 43135
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4446222222222224e-05,
+      "loss": 1.6084,
+      "step": 43140
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.444325925925926e-05,
+      "loss": 1.7005,
+      "step": 43145
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.44402962962963e-05,
+      "loss": 1.6509,
+      "step": 43150
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4437333333333334e-05,
+      "loss": 1.675,
+      "step": 43155
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4434370370370373e-05,
+      "loss": 1.7306,
+      "step": 43160
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4431407407407408e-05,
+      "loss": 1.6103,
+      "step": 43165
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.4428444444444447e-05,
+      "loss": 1.518,
+      "step": 43170
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4425481481481483e-05,
+      "loss": 1.6322,
+      "step": 43175
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4422518518518518e-05,
+      "loss": 1.7166,
+      "step": 43180
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4419555555555557e-05,
+      "loss": 1.6174,
+      "step": 43185
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4416592592592592e-05,
+      "loss": 1.7199,
+      "step": 43190
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.441362962962963e-05,
+      "loss": 1.5637,
+      "step": 43195
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4410666666666667e-05,
+      "loss": 1.7475,
+      "step": 43200
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4407703703703705e-05,
+      "loss": 1.6044,
+      "step": 43205
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.440474074074074e-05,
+      "loss": 1.5536,
+      "step": 43210
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.440177777777778e-05,
+      "loss": 1.6782,
+      "step": 43215
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4398814814814815e-05,
+      "loss": 1.6251,
+      "step": 43220
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.4395851851851854e-05,
+      "loss": 1.712,
+      "step": 43225
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.439288888888889e-05,
+      "loss": 1.6403,
+      "step": 43230
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4389925925925928e-05,
+      "loss": 1.6092,
+      "step": 43235
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4386962962962964e-05,
+      "loss": 1.6144,
+      "step": 43240
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4384000000000002e-05,
+      "loss": 1.694,
+      "step": 43245
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4381037037037038e-05,
+      "loss": 1.6376,
+      "step": 43250
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4378074074074077e-05,
+      "loss": 1.7075,
+      "step": 43255
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4375111111111112e-05,
+      "loss": 1.5395,
+      "step": 43260
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.437214814814815e-05,
+      "loss": 1.6706,
+      "step": 43265
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4369185185185186e-05,
+      "loss": 1.7421,
+      "step": 43270
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.4366222222222225e-05,
+      "loss": 1.5826,
+      "step": 43275
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.436325925925926e-05,
+      "loss": 1.6134,
+      "step": 43280
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.43602962962963e-05,
+      "loss": 1.6581,
+      "step": 43285
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4357333333333335e-05,
+      "loss": 1.6876,
+      "step": 43290
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.435437037037037e-05,
+      "loss": 1.6236,
+      "step": 43295
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.435140740740741e-05,
+      "loss": 1.6006,
+      "step": 43300
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4348444444444445e-05,
+      "loss": 1.5795,
+      "step": 43305
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4345481481481483e-05,
+      "loss": 1.4963,
+      "step": 43310
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.434251851851852e-05,
+      "loss": 1.7048,
+      "step": 43315
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4339555555555558e-05,
+      "loss": 1.606,
+      "step": 43320
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4336592592592593e-05,
+      "loss": 1.6692,
+      "step": 43325
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4333629629629632e-05,
+      "loss": 1.6537,
+      "step": 43330
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4330666666666667e-05,
+      "loss": 1.5897,
+      "step": 43335
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.4327703703703706e-05,
+      "loss": 1.5532,
+      "step": 43340
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.432474074074074e-05,
+      "loss": 1.7561,
+      "step": 43345
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.432177777777778e-05,
+      "loss": 1.7181,
+      "step": 43350
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4318814814814816e-05,
+      "loss": 1.6277,
+      "step": 43355
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4315851851851855e-05,
+      "loss": 1.6867,
+      "step": 43360
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.431288888888889e-05,
+      "loss": 1.7926,
+      "step": 43365
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.430992592592593e-05,
+      "loss": 1.6746,
+      "step": 43370
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4306962962962964e-05,
+      "loss": 1.5804,
+      "step": 43375
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4304000000000003e-05,
+      "loss": 1.6194,
+      "step": 43380
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.430103703703704e-05,
+      "loss": 1.6775,
+      "step": 43385
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4298074074074077e-05,
+      "loss": 1.6694,
+      "step": 43390
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.4295111111111113e-05,
+      "loss": 1.628,
+      "step": 43395
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.429214814814815e-05,
+      "loss": 1.5887,
+      "step": 43400
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4289185185185187e-05,
+      "loss": 1.643,
+      "step": 43405
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4286222222222222e-05,
+      "loss": 1.5909,
+      "step": 43410
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.428325925925926e-05,
+      "loss": 1.5589,
+      "step": 43415
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4280296296296297e-05,
+      "loss": 1.6534,
+      "step": 43420
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4277333333333336e-05,
+      "loss": 1.6822,
+      "step": 43425
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.427437037037037e-05,
+      "loss": 1.7091,
+      "step": 43430
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.427140740740741e-05,
+      "loss": 1.6612,
+      "step": 43435
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4268444444444445e-05,
+      "loss": 1.7228,
+      "step": 43440
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.4265481481481484e-05,
+      "loss": 1.6556,
+      "step": 43445
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.426251851851852e-05,
+      "loss": 1.6636,
+      "step": 43450
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.425955555555556e-05,
+      "loss": 1.5737,
+      "step": 43455
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4256592592592594e-05,
+      "loss": 1.648,
+      "step": 43460
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4253629629629633e-05,
+      "loss": 1.6036,
+      "step": 43465
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4250666666666668e-05,
+      "loss": 1.7201,
+      "step": 43470
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4247703703703707e-05,
+      "loss": 1.5684,
+      "step": 43475
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4244740740740742e-05,
+      "loss": 1.6336,
+      "step": 43480
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.424177777777778e-05,
+      "loss": 1.8631,
+      "step": 43485
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4238814814814817e-05,
+      "loss": 1.7817,
+      "step": 43490
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.4235851851851855e-05,
+      "loss": 1.546,
+      "step": 43495
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.423288888888889e-05,
+      "loss": 1.513,
+      "step": 43500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.422992592592593e-05,
+      "loss": 1.4688,
+      "step": 43505
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4226962962962965e-05,
+      "loss": 1.6324,
+      "step": 43510
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4224000000000004e-05,
+      "loss": 1.6673,
+      "step": 43515
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.422103703703704e-05,
+      "loss": 1.7393,
+      "step": 43520
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4218074074074075e-05,
+      "loss": 1.6036,
+      "step": 43525
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4215111111111114e-05,
+      "loss": 1.7096,
+      "step": 43530
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.421214814814815e-05,
+      "loss": 1.5361,
+      "step": 43535
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4209185185185188e-05,
+      "loss": 1.605,
+      "step": 43540
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4206222222222223e-05,
+      "loss": 1.7471,
+      "step": 43545
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4203259259259262e-05,
+      "loss": 1.6791,
+      "step": 43550
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4200296296296298e-05,
+      "loss": 1.6465,
+      "step": 43555
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4197333333333336e-05,
+      "loss": 1.6986,
+      "step": 43560
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.4194370370370372e-05,
+      "loss": 1.56,
+      "step": 43565
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.419140740740741e-05,
+      "loss": 1.668,
+      "step": 43570
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4188444444444446e-05,
+      "loss": 1.5966,
+      "step": 43575
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4185481481481485e-05,
+      "loss": 1.665,
+      "step": 43580
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.418251851851852e-05,
+      "loss": 1.6944,
+      "step": 43585
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.417955555555556e-05,
+      "loss": 1.6117,
+      "step": 43590
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4176592592592595e-05,
+      "loss": 1.6165,
+      "step": 43595
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.417362962962963e-05,
+      "loss": 1.717,
+      "step": 43600
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4170666666666665e-05,
+      "loss": 1.7191,
+      "step": 43605
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.4167703703703704e-05,
+      "loss": 1.6871,
+      "step": 43610
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.416474074074074e-05,
+      "loss": 1.5964,
+      "step": 43615
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.416177777777778e-05,
+      "loss": 1.526,
+      "step": 43620
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4158814814814814e-05,
+      "loss": 1.6351,
+      "step": 43625
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4155851851851853e-05,
+      "loss": 1.68,
+      "step": 43630
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4152888888888888e-05,
+      "loss": 1.5641,
+      "step": 43635
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4149925925925927e-05,
+      "loss": 1.6968,
+      "step": 43640
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4146962962962962e-05,
+      "loss": 1.7593,
+      "step": 43645
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4144e-05,
+      "loss": 1.7301,
+      "step": 43650
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4141037037037037e-05,
+      "loss": 1.7554,
+      "step": 43655
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4138074074074076e-05,
+      "loss": 1.6871,
+      "step": 43660
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.413511111111111e-05,
+      "loss": 1.5504,
+      "step": 43665
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.413214814814815e-05,
+      "loss": 1.7713,
+      "step": 43670
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.4129185185185185e-05,
+      "loss": 1.6444,
+      "step": 43675
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.412622222222222e-05,
+      "loss": 1.6298,
+      "step": 43680
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.412325925925926e-05,
+      "loss": 1.6716,
+      "step": 43685
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4120296296296295e-05,
+      "loss": 1.698,
+      "step": 43690
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4117333333333334e-05,
+      "loss": 1.6255,
+      "step": 43695
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.411437037037037e-05,
+      "loss": 1.6152,
+      "step": 43700
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4111407407407408e-05,
+      "loss": 1.6757,
+      "step": 43705
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4108444444444443e-05,
+      "loss": 1.6195,
+      "step": 43710
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4105481481481482e-05,
+      "loss": 1.5766,
+      "step": 43715
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4102518518518518e-05,
+      "loss": 1.7298,
+      "step": 43720
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4099555555555556e-05,
+      "loss": 1.6454,
+      "step": 43725
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4096592592592592e-05,
+      "loss": 1.6131,
+      "step": 43730
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.409362962962963e-05,
+      "loss": 1.7367,
+      "step": 43735
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4090666666666666e-05,
+      "loss": 1.6687,
+      "step": 43740
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4087703703703705e-05,
+      "loss": 1.6111,
+      "step": 43745
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.408474074074074e-05,
+      "loss": 1.5982,
+      "step": 43750
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.408177777777778e-05,
+      "loss": 1.5327,
+      "step": 43755
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4078814814814815e-05,
+      "loss": 1.5602,
+      "step": 43760
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4075851851851854e-05,
+      "loss": 1.5002,
+      "step": 43765
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.407288888888889e-05,
+      "loss": 1.5914,
+      "step": 43770
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4069925925925928e-05,
+      "loss": 1.6102,
+      "step": 43775
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4066962962962963e-05,
+      "loss": 1.6541,
+      "step": 43780
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4064000000000002e-05,
+      "loss": 1.4847,
+      "step": 43785
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4061037037037037e-05,
+      "loss": 1.7369,
+      "step": 43790
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4058074074074073e-05,
+      "loss": 1.5941,
+      "step": 43795
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4055111111111112e-05,
+      "loss": 1.5949,
+      "step": 43800
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4052148148148147e-05,
+      "loss": 1.7215,
+      "step": 43805
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4049185185185186e-05,
+      "loss": 1.5552,
+      "step": 43810
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.404622222222222e-05,
+      "loss": 1.7436,
+      "step": 43815
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.404325925925926e-05,
+      "loss": 1.6751,
+      "step": 43820
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4040296296296296e-05,
+      "loss": 1.7211,
+      "step": 43825
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4037333333333334e-05,
+      "loss": 1.6741,
+      "step": 43830
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.403437037037037e-05,
+      "loss": 1.7234,
+      "step": 43835
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.403140740740741e-05,
+      "loss": 1.7538,
+      "step": 43840
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4028444444444444e-05,
+      "loss": 1.5806,
+      "step": 43845
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4025481481481483e-05,
+      "loss": 1.6546,
+      "step": 43850
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.402251851851852e-05,
+      "loss": 1.7191,
+      "step": 43855
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4019555555555557e-05,
+      "loss": 1.5458,
+      "step": 43860
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4016592592592593e-05,
+      "loss": 1.7858,
+      "step": 43865
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.401362962962963e-05,
+      "loss": 1.6788,
+      "step": 43870
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4010666666666667e-05,
+      "loss": 1.593,
+      "step": 43875
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4007703703703706e-05,
+      "loss": 1.7663,
+      "step": 43880
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.400474074074074e-05,
+      "loss": 1.5753,
+      "step": 43885
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.400177777777778e-05,
+      "loss": 1.6321,
+      "step": 43890
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3998814814814815e-05,
+      "loss": 1.6525,
+      "step": 43895
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.3995851851851854e-05,
+      "loss": 1.6741,
+      "step": 43900
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.399288888888889e-05,
+      "loss": 1.6693,
+      "step": 43905
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3989925925925925e-05,
+      "loss": 1.587,
+      "step": 43910
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3986962962962964e-05,
+      "loss": 1.6631,
+      "step": 43915
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3984e-05,
+      "loss": 1.7355,
+      "step": 43920
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3981037037037038e-05,
+      "loss": 1.6301,
+      "step": 43925
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3978074074074074e-05,
+      "loss": 1.6278,
+      "step": 43930
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3975111111111112e-05,
+      "loss": 1.5621,
+      "step": 43935
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3972148148148148e-05,
+      "loss": 1.7544,
+      "step": 43940
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3969185185185187e-05,
+      "loss": 1.5575,
+      "step": 43945
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3966222222222222e-05,
+      "loss": 1.5192,
+      "step": 43950
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.396325925925926e-05,
+      "loss": 1.7646,
+      "step": 43955
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3960296296296296e-05,
+      "loss": 1.7212,
+      "step": 43960
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3957333333333335e-05,
+      "loss": 1.6293,
+      "step": 43965
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.395437037037037e-05,
+      "loss": 1.7256,
+      "step": 43970
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.395140740740741e-05,
+      "loss": 1.6934,
+      "step": 43975
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3948444444444445e-05,
+      "loss": 1.5999,
+      "step": 43980
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3945481481481484e-05,
+      "loss": 1.7327,
+      "step": 43985
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.394251851851852e-05,
+      "loss": 1.6156,
+      "step": 43990
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3939555555555558e-05,
+      "loss": 1.6161,
+      "step": 43995
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3936592592592593e-05,
+      "loss": 1.6907,
+      "step": 44000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3933629629629632e-05,
+      "loss": 1.6608,
+      "step": 44005
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3930666666666668e-05,
+      "loss": 1.6311,
+      "step": 44010
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3927703703703707e-05,
+      "loss": 1.5601,
+      "step": 44015
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3924740740740742e-05,
+      "loss": 1.6394,
+      "step": 44020
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3921777777777777e-05,
+      "loss": 1.7379,
+      "step": 44025
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3918814814814816e-05,
+      "loss": 1.7504,
+      "step": 44030
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.391585185185185e-05,
+      "loss": 1.6393,
+      "step": 44035
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.391288888888889e-05,
+      "loss": 1.6309,
+      "step": 44040
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3909925925925926e-05,
+      "loss": 1.78,
+      "step": 44045
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3906962962962965e-05,
+      "loss": 1.6029,
+      "step": 44050
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3904e-05,
+      "loss": 1.6456,
+      "step": 44055
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.390103703703704e-05,
+      "loss": 1.6269,
+      "step": 44060
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3898074074074074e-05,
+      "loss": 1.6041,
+      "step": 44065
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3895111111111113e-05,
+      "loss": 1.554,
+      "step": 44070
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.389214814814815e-05,
+      "loss": 1.6295,
+      "step": 44075
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3889185185185188e-05,
+      "loss": 1.5618,
+      "step": 44080
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3886222222222223e-05,
+      "loss": 1.5222,
+      "step": 44085
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3883259259259262e-05,
+      "loss": 1.6033,
+      "step": 44090
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3880296296296297e-05,
+      "loss": 1.6292,
+      "step": 44095
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3877333333333336e-05,
+      "loss": 1.713,
+      "step": 44100
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.387437037037037e-05,
+      "loss": 1.643,
+      "step": 44105
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.387140740740741e-05,
+      "loss": 1.5765,
+      "step": 44110
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3868444444444446e-05,
+      "loss": 1.6238,
+      "step": 44115
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3865481481481485e-05,
+      "loss": 1.52,
+      "step": 44120
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.386251851851852e-05,
+      "loss": 1.5639,
+      "step": 44125
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.385955555555556e-05,
+      "loss": 1.5887,
+      "step": 44130
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3856592592592594e-05,
+      "loss": 1.618,
+      "step": 44135
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.385362962962963e-05,
+      "loss": 1.7144,
+      "step": 44140
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.385066666666667e-05,
+      "loss": 1.6376,
+      "step": 44145
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3847703703703704e-05,
+      "loss": 1.7256,
+      "step": 44150
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3844740740740743e-05,
+      "loss": 1.6934,
+      "step": 44155
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3841777777777778e-05,
+      "loss": 1.5969,
+      "step": 44160
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3838814814814817e-05,
+      "loss": 1.7101,
+      "step": 44165
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3835851851851852e-05,
+      "loss": 1.6164,
+      "step": 44170
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.383288888888889e-05,
+      "loss": 1.6782,
+      "step": 44175
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3829925925925927e-05,
+      "loss": 1.7189,
+      "step": 44180
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3826962962962966e-05,
+      "loss": 1.6555,
+      "step": 44185
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3824e-05,
+      "loss": 1.678,
+      "step": 44190
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.382103703703704e-05,
+      "loss": 1.7067,
+      "step": 44195
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3818074074074075e-05,
+      "loss": 1.7777,
+      "step": 44200
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3815111111111114e-05,
+      "loss": 1.6683,
+      "step": 44205
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.381214814814815e-05,
+      "loss": 1.6967,
+      "step": 44210
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3809185185185188e-05,
+      "loss": 1.6287,
+      "step": 44215
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3806222222222224e-05,
+      "loss": 1.5928,
+      "step": 44220
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3803259259259263e-05,
+      "loss": 1.7359,
+      "step": 44225
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3800296296296298e-05,
+      "loss": 1.678,
+      "step": 44230
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3797333333333337e-05,
+      "loss": 1.477,
+      "step": 44235
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.3794370370370372e-05,
+      "loss": 1.7596,
+      "step": 44240
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.379140740740741e-05,
+      "loss": 1.6922,
+      "step": 44245
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3788444444444446e-05,
+      "loss": 1.7037,
+      "step": 44250
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3785481481481482e-05,
+      "loss": 1.4522,
+      "step": 44255
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.378251851851852e-05,
+      "loss": 1.7415,
+      "step": 44260
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3779555555555556e-05,
+      "loss": 1.6058,
+      "step": 44265
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3776592592592595e-05,
+      "loss": 1.6538,
+      "step": 44270
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.377362962962963e-05,
+      "loss": 1.6966,
+      "step": 44275
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.377066666666667e-05,
+      "loss": 1.7313,
+      "step": 44280
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3767703703703705e-05,
+      "loss": 1.8092,
+      "step": 44285
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.3764740740740744e-05,
+      "loss": 1.7834,
+      "step": 44290
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.376177777777778e-05,
+      "loss": 1.5814,
+      "step": 44295
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3758814814814818e-05,
+      "loss": 1.6047,
+      "step": 44300
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3755851851851853e-05,
+      "loss": 1.5985,
+      "step": 44305
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3752888888888892e-05,
+      "loss": 1.4982,
+      "step": 44310
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3749925925925927e-05,
+      "loss": 1.6979,
+      "step": 44315
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3746962962962966e-05,
+      "loss": 1.6143,
+      "step": 44320
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3744000000000002e-05,
+      "loss": 1.6739,
+      "step": 44325
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.374103703703704e-05,
+      "loss": 1.684,
+      "step": 44330
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3738074074074076e-05,
+      "loss": 1.6024,
+      "step": 44335
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.3735111111111115e-05,
+      "loss": 1.6681,
+      "step": 44340
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.373214814814815e-05,
+      "loss": 1.7536,
+      "step": 44345
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.372918518518519e-05,
+      "loss": 1.6802,
+      "step": 44350
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3726222222222224e-05,
+      "loss": 1.6843,
+      "step": 44355
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3723259259259263e-05,
+      "loss": 1.6141,
+      "step": 44360
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.37202962962963e-05,
+      "loss": 1.6366,
+      "step": 44365
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3717333333333334e-05,
+      "loss": 1.6647,
+      "step": 44370
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.371437037037037e-05,
+      "loss": 1.5831,
+      "step": 44375
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.371140740740741e-05,
+      "loss": 1.7328,
+      "step": 44380
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3708444444444444e-05,
+      "loss": 1.694,
+      "step": 44385
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3705481481481483e-05,
+      "loss": 1.7438,
+      "step": 44390
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3702518518518518e-05,
+      "loss": 1.5709,
+      "step": 44395
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3699555555555557e-05,
+      "loss": 1.7062,
+      "step": 44400
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.3696592592592592e-05,
+      "loss": 1.6691,
+      "step": 44405
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.369362962962963e-05,
+      "loss": 1.6627,
+      "step": 44410
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3690666666666667e-05,
+      "loss": 1.5864,
+      "step": 44415
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3687703703703702e-05,
+      "loss": 1.7458,
+      "step": 44420
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.368474074074074e-05,
+      "loss": 1.6351,
+      "step": 44425
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3681777777777776e-05,
+      "loss": 1.7145,
+      "step": 44430
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3678814814814815e-05,
+      "loss": 1.7232,
+      "step": 44435
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.367585185185185e-05,
+      "loss": 1.6018,
+      "step": 44440
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.367288888888889e-05,
+      "loss": 1.7669,
+      "step": 44445
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3669925925925925e-05,
+      "loss": 1.8388,
+      "step": 44450
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3666962962962964e-05,
+      "loss": 1.6238,
+      "step": 44455
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3664e-05,
+      "loss": 1.5801,
+      "step": 44460
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.3661037037037038e-05,
+      "loss": 1.7535,
+      "step": 44465
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3658074074074073e-05,
+      "loss": 1.7648,
+      "step": 44470
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3655111111111112e-05,
+      "loss": 1.6927,
+      "step": 44475
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3652148148148148e-05,
+      "loss": 1.6763,
+      "step": 44480
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3649185185185186e-05,
+      "loss": 1.6979,
+      "step": 44485
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3646222222222222e-05,
+      "loss": 1.5611,
+      "step": 44490
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.364325925925926e-05,
+      "loss": 1.6492,
+      "step": 44495
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3640296296296296e-05,
+      "loss": 1.6176,
+      "step": 44500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3637333333333335e-05,
+      "loss": 1.7075,
+      "step": 44505
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.363437037037037e-05,
+      "loss": 1.6913,
+      "step": 44510
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.363140740740741e-05,
+      "loss": 1.687,
+      "step": 44515
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.3628444444444445e-05,
+      "loss": 1.6182,
+      "step": 44520
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3625481481481483e-05,
+      "loss": 1.5335,
+      "step": 44525
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.362251851851852e-05,
+      "loss": 1.4918,
+      "step": 44530
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3619555555555554e-05,
+      "loss": 1.682,
+      "step": 44535
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3616592592592593e-05,
+      "loss": 1.7019,
+      "step": 44540
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.361362962962963e-05,
+      "loss": 1.7195,
+      "step": 44545
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3610666666666667e-05,
+      "loss": 1.5996,
+      "step": 44550
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3607703703703703e-05,
+      "loss": 1.5595,
+      "step": 44555
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.360474074074074e-05,
+      "loss": 1.6246,
+      "step": 44560
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3601777777777777e-05,
+      "loss": 1.5444,
+      "step": 44565
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.3598814814814816e-05,
+      "loss": 1.6714,
+      "step": 44570
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.359585185185185e-05,
+      "loss": 1.6278,
+      "step": 44575
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.359288888888889e-05,
+      "loss": 1.6185,
+      "step": 44580
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3589925925925926e-05,
+      "loss": 1.6117,
+      "step": 44585
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3586962962962964e-05,
+      "loss": 1.6647,
+      "step": 44590
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3584e-05,
+      "loss": 1.6413,
+      "step": 44595
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.358103703703704e-05,
+      "loss": 1.7219,
+      "step": 44600
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3578074074074074e-05,
+      "loss": 1.5731,
+      "step": 44605
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3575111111111113e-05,
+      "loss": 1.6878,
+      "step": 44610
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.357214814814815e-05,
+      "loss": 1.579,
+      "step": 44615
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3569185185185187e-05,
+      "loss": 1.5995,
+      "step": 44620
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.3566222222222223e-05,
+      "loss": 1.6012,
+      "step": 44625
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.356325925925926e-05,
+      "loss": 1.6608,
+      "step": 44630
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3560296296296297e-05,
+      "loss": 1.482,
+      "step": 44635
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3557333333333336e-05,
+      "loss": 1.5583,
+      "step": 44640
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.355437037037037e-05,
+      "loss": 1.7036,
+      "step": 44645
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3551407407407407e-05,
+      "loss": 1.6458,
+      "step": 44650
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3548444444444445e-05,
+      "loss": 1.6258,
+      "step": 44655
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.354548148148148e-05,
+      "loss": 1.6745,
+      "step": 44660
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.354251851851852e-05,
+      "loss": 1.6146,
+      "step": 44665
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3539555555555555e-05,
+      "loss": 1.7411,
+      "step": 44670
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3536592592592594e-05,
+      "loss": 1.7513,
+      "step": 44675
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.353362962962963e-05,
+      "loss": 1.6785,
+      "step": 44680
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3530666666666668e-05,
+      "loss": 1.653,
+      "step": 44685
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.3527703703703704e-05,
+      "loss": 1.6472,
+      "step": 44690
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3524740740740742e-05,
+      "loss": 1.6399,
+      "step": 44695
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3521777777777778e-05,
+      "loss": 1.4768,
+      "step": 44700
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3518814814814817e-05,
+      "loss": 1.7246,
+      "step": 44705
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3515851851851852e-05,
+      "loss": 1.6618,
+      "step": 44710
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.351288888888889e-05,
+      "loss": 1.5396,
+      "step": 44715
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3509925925925926e-05,
+      "loss": 1.6135,
+      "step": 44720
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3506962962962965e-05,
+      "loss": 1.6407,
+      "step": 44725
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3504e-05,
+      "loss": 1.7406,
+      "step": 44730
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.350103703703704e-05,
+      "loss": 1.7026,
+      "step": 44735
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3498074074074075e-05,
+      "loss": 1.643,
+      "step": 44740
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.3495111111111114e-05,
+      "loss": 1.5974,
+      "step": 44745
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.349214814814815e-05,
+      "loss": 1.6868,
+      "step": 44750
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3489185185185188e-05,
+      "loss": 1.8256,
+      "step": 44755
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3486222222222223e-05,
+      "loss": 1.6032,
+      "step": 44760
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.348325925925926e-05,
+      "loss": 1.639,
+      "step": 44765
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3480296296296298e-05,
+      "loss": 1.745,
+      "step": 44770
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3477333333333333e-05,
+      "loss": 1.6011,
+      "step": 44775
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3474370370370372e-05,
+      "loss": 1.6498,
+      "step": 44780
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3471407407407407e-05,
+      "loss": 1.6198,
+      "step": 44785
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.3468444444444446e-05,
+      "loss": 1.6809,
+      "step": 44790
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.346548148148148e-05,
+      "loss": 1.6093,
+      "step": 44795
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.346251851851852e-05,
+      "loss": 1.6004,
+      "step": 44800
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3459555555555556e-05,
+      "loss": 1.7816,
+      "step": 44805
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3456592592592595e-05,
+      "loss": 1.7337,
+      "step": 44810
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.345362962962963e-05,
+      "loss": 1.531,
+      "step": 44815
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.345066666666667e-05,
+      "loss": 1.5797,
+      "step": 44820
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3447703703703704e-05,
+      "loss": 1.6766,
+      "step": 44825
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3444740740740743e-05,
+      "loss": 1.642,
+      "step": 44830
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.344177777777778e-05,
+      "loss": 1.5744,
+      "step": 44835
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3438814814814817e-05,
+      "loss": 1.662,
+      "step": 44840
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3435851851851853e-05,
+      "loss": 1.6339,
+      "step": 44845
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.343288888888889e-05,
+      "loss": 1.6259,
+      "step": 44850
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.3429925925925927e-05,
+      "loss": 1.7587,
+      "step": 44855
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3426962962962966e-05,
+      "loss": 1.6832,
+      "step": 44860
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3424e-05,
+      "loss": 1.592,
+      "step": 44865
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.342103703703704e-05,
+      "loss": 1.5972,
+      "step": 44870
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3418074074074076e-05,
+      "loss": 1.6299,
+      "step": 44875
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.341511111111111e-05,
+      "loss": 1.5961,
+      "step": 44880
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.341214814814815e-05,
+      "loss": 1.6918,
+      "step": 44885
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3409185185185185e-05,
+      "loss": 1.7329,
+      "step": 44890
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3406222222222224e-05,
+      "loss": 1.6431,
+      "step": 44895
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.340325925925926e-05,
+      "loss": 1.5081,
+      "step": 44900
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.34002962962963e-05,
+      "loss": 1.6562,
+      "step": 44905
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3397333333333334e-05,
+      "loss": 1.6081,
+      "step": 44910
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.3394370370370373e-05,
+      "loss": 1.5982,
+      "step": 44915
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3391407407407408e-05,
+      "loss": 1.5854,
+      "step": 44920
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3388444444444447e-05,
+      "loss": 1.7278,
+      "step": 44925
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3385481481481482e-05,
+      "loss": 1.7261,
+      "step": 44930
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.338251851851852e-05,
+      "loss": 1.6154,
+      "step": 44935
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3379555555555557e-05,
+      "loss": 1.6804,
+      "step": 44940
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3376592592592595e-05,
+      "loss": 1.7905,
+      "step": 44945
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.337362962962963e-05,
+      "loss": 1.6661,
+      "step": 44950
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.337066666666667e-05,
+      "loss": 1.594,
+      "step": 44955
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3367703703703705e-05,
+      "loss": 1.5908,
+      "step": 44960
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.3364740740740744e-05,
+      "loss": 1.6451,
+      "step": 44965
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.336177777777778e-05,
+      "loss": 1.6804,
+      "step": 44970
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3358814814814818e-05,
+      "loss": 1.6909,
+      "step": 44975
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3355851851851854e-05,
+      "loss": 1.7838,
+      "step": 44980
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3352888888888892e-05,
+      "loss": 1.7537,
+      "step": 44985
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3349925925925928e-05,
+      "loss": 1.7058,
+      "step": 44990
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3346962962962963e-05,
+      "loss": 1.6626,
+      "step": 44995
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3344000000000002e-05,
+      "loss": 1.6149,
+      "step": 45000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3341037037037038e-05,
+      "loss": 1.5319,
+      "step": 45005
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3338074074074076e-05,
+      "loss": 1.5372,
+      "step": 45010
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3335111111111112e-05,
+      "loss": 1.5558,
+      "step": 45015
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.333214814814815e-05,
+      "loss": 1.4722,
+      "step": 45020
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.3329185185185186e-05,
+      "loss": 1.6423,
+      "step": 45025
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3326222222222225e-05,
+      "loss": 1.5207,
+      "step": 45030
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.332325925925926e-05,
+      "loss": 1.4872,
+      "step": 45035
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.33202962962963e-05,
+      "loss": 1.5813,
+      "step": 45040
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3317333333333335e-05,
+      "loss": 1.6228,
+      "step": 45045
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3314370370370373e-05,
+      "loss": 1.4591,
+      "step": 45050
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.331140740740741e-05,
+      "loss": 1.5458,
+      "step": 45055
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3308444444444448e-05,
+      "loss": 1.6699,
+      "step": 45060
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3305481481481483e-05,
+      "loss": 1.5123,
+      "step": 45065
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3302518518518522e-05,
+      "loss": 1.5467,
+      "step": 45070
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3299555555555557e-05,
+      "loss": 1.5179,
+      "step": 45075
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.3296592592592596e-05,
+      "loss": 1.6134,
+      "step": 45080
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.329362962962963e-05,
+      "loss": 1.5591,
+      "step": 45085
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.329066666666667e-05,
+      "loss": 1.5257,
+      "step": 45090
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3287703703703706e-05,
+      "loss": 1.4926,
+      "step": 45095
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3284740740740745e-05,
+      "loss": 1.6152,
+      "step": 45100
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.328177777777778e-05,
+      "loss": 1.5089,
+      "step": 45105
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3278814814814816e-05,
+      "loss": 1.6092,
+      "step": 45110
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3275851851851854e-05,
+      "loss": 1.4785,
+      "step": 45115
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.327288888888889e-05,
+      "loss": 1.532,
+      "step": 45120
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.326992592592593e-05,
+      "loss": 1.4919,
+      "step": 45125
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3266962962962964e-05,
+      "loss": 1.5407,
+      "step": 45130
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.3264000000000003e-05,
+      "loss": 1.6088,
+      "step": 45135
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 2.326103703703704e-05,
+      "loss": 1.5379,
+      "step": 45140
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3258074074074077e-05,
+      "loss": 1.5356,
+      "step": 45145
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.325511111111111e-05,
+      "loss": 1.5655,
+      "step": 45150
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3252148148148148e-05,
+      "loss": 1.5902,
+      "step": 45155
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3249185185185183e-05,
+      "loss": 1.5304,
+      "step": 45160
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3246222222222222e-05,
+      "loss": 1.4247,
+      "step": 45165
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3243259259259258e-05,
+      "loss": 1.4893,
+      "step": 45170
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3240296296296297e-05,
+      "loss": 1.5992,
+      "step": 45175
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3237333333333332e-05,
+      "loss": 1.5439,
+      "step": 45180
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.323437037037037e-05,
+      "loss": 1.5346,
+      "step": 45185
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3231407407407406e-05,
+      "loss": 1.5502,
+      "step": 45190
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 2.3228444444444445e-05,
+      "loss": 1.6376,
+      "step": 45195
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.322548148148148e-05,
+      "loss": 1.6687,
+      "step": 45200
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.322251851851852e-05,
+      "loss": 1.5223,
+      "step": 45205
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3219555555555555e-05,
+      "loss": 1.5763,
+      "step": 45210
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3216592592592594e-05,
+      "loss": 1.5428,
+      "step": 45215
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.321362962962963e-05,
+      "loss": 1.455,
+      "step": 45220
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3210666666666668e-05,
+      "loss": 1.5248,
+      "step": 45225
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3207703703703703e-05,
+      "loss": 1.5887,
+      "step": 45230
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3204740740740742e-05,
+      "loss": 1.5292,
+      "step": 45235
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3201777777777778e-05,
+      "loss": 1.5325,
+      "step": 45240
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3198814814814816e-05,
+      "loss": 1.5418,
+      "step": 45245
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 2.3195851851851852e-05,
+      "loss": 1.563,
+      "step": 45250
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.319288888888889e-05,
+      "loss": 1.6303,
+      "step": 45255
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3189925925925926e-05,
+      "loss": 1.6107,
+      "step": 45260
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.318696296296296e-05,
+      "loss": 1.5798,
+      "step": 45265
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3184e-05,
+      "loss": 1.6505,
+      "step": 45270
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3181037037037036e-05,
+      "loss": 1.5395,
+      "step": 45275
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3178074074074075e-05,
+      "loss": 1.624,
+      "step": 45280
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.317511111111111e-05,
+      "loss": 1.5726,
+      "step": 45285
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.317214814814815e-05,
+      "loss": 1.7139,
+      "step": 45290
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3169185185185184e-05,
+      "loss": 1.4561,
+      "step": 45295
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.3166222222222223e-05,
+      "loss": 1.503,
+      "step": 45300
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 2.316325925925926e-05,
+      "loss": 1.5384,
+      "step": 45305
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3160296296296297e-05,
+      "loss": 1.5321,
+      "step": 45310
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3157333333333333e-05,
+      "loss": 1.6034,
+      "step": 45315
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.315437037037037e-05,
+      "loss": 1.4981,
+      "step": 45320
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3151407407407407e-05,
+      "loss": 1.6399,
+      "step": 45325
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3148444444444446e-05,
+      "loss": 1.581,
+      "step": 45330
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.314548148148148e-05,
+      "loss": 1.6299,
+      "step": 45335
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.314251851851852e-05,
+      "loss": 1.5266,
+      "step": 45340
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3139555555555556e-05,
+      "loss": 1.4729,
+      "step": 45345
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3136592592592594e-05,
+      "loss": 1.4905,
+      "step": 45350
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.313362962962963e-05,
+      "loss": 1.6046,
+      "step": 45355
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.313066666666667e-05,
+      "loss": 1.5316,
+      "step": 45360
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 2.3127703703703704e-05,
+      "loss": 1.3822,
+      "step": 45365
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3124740740740743e-05,
+      "loss": 1.5819,
+      "step": 45370
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3121777777777778e-05,
+      "loss": 1.4971,
+      "step": 45375
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3118814814814814e-05,
+      "loss": 1.7153,
+      "step": 45380
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3115851851851853e-05,
+      "loss": 1.5087,
+      "step": 45385
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3112888888888888e-05,
+      "loss": 1.5463,
+      "step": 45390
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3109925925925927e-05,
+      "loss": 1.586,
+      "step": 45395
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3106962962962962e-05,
+      "loss": 1.5333,
+      "step": 45400
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3104e-05,
+      "loss": 1.6281,
+      "step": 45405
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3101037037037036e-05,
+      "loss": 1.6642,
+      "step": 45410
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.3098074074074075e-05,
+      "loss": 1.5242,
+      "step": 45415
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 2.309511111111111e-05,
+      "loss": 1.6828,
+      "step": 45420
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.309214814814815e-05,
+      "loss": 1.6406,
+      "step": 45425
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3089185185185185e-05,
+      "loss": 1.6344,
+      "step": 45430
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3086222222222224e-05,
+      "loss": 1.6432,
+      "step": 45435
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.308325925925926e-05,
+      "loss": 1.5434,
+      "step": 45440
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3080296296296298e-05,
+      "loss": 1.5959,
+      "step": 45445
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3077333333333334e-05,
+      "loss": 1.662,
+      "step": 45450
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3074370370370372e-05,
+      "loss": 1.5361,
+      "step": 45455
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3071407407407408e-05,
+      "loss": 1.4627,
+      "step": 45460
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3068444444444447e-05,
+      "loss": 1.4907,
+      "step": 45465
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.3065481481481482e-05,
+      "loss": 1.4775,
+      "step": 45470
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 2.306251851851852e-05,
+      "loss": 1.5532,
+      "step": 45475
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3059555555555556e-05,
+      "loss": 1.5261,
+      "step": 45480
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3056592592592595e-05,
+      "loss": 1.5028,
+      "step": 45485
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.305362962962963e-05,
+      "loss": 1.5896,
+      "step": 45490
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3050666666666666e-05,
+      "loss": 1.5869,
+      "step": 45495
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3047703703703705e-05,
+      "loss": 1.5967,
+      "step": 45500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.304474074074074e-05,
+      "loss": 1.5541,
+      "step": 45505
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.304177777777778e-05,
+      "loss": 1.4754,
+      "step": 45510
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3038814814814814e-05,
+      "loss": 1.5697,
+      "step": 45515
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3035851851851853e-05,
+      "loss": 1.5145,
+      "step": 45520
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.303288888888889e-05,
+      "loss": 1.5985,
+      "step": 45525
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 2.3029925925925928e-05,
+      "loss": 1.4577,
+      "step": 45530
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3026962962962963e-05,
+      "loss": 1.4819,
+      "step": 45535
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3024000000000002e-05,
+      "loss": 1.5791,
+      "step": 45540
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3021037037037037e-05,
+      "loss": 1.6249,
+      "step": 45545
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3018074074074076e-05,
+      "loss": 1.4333,
+      "step": 45550
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.301511111111111e-05,
+      "loss": 1.6504,
+      "step": 45555
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.301214814814815e-05,
+      "loss": 1.5436,
+      "step": 45560
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3009185185185186e-05,
+      "loss": 1.6394,
+      "step": 45565
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.3006222222222225e-05,
+      "loss": 1.4945,
+      "step": 45570
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.300325925925926e-05,
+      "loss": 1.658,
+      "step": 45575
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.30002962962963e-05,
+      "loss": 1.4466,
+      "step": 45580
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.2997333333333334e-05,
+      "loss": 1.4792,
+      "step": 45585
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 2.2994370370370373e-05,
+      "loss": 1.5259,
+      "step": 45590
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.299140740740741e-05,
+      "loss": 1.5962,
+      "step": 45595
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2988444444444447e-05,
+      "loss": 1.5246,
+      "step": 45600
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2985481481481483e-05,
+      "loss": 1.516,
+      "step": 45605
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2982518518518518e-05,
+      "loss": 1.575,
+      "step": 45610
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2979555555555557e-05,
+      "loss": 1.6239,
+      "step": 45615
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2976592592592592e-05,
+      "loss": 1.4821,
+      "step": 45620
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.297362962962963e-05,
+      "loss": 1.5263,
+      "step": 45625
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2970666666666667e-05,
+      "loss": 1.518,
+      "step": 45630
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.2967703703703706e-05,
+      "loss": 1.4405,
+      "step": 45635
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.296474074074074e-05,
+      "loss": 1.5605,
+      "step": 45640
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 2.296177777777778e-05,
+      "loss": 1.5732,
+      "step": 45645
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2958814814814815e-05,
+      "loss": 1.4171,
+      "step": 45650
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2955851851851854e-05,
+      "loss": 1.593,
+      "step": 45655
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.295288888888889e-05,
+      "loss": 1.5186,
+      "step": 45660
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.294992592592593e-05,
+      "loss": 1.5421,
+      "step": 45665
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2946962962962964e-05,
+      "loss": 1.5188,
+      "step": 45670
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2944000000000003e-05,
+      "loss": 1.5741,
+      "step": 45675
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2941037037037038e-05,
+      "loss": 1.6651,
+      "step": 45680
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2938074074074077e-05,
+      "loss": 1.5427,
+      "step": 45685
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2935111111111112e-05,
+      "loss": 1.4956,
+      "step": 45690
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.293214814814815e-05,
+      "loss": 1.554,
+      "step": 45695
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 2.2929185185185187e-05,
+      "loss": 1.4999,
+      "step": 45700
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2926222222222225e-05,
+      "loss": 1.497,
+      "step": 45705
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.292325925925926e-05,
+      "loss": 1.6863,
+      "step": 45710
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.29202962962963e-05,
+      "loss": 1.5291,
+      "step": 45715
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2917333333333335e-05,
+      "loss": 1.4117,
+      "step": 45720
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.291437037037037e-05,
+      "loss": 1.6449,
+      "step": 45725
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.291140740740741e-05,
+      "loss": 1.5422,
+      "step": 45730
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2908444444444445e-05,
+      "loss": 1.582,
+      "step": 45735
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2905481481481484e-05,
+      "loss": 1.5755,
+      "step": 45740
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.290251851851852e-05,
+      "loss": 1.5205,
+      "step": 45745
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2899555555555558e-05,
+      "loss": 1.5642,
+      "step": 45750
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 2.2896592592592593e-05,
+      "loss": 1.4762,
+      "step": 45755
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2893629629629632e-05,
+      "loss": 1.6037,
+      "step": 45760
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2890666666666668e-05,
+      "loss": 1.4866,
+      "step": 45765
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2887703703703706e-05,
+      "loss": 1.5811,
+      "step": 45770
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2884740740740742e-05,
+      "loss": 1.5323,
+      "step": 45775
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.288177777777778e-05,
+      "loss": 1.672,
+      "step": 45780
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2878814814814816e-05,
+      "loss": 1.598,
+      "step": 45785
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2875851851851855e-05,
+      "loss": 1.5632,
+      "step": 45790
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.287288888888889e-05,
+      "loss": 1.475,
+      "step": 45795
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.286992592592593e-05,
+      "loss": 1.5667,
+      "step": 45800
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2866962962962965e-05,
+      "loss": 1.5672,
+      "step": 45805
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.2864000000000003e-05,
+      "loss": 1.4992,
+      "step": 45810
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 2.286103703703704e-05,
+      "loss": 1.7117,
+      "step": 45815
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2858074074074078e-05,
+      "loss": 1.6201,
+      "step": 45820
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2855111111111113e-05,
+      "loss": 1.5594,
+      "step": 45825
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2852148148148152e-05,
+      "loss": 1.5995,
+      "step": 45830
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2849185185185187e-05,
+      "loss": 1.5737,
+      "step": 45835
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2846222222222223e-05,
+      "loss": 1.5068,
+      "step": 45840
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.284325925925926e-05,
+      "loss": 1.5644,
+      "step": 45845
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2840296296296297e-05,
+      "loss": 1.4801,
+      "step": 45850
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2837333333333336e-05,
+      "loss": 1.5289,
+      "step": 45855
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.283437037037037e-05,
+      "loss": 1.6561,
+      "step": 45860
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.283140740740741e-05,
+      "loss": 1.562,
+      "step": 45865
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 2.2828444444444446e-05,
+      "loss": 1.4998,
+      "step": 45870
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2825481481481484e-05,
+      "loss": 1.5373,
+      "step": 45875
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.282251851851852e-05,
+      "loss": 1.4284,
+      "step": 45880
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.281955555555556e-05,
+      "loss": 1.474,
+      "step": 45885
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2816592592592594e-05,
+      "loss": 1.5559,
+      "step": 45890
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2813629629629633e-05,
+      "loss": 1.5984,
+      "step": 45895
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2810666666666668e-05,
+      "loss": 1.5282,
+      "step": 45900
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2807703703703707e-05,
+      "loss": 1.5808,
+      "step": 45905
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2804740740740743e-05,
+      "loss": 1.5109,
+      "step": 45910
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.280177777777778e-05,
+      "loss": 1.5597,
+      "step": 45915
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2798814814814817e-05,
+      "loss": 1.4594,
+      "step": 45920
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 2.2795851851851852e-05,
+      "loss": 1.5634,
+      "step": 45925
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2792888888888888e-05,
+      "loss": 1.6347,
+      "step": 45930
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2789925925925926e-05,
+      "loss": 1.6286,
+      "step": 45935
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2786962962962962e-05,
+      "loss": 1.4358,
+      "step": 45940
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2784e-05,
+      "loss": 1.5639,
+      "step": 45945
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2781037037037036e-05,
+      "loss": 1.5817,
+      "step": 45950
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2778074074074075e-05,
+      "loss": 1.5448,
+      "step": 45955
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.277511111111111e-05,
+      "loss": 1.4816,
+      "step": 45960
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.277214814814815e-05,
+      "loss": 1.5953,
+      "step": 45965
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2769185185185185e-05,
+      "loss": 1.5311,
+      "step": 45970
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.2766222222222223e-05,
+      "loss": 1.6437,
+      "step": 45975
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 2.276325925925926e-05,
+      "loss": 1.4537,
+      "step": 45980
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2760296296296298e-05,
+      "loss": 1.5871,
+      "step": 45985
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2757333333333333e-05,
+      "loss": 1.631,
+      "step": 45990
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2754370370370372e-05,
+      "loss": 1.5192,
+      "step": 45995
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2751407407407407e-05,
+      "loss": 1.6275,
+      "step": 46000
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2748444444444443e-05,
+      "loss": 1.5853,
+      "step": 46005
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2745481481481482e-05,
+      "loss": 1.6075,
+      "step": 46010
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2742518518518517e-05,
+      "loss": 1.5914,
+      "step": 46015
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2739555555555556e-05,
+      "loss": 1.5168,
+      "step": 46020
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.273659259259259e-05,
+      "loss": 1.3981,
+      "step": 46025
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.273362962962963e-05,
+      "loss": 1.4735,
+      "step": 46030
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2730666666666666e-05,
+      "loss": 1.6041,
+      "step": 46035
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 2.2727703703703704e-05,
+      "loss": 1.5408,
+      "step": 46040
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.272474074074074e-05,
+      "loss": 1.6497,
+      "step": 46045
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.272177777777778e-05,
+      "loss": 1.4999,
+      "step": 46050
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2718814814814814e-05,
+      "loss": 1.478,
+      "step": 46055
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2715851851851853e-05,
+      "loss": 1.516,
+      "step": 46060
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.271288888888889e-05,
+      "loss": 1.4681,
+      "step": 46065
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2709925925925927e-05,
+      "loss": 1.5662,
+      "step": 46070
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2706962962962963e-05,
+      "loss": 1.5806,
+      "step": 46075
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2704e-05,
+      "loss": 1.4732,
+      "step": 46080
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2701037037037037e-05,
+      "loss": 1.4605,
+      "step": 46085
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.2698074074074076e-05,
+      "loss": 1.6052,
+      "step": 46090
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 2.269511111111111e-05,
+      "loss": 1.5897,
+      "step": 46095
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.269214814814815e-05,
+      "loss": 1.659,
+      "step": 46100
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2689185185185185e-05,
+      "loss": 1.4688,
+      "step": 46105
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2686222222222224e-05,
+      "loss": 1.4643,
+      "step": 46110
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.268325925925926e-05,
+      "loss": 1.5818,
+      "step": 46115
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2680296296296295e-05,
+      "loss": 1.6606,
+      "step": 46120
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2677333333333334e-05,
+      "loss": 1.5478,
+      "step": 46125
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.267437037037037e-05,
+      "loss": 1.6057,
+      "step": 46130
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2671407407407408e-05,
+      "loss": 1.5544,
+      "step": 46135
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2668444444444444e-05,
+      "loss": 1.5777,
+      "step": 46140
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2665481481481482e-05,
+      "loss": 1.4771,
+      "step": 46145
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 2.2662518518518518e-05,
+      "loss": 1.5987,
+      "step": 46150
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2659555555555557e-05,
+      "loss": 1.4826,
+      "step": 46155
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2656592592592592e-05,
+      "loss": 1.5478,
+      "step": 46160
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.265362962962963e-05,
+      "loss": 1.6778,
+      "step": 46165
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2650666666666666e-05,
+      "loss": 1.6293,
+      "step": 46170
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2647703703703705e-05,
+      "loss": 1.6275,
+      "step": 46175
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.264474074074074e-05,
+      "loss": 1.5051,
+      "step": 46180
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.264177777777778e-05,
+      "loss": 1.5817,
+      "step": 46185
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2638814814814815e-05,
+      "loss": 1.7326,
+      "step": 46190
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2635851851851854e-05,
+      "loss": 1.6832,
+      "step": 46195
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.263288888888889e-05,
+      "loss": 1.4561,
+      "step": 46200
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 2.2629925925925928e-05,
+      "loss": 1.5981,
+      "step": 46205
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2626962962962963e-05,
+      "loss": 1.5215,
+      "step": 46210
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2624000000000002e-05,
+      "loss": 1.5888,
+      "step": 46215
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2621037037037038e-05,
+      "loss": 1.6355,
+      "step": 46220
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2618074074074077e-05,
+      "loss": 1.5591,
+      "step": 46225
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2615111111111112e-05,
+      "loss": 1.6067,
+      "step": 46230
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2612148148148147e-05,
+      "loss": 1.6062,
+      "step": 46235
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2609185185185186e-05,
+      "loss": 1.6973,
+      "step": 46240
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.260622222222222e-05,
+      "loss": 1.4996,
+      "step": 46245
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.260325925925926e-05,
+      "loss": 1.5403,
+      "step": 46250
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2600296296296296e-05,
+      "loss": 1.5228,
+      "step": 46255
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.2597333333333335e-05,
+      "loss": 1.472,
+      "step": 46260
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 2.259437037037037e-05,
+      "loss": 1.6462,
+      "step": 46265
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.259140740740741e-05,
+      "loss": 1.5733,
+      "step": 46270
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2588444444444444e-05,
+      "loss": 1.6584,
+      "step": 46275
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2585481481481483e-05,
+      "loss": 1.5607,
+      "step": 46280
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.258251851851852e-05,
+      "loss": 1.5951,
+      "step": 46285
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2579555555555557e-05,
+      "loss": 1.5419,
+      "step": 46290
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2576592592592593e-05,
+      "loss": 1.518,
+      "step": 46295
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2573629629629632e-05,
+      "loss": 1.4671,
+      "step": 46300
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2570666666666667e-05,
+      "loss": 1.6407,
+      "step": 46305
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.2567703703703706e-05,
+      "loss": 1.5801,
+      "step": 46310
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.256474074074074e-05,
+      "loss": 1.5516,
+      "step": 46315
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 2.256177777777778e-05,
+      "loss": 1.5671,
+      "step": 46320
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2558814814814816e-05,
+      "loss": 1.5578,
+      "step": 46325
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2555851851851855e-05,
+      "loss": 1.715,
+      "step": 46330
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.255288888888889e-05,
+      "loss": 1.477,
+      "step": 46335
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.254992592592593e-05,
+      "loss": 1.6092,
+      "step": 46340
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2546962962962964e-05,
+      "loss": 1.5662,
+      "step": 46345
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2544e-05,
+      "loss": 1.4304,
+      "step": 46350
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.254103703703704e-05,
+      "loss": 1.4583,
+      "step": 46355
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2538074074074074e-05,
+      "loss": 1.5449,
+      "step": 46360
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2535111111111113e-05,
+      "loss": 1.5556,
+      "step": 46365
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2532148148148148e-05,
+      "loss": 1.5434,
+      "step": 46370
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 2.2529185185185187e-05,
+      "loss": 1.5861,
+      "step": 46375
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2526222222222222e-05,
+      "loss": 1.6547,
+      "step": 46380
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.252325925925926e-05,
+      "loss": 1.5264,
+      "step": 46385
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2520296296296297e-05,
+      "loss": 1.5013,
+      "step": 46390
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2517333333333335e-05,
+      "loss": 1.5383,
+      "step": 46395
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.251437037037037e-05,
+      "loss": 1.5927,
+      "step": 46400
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.251140740740741e-05,
+      "loss": 1.6724,
+      "step": 46405
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2508444444444445e-05,
+      "loss": 1.6415,
+      "step": 46410
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2505481481481484e-05,
+      "loss": 1.6483,
+      "step": 46415
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.250251851851852e-05,
+      "loss": 1.5451,
+      "step": 46420
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2499555555555558e-05,
+      "loss": 1.6542,
+      "step": 46425
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 2.2496592592592594e-05,
+      "loss": 1.532,
+      "step": 46430
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2493629629629633e-05,
+      "loss": 1.5155,
+      "step": 46435
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2490666666666668e-05,
+      "loss": 1.5673,
+      "step": 46440
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2487703703703707e-05,
+      "loss": 1.5198,
+      "step": 46445
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2484740740740742e-05,
+      "loss": 1.5584,
+      "step": 46450
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.248177777777778e-05,
+      "loss": 1.5733,
+      "step": 46455
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2478814814814816e-05,
+      "loss": 1.4792,
+      "step": 46460
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2475851851851852e-05,
+      "loss": 1.4191,
+      "step": 46465
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.247288888888889e-05,
+      "loss": 1.6312,
+      "step": 46470
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2469925925925926e-05,
+      "loss": 1.5587,
+      "step": 46475
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2466962962962965e-05,
+      "loss": 1.6761,
+      "step": 46480
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.2464e-05,
+      "loss": 1.5225,
+      "step": 46485
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 2.246103703703704e-05,
+      "loss": 1.5518,
+      "step": 46490
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2458074074074075e-05,
+      "loss": 1.625,
+      "step": 46495
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2455111111111113e-05,
+      "loss": 1.5233,
+      "step": 46500
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.245214814814815e-05,
+      "loss": 1.5217,
+      "step": 46505
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2449185185185188e-05,
+      "loss": 1.7396,
+      "step": 46510
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2446222222222223e-05,
+      "loss": 1.6406,
+      "step": 46515
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2443259259259262e-05,
+      "loss": 1.6327,
+      "step": 46520
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2440296296296297e-05,
+      "loss": 1.5149,
+      "step": 46525
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2437333333333336e-05,
+      "loss": 1.5056,
+      "step": 46530
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.243437037037037e-05,
+      "loss": 1.5806,
+      "step": 46535
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.243140740740741e-05,
+      "loss": 1.6433,
+      "step": 46540
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 2.2428444444444446e-05,
+      "loss": 1.5627,
+      "step": 46545
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2425481481481485e-05,
+      "loss": 1.5305,
+      "step": 46550
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.242251851851852e-05,
+      "loss": 1.5145,
+      "step": 46555
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.241955555555556e-05,
+      "loss": 1.6992,
+      "step": 46560
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2416592592592594e-05,
+      "loss": 1.5639,
+      "step": 46565
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2413629629629633e-05,
+      "loss": 1.6389,
+      "step": 46570
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.241066666666667e-05,
+      "loss": 1.7074,
+      "step": 46575
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2407703703703704e-05,
+      "loss": 1.5138,
+      "step": 46580
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2404740740740743e-05,
+      "loss": 1.6698,
+      "step": 46585
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.240177777777778e-05,
+      "loss": 1.5069,
+      "step": 46590
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2398814814814817e-05,
+      "loss": 1.5698,
+      "step": 46595
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 2.2395851851851853e-05,
+      "loss": 1.6,
+      "step": 46600
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.239288888888889e-05,
+      "loss": 1.5743,
+      "step": 46605
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2389925925925927e-05,
+      "loss": 1.6339,
+      "step": 46610
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2386962962962966e-05,
+      "loss": 1.5836,
+      "step": 46615
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2384e-05,
+      "loss": 1.6186,
+      "step": 46620
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.238103703703704e-05,
+      "loss": 1.5696,
+      "step": 46625
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2378074074074075e-05,
+      "loss": 1.7069,
+      "step": 46630
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2375111111111114e-05,
+      "loss": 1.5354,
+      "step": 46635
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.237214814814815e-05,
+      "loss": 1.549,
+      "step": 46640
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.236918518518519e-05,
+      "loss": 1.5874,
+      "step": 46645
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2366222222222224e-05,
+      "loss": 1.5482,
+      "step": 46650
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 2.2363259259259263e-05,
+      "loss": 1.5749,
+      "step": 46655
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2360296296296298e-05,
+      "loss": 1.649,
+      "step": 46660
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2357333333333337e-05,
+      "loss": 1.461,
+      "step": 46665
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2354370370370372e-05,
+      "loss": 1.6105,
+      "step": 46670
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.235140740740741e-05,
+      "loss": 1.616,
+      "step": 46675
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2348444444444447e-05,
+      "loss": 1.3968,
+      "step": 46680
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2345481481481486e-05,
+      "loss": 1.585,
+      "step": 46685
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.234251851851852e-05,
+      "loss": 1.4973,
+      "step": 46690
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2339555555555556e-05,
+      "loss": 1.5986,
+      "step": 46695
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2336592592592592e-05,
+      "loss": 1.4107,
+      "step": 46700
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.233362962962963e-05,
+      "loss": 1.621,
+      "step": 46705
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2330666666666666e-05,
+      "loss": 1.484,
+      "step": 46710
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 2.2327703703703705e-05,
+      "loss": 1.6283,
+      "step": 46715
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.232474074074074e-05,
+      "loss": 1.6551,
+      "step": 46720
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.232177777777778e-05,
+      "loss": 1.524,
+      "step": 46725
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2318814814814815e-05,
+      "loss": 1.488,
+      "step": 46730
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.231585185185185e-05,
+      "loss": 1.6253,
+      "step": 46735
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.231288888888889e-05,
+      "loss": 1.6196,
+      "step": 46740
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2309925925925924e-05,
+      "loss": 1.5229,
+      "step": 46745
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2306962962962963e-05,
+      "loss": 1.6012,
+      "step": 46750
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2304e-05,
+      "loss": 1.4397,
+      "step": 46755
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2301037037037037e-05,
+      "loss": 1.5493,
+      "step": 46760
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.2298074074074073e-05,
+      "loss": 1.6227,
+      "step": 46765
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 2.229511111111111e-05,
+      "loss": 1.5586,
+      "step": 46770
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2292148148148147e-05,
+      "loss": 1.4732,
+      "step": 46775
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2289185185185186e-05,
+      "loss": 1.5947,
+      "step": 46780
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.228622222222222e-05,
+      "loss": 1.7021,
+      "step": 46785
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.228325925925926e-05,
+      "loss": 1.5885,
+      "step": 46790
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2280296296296296e-05,
+      "loss": 1.4872,
+      "step": 46795
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2277333333333334e-05,
+      "loss": 1.6864,
+      "step": 46800
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.227437037037037e-05,
+      "loss": 1.5305,
+      "step": 46805
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.227140740740741e-05,
+      "loss": 1.524,
+      "step": 46810
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2268444444444444e-05,
+      "loss": 1.5166,
+      "step": 46815
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.2265481481481483e-05,
+      "loss": 1.6664,
+      "step": 46820
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 2.226251851851852e-05,
+      "loss": 1.4594,
+      "step": 46825
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2259555555555557e-05,
+      "loss": 1.5694,
+      "step": 46830
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2256592592592593e-05,
+      "loss": 1.5973,
+      "step": 46835
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.225362962962963e-05,
+      "loss": 1.6549,
+      "step": 46840
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2250666666666667e-05,
+      "loss": 1.5489,
+      "step": 46845
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2247703703703702e-05,
+      "loss": 1.5598,
+      "step": 46850
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.224474074074074e-05,
+      "loss": 1.6193,
+      "step": 46855
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2241777777777777e-05,
+      "loss": 1.3867,
+      "step": 46860
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2238814814814815e-05,
+      "loss": 1.6404,
+      "step": 46865
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.223585185185185e-05,
+      "loss": 1.4696,
+      "step": 46870
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.223288888888889e-05,
+      "loss": 1.7392,
+      "step": 46875
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 2.2229925925925925e-05,
+      "loss": 1.6278,
+      "step": 46880
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2226962962962964e-05,
+      "loss": 1.6228,
+      "step": 46885
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2224e-05,
+      "loss": 1.6901,
+      "step": 46890
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2221037037037038e-05,
+      "loss": 1.6052,
+      "step": 46895
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2218074074074074e-05,
+      "loss": 1.6018,
+      "step": 46900
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2215111111111112e-05,
+      "loss": 1.5572,
+      "step": 46905
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2212148148148148e-05,
+      "loss": 1.6287,
+      "step": 46910
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2209185185185187e-05,
+      "loss": 1.5911,
+      "step": 46915
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2206222222222222e-05,
+      "loss": 1.4826,
+      "step": 46920
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.220325925925926e-05,
+      "loss": 1.581,
+      "step": 46925
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2200296296296296e-05,
+      "loss": 1.5588,
+      "step": 46930
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.2197333333333335e-05,
+      "loss": 1.6054,
+      "step": 46935
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 2.219437037037037e-05,
+      "loss": 1.5383,
+      "step": 46940
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.219140740740741e-05,
+      "loss": 1.5289,
+      "step": 46945
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2188444444444445e-05,
+      "loss": 1.5351,
+      "step": 46950
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2185481481481484e-05,
+      "loss": 1.6189,
+      "step": 46955
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.218251851851852e-05,
+      "loss": 1.607,
+      "step": 46960
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2179555555555555e-05,
+      "loss": 1.615,
+      "step": 46965
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2176592592592593e-05,
+      "loss": 1.4924,
+      "step": 46970
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.217362962962963e-05,
+      "loss": 1.5961,
+      "step": 46975
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2170666666666668e-05,
+      "loss": 1.5244,
+      "step": 46980
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2167703703703703e-05,
+      "loss": 1.5813,
+      "step": 46985
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2164740740740742e-05,
+      "loss": 1.5784,
+      "step": 46990
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 2.2161777777777777e-05,
+      "loss": 1.5727,
+      "step": 46995
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2158814814814816e-05,
+      "loss": 1.5636,
+      "step": 47000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.215585185185185e-05,
+      "loss": 1.583,
+      "step": 47005
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.215288888888889e-05,
+      "loss": 1.4832,
+      "step": 47010
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2149925925925926e-05,
+      "loss": 1.5225,
+      "step": 47015
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2146962962962965e-05,
+      "loss": 1.4656,
+      "step": 47020
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2144e-05,
+      "loss": 1.571,
+      "step": 47025
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.214103703703704e-05,
+      "loss": 1.5867,
+      "step": 47030
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2138074074074074e-05,
+      "loss": 1.4768,
+      "step": 47035
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2135111111111113e-05,
+      "loss": 1.5745,
+      "step": 47040
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.213214814814815e-05,
+      "loss": 1.5118,
+      "step": 47045
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 2.2129185185185187e-05,
+      "loss": 1.548,
+      "step": 47050
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2126222222222223e-05,
+      "loss": 1.6263,
+      "step": 47055
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.212325925925926e-05,
+      "loss": 1.6887,
+      "step": 47060
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2120296296296297e-05,
+      "loss": 1.6418,
+      "step": 47065
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2117333333333336e-05,
+      "loss": 1.6527,
+      "step": 47070
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.211437037037037e-05,
+      "loss": 1.5866,
+      "step": 47075
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2111407407407407e-05,
+      "loss": 1.4501,
+      "step": 47080
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2108444444444446e-05,
+      "loss": 1.6031,
+      "step": 47085
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.210548148148148e-05,
+      "loss": 1.6453,
+      "step": 47090
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.210251851851852e-05,
+      "loss": 1.6063,
+      "step": 47095
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2099555555555555e-05,
+      "loss": 1.5585,
+      "step": 47100
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 2.2096592592592594e-05,
+      "loss": 1.5528,
+      "step": 47105
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.209362962962963e-05,
+      "loss": 1.5844,
+      "step": 47110
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.209066666666667e-05,
+      "loss": 1.6351,
+      "step": 47115
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2087703703703704e-05,
+      "loss": 1.5701,
+      "step": 47120
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2084740740740743e-05,
+      "loss": 1.6592,
+      "step": 47125
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2081777777777778e-05,
+      "loss": 1.5281,
+      "step": 47130
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2078814814814817e-05,
+      "loss": 1.5523,
+      "step": 47135
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2075851851851852e-05,
+      "loss": 1.5014,
+      "step": 47140
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.207288888888889e-05,
+      "loss": 1.5398,
+      "step": 47145
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2069925925925927e-05,
+      "loss": 1.6268,
+      "step": 47150
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2066962962962965e-05,
+      "loss": 1.4891,
+      "step": 47155
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.2064e-05,
+      "loss": 1.5367,
+      "step": 47160
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 2.206103703703704e-05,
+      "loss": 1.6897,
+      "step": 47165
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2058074074074075e-05,
+      "loss": 1.6075,
+      "step": 47170
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2055111111111114e-05,
+      "loss": 1.6416,
+      "step": 47175
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.205214814814815e-05,
+      "loss": 1.6006,
+      "step": 47180
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2049185185185188e-05,
+      "loss": 1.5777,
+      "step": 47185
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2046222222222224e-05,
+      "loss": 1.6031,
+      "step": 47190
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.204325925925926e-05,
+      "loss": 1.6258,
+      "step": 47195
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2040296296296298e-05,
+      "loss": 1.6051,
+      "step": 47200
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2037333333333333e-05,
+      "loss": 1.6267,
+      "step": 47205
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2034370370370372e-05,
+      "loss": 1.5893,
+      "step": 47210
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2031407407407408e-05,
+      "loss": 1.6302,
+      "step": 47215
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 2.2028444444444446e-05,
+      "loss": 1.6059,
+      "step": 47220
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2025481481481482e-05,
+      "loss": 1.5594,
+      "step": 47225
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.202251851851852e-05,
+      "loss": 1.6561,
+      "step": 47230
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2019555555555556e-05,
+      "loss": 1.5227,
+      "step": 47235
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2016592592592595e-05,
+      "loss": 1.5793,
+      "step": 47240
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.201362962962963e-05,
+      "loss": 1.5484,
+      "step": 47245
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.201066666666667e-05,
+      "loss": 1.5301,
+      "step": 47250
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2007703703703705e-05,
+      "loss": 1.5685,
+      "step": 47255
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.2004740740740743e-05,
+      "loss": 1.6943,
+      "step": 47260
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.200177777777778e-05,
+      "loss": 1.6373,
+      "step": 47265
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1998814814814818e-05,
+      "loss": 1.5516,
+      "step": 47270
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 2.1995851851851853e-05,
+      "loss": 1.6012,
+      "step": 47275
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1992888888888892e-05,
+      "loss": 1.4865,
+      "step": 47280
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1989925925925927e-05,
+      "loss": 1.5459,
+      "step": 47285
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1986962962962966e-05,
+      "loss": 1.5592,
+      "step": 47290
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1984e-05,
+      "loss": 1.5472,
+      "step": 47295
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.198103703703704e-05,
+      "loss": 1.4812,
+      "step": 47300
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1978074074074076e-05,
+      "loss": 1.5071,
+      "step": 47305
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.197511111111111e-05,
+      "loss": 1.5342,
+      "step": 47310
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.197214814814815e-05,
+      "loss": 1.3157,
+      "step": 47315
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1969185185185186e-05,
+      "loss": 1.6173,
+      "step": 47320
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.1966222222222224e-05,
+      "loss": 1.6164,
+      "step": 47325
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 2.196325925925926e-05,
+      "loss": 1.6466,
+      "step": 47330
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.19602962962963e-05,
+      "loss": 1.4472,
+      "step": 47335
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1957333333333334e-05,
+      "loss": 1.6237,
+      "step": 47340
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1954370370370373e-05,
+      "loss": 1.5209,
+      "step": 47345
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.195140740740741e-05,
+      "loss": 1.6338,
+      "step": 47350
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1948444444444447e-05,
+      "loss": 1.5769,
+      "step": 47355
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1945481481481483e-05,
+      "loss": 1.6905,
+      "step": 47360
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.194251851851852e-05,
+      "loss": 1.5773,
+      "step": 47365
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1939555555555557e-05,
+      "loss": 1.6069,
+      "step": 47370
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1936592592592596e-05,
+      "loss": 1.6848,
+      "step": 47375
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.193362962962963e-05,
+      "loss": 1.6073,
+      "step": 47380
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.193066666666667e-05,
+      "loss": 1.6285,
+      "step": 47385
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 2.1927703703703705e-05,
+      "loss": 1.575,
+      "step": 47390
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1924740740740744e-05,
+      "loss": 1.5736,
+      "step": 47395
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.192177777777778e-05,
+      "loss": 1.5025,
+      "step": 47400
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.191881481481482e-05,
+      "loss": 1.5521,
+      "step": 47405
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1915851851851854e-05,
+      "loss": 1.6113,
+      "step": 47410
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1912888888888893e-05,
+      "loss": 1.7129,
+      "step": 47415
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1909925925925928e-05,
+      "loss": 1.6043,
+      "step": 47420
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1906962962962964e-05,
+      "loss": 1.5439,
+      "step": 47425
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1904000000000002e-05,
+      "loss": 1.6108,
+      "step": 47430
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1901037037037038e-05,
+      "loss": 1.5353,
+      "step": 47435
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1898074074074077e-05,
+      "loss": 1.5844,
+      "step": 47440
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 2.1895111111111112e-05,
+      "loss": 1.5994,
+      "step": 47445
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.189214814814815e-05,
+      "loss": 1.6359,
+      "step": 47450
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1889185185185186e-05,
+      "loss": 1.5686,
+      "step": 47455
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1886222222222225e-05,
+      "loss": 1.6189,
+      "step": 47460
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.188325925925926e-05,
+      "loss": 1.5815,
+      "step": 47465
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1880296296296296e-05,
+      "loss": 1.687,
+      "step": 47470
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.187733333333333e-05,
+      "loss": 1.5858,
+      "step": 47475
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.187437037037037e-05,
+      "loss": 1.6059,
+      "step": 47480
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1871407407407406e-05,
+      "loss": 1.5492,
+      "step": 47485
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.1868444444444445e-05,
+      "loss": 1.5101,
+      "step": 47490
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.186548148148148e-05,
+      "loss": 1.6404,
+      "step": 47495
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 2.186251851851852e-05,
+      "loss": 1.6206,
+      "step": 47500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1859555555555554e-05,
+      "loss": 1.6501,
+      "step": 47505
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1856592592592593e-05,
+      "loss": 1.6834,
+      "step": 47510
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.185362962962963e-05,
+      "loss": 1.5677,
+      "step": 47515
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1850666666666667e-05,
+      "loss": 1.6211,
+      "step": 47520
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1847703703703703e-05,
+      "loss": 1.5655,
+      "step": 47525
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.184474074074074e-05,
+      "loss": 1.5535,
+      "step": 47530
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1841777777777777e-05,
+      "loss": 1.5823,
+      "step": 47535
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1838814814814816e-05,
+      "loss": 1.502,
+      "step": 47540
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.183585185185185e-05,
+      "loss": 1.5056,
+      "step": 47545
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.183288888888889e-05,
+      "loss": 1.5667,
+      "step": 47550
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 2.1829925925925926e-05,
+      "loss": 1.5553,
+      "step": 47555
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1826962962962964e-05,
+      "loss": 1.5297,
+      "step": 47560
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1824e-05,
+      "loss": 1.6318,
+      "step": 47565
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.182103703703704e-05,
+      "loss": 1.5296,
+      "step": 47570
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1818074074074074e-05,
+      "loss": 1.6722,
+      "step": 47575
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1815111111111113e-05,
+      "loss": 1.4001,
+      "step": 47580
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1812148148148148e-05,
+      "loss": 1.568,
+      "step": 47585
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1809185185185184e-05,
+      "loss": 1.4917,
+      "step": 47590
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1806222222222223e-05,
+      "loss": 1.5756,
+      "step": 47595
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1803259259259258e-05,
+      "loss": 1.5211,
+      "step": 47600
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1800296296296297e-05,
+      "loss": 1.6027,
+      "step": 47605
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.1797333333333332e-05,
+      "loss": 1.5355,
+      "step": 47610
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 2.179437037037037e-05,
+      "loss": 1.6005,
+      "step": 47615
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1791407407407406e-05,
+      "loss": 1.5999,
+      "step": 47620
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1788444444444445e-05,
+      "loss": 1.6756,
+      "step": 47625
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.178548148148148e-05,
+      "loss": 1.5878,
+      "step": 47630
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.178251851851852e-05,
+      "loss": 1.5147,
+      "step": 47635
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1779555555555555e-05,
+      "loss": 1.5881,
+      "step": 47640
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1776592592592594e-05,
+      "loss": 1.7187,
+      "step": 47645
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.177362962962963e-05,
+      "loss": 1.6254,
+      "step": 47650
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1770666666666668e-05,
+      "loss": 1.5287,
+      "step": 47655
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1767703703703703e-05,
+      "loss": 1.6405,
+      "step": 47660
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1764740740740742e-05,
+      "loss": 1.5883,
+      "step": 47665
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 2.1761777777777778e-05,
+      "loss": 1.5234,
+      "step": 47670
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1758814814814817e-05,
+      "loss": 1.5134,
+      "step": 47675
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1755851851851852e-05,
+      "loss": 1.6219,
+      "step": 47680
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.175288888888889e-05,
+      "loss": 1.6502,
+      "step": 47685
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1749925925925926e-05,
+      "loss": 1.5421,
+      "step": 47690
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1746962962962965e-05,
+      "loss": 1.6259,
+      "step": 47695
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1744e-05,
+      "loss": 1.6538,
+      "step": 47700
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1741037037037036e-05,
+      "loss": 1.5831,
+      "step": 47705
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1738074074074075e-05,
+      "loss": 1.5366,
+      "step": 47710
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.173511111111111e-05,
+      "loss": 1.6141,
+      "step": 47715
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.173214814814815e-05,
+      "loss": 1.6115,
+      "step": 47720
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 2.1729185185185184e-05,
+      "loss": 1.4801,
+      "step": 47725
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1726222222222223e-05,
+      "loss": 1.5836,
+      "step": 47730
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.172325925925926e-05,
+      "loss": 1.5643,
+      "step": 47735
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1720296296296298e-05,
+      "loss": 1.5182,
+      "step": 47740
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1717333333333333e-05,
+      "loss": 1.5676,
+      "step": 47745
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1714370370370372e-05,
+      "loss": 1.6412,
+      "step": 47750
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1711407407407407e-05,
+      "loss": 1.5194,
+      "step": 47755
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1708444444444446e-05,
+      "loss": 1.5391,
+      "step": 47760
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.170548148148148e-05,
+      "loss": 1.6565,
+      "step": 47765
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.170251851851852e-05,
+      "loss": 1.5634,
+      "step": 47770
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1699555555555556e-05,
+      "loss": 1.6943,
+      "step": 47775
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 2.1696592592592595e-05,
+      "loss": 1.5128,
+      "step": 47780
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.169362962962963e-05,
+      "loss": 1.5908,
+      "step": 47785
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.169066666666667e-05,
+      "loss": 1.5933,
+      "step": 47790
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1687703703703704e-05,
+      "loss": 1.5426,
+      "step": 47795
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1684740740740743e-05,
+      "loss": 1.6647,
+      "step": 47800
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.168177777777778e-05,
+      "loss": 1.644,
+      "step": 47805
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1678814814814817e-05,
+      "loss": 1.6176,
+      "step": 47810
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1675851851851853e-05,
+      "loss": 1.5917,
+      "step": 47815
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1672888888888888e-05,
+      "loss": 1.5366,
+      "step": 47820
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1669925925925927e-05,
+      "loss": 1.5441,
+      "step": 47825
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1666962962962962e-05,
+      "loss": 1.5943,
+      "step": 47830
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1664e-05,
+      "loss": 1.6253,
+      "step": 47835
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 2.1661037037037037e-05,
+      "loss": 1.5159,
+      "step": 47840
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1658074074074076e-05,
+      "loss": 1.7313,
+      "step": 47845
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.165511111111111e-05,
+      "loss": 1.5913,
+      "step": 47850
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.165214814814815e-05,
+      "loss": 1.5954,
+      "step": 47855
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1649185185185185e-05,
+      "loss": 1.715,
+      "step": 47860
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1646222222222224e-05,
+      "loss": 1.6193,
+      "step": 47865
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.164325925925926e-05,
+      "loss": 1.6036,
+      "step": 47870
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.16402962962963e-05,
+      "loss": 1.5861,
+      "step": 47875
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1637333333333334e-05,
+      "loss": 1.5645,
+      "step": 47880
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1634370370370373e-05,
+      "loss": 1.5977,
+      "step": 47885
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1631407407407408e-05,
+      "loss": 1.5133,
+      "step": 47890
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 2.1628444444444447e-05,
+      "loss": 1.6485,
+      "step": 47895
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1625481481481482e-05,
+      "loss": 1.5182,
+      "step": 47900
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.162251851851852e-05,
+      "loss": 1.5743,
+      "step": 47905
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1619555555555557e-05,
+      "loss": 1.6121,
+      "step": 47910
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1616592592592595e-05,
+      "loss": 1.4969,
+      "step": 47915
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.161362962962963e-05,
+      "loss": 1.4923,
+      "step": 47920
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.161066666666667e-05,
+      "loss": 1.6338,
+      "step": 47925
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1607703703703705e-05,
+      "loss": 1.6001,
+      "step": 47930
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.160474074074074e-05,
+      "loss": 1.598,
+      "step": 47935
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.160177777777778e-05,
+      "loss": 1.5255,
+      "step": 47940
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1598814814814815e-05,
+      "loss": 1.6589,
+      "step": 47945
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 2.1595851851851854e-05,
+      "loss": 1.4474,
+      "step": 47950
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.159288888888889e-05,
+      "loss": 1.5267,
+      "step": 47955
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1589925925925928e-05,
+      "loss": 1.6272,
+      "step": 47960
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1586962962962963e-05,
+      "loss": 1.5216,
+      "step": 47965
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1584000000000002e-05,
+      "loss": 1.6175,
+      "step": 47970
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1581037037037037e-05,
+      "loss": 1.6078,
+      "step": 47975
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1578074074074076e-05,
+      "loss": 1.5719,
+      "step": 47980
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1575111111111112e-05,
+      "loss": 1.6192,
+      "step": 47985
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.157214814814815e-05,
+      "loss": 1.5284,
+      "step": 47990
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1569185185185186e-05,
+      "loss": 1.5808,
+      "step": 47995
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.1566222222222225e-05,
+      "loss": 1.5719,
+      "step": 48000
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 2.156325925925926e-05,
+      "loss": 1.4008,
+      "step": 48005
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.15602962962963e-05,
+      "loss": 1.5677,
+      "step": 48010
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1557333333333335e-05,
+      "loss": 1.5654,
+      "step": 48015
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1554370370370373e-05,
+      "loss": 1.5501,
+      "step": 48020
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.155140740740741e-05,
+      "loss": 1.641,
+      "step": 48025
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1548444444444448e-05,
+      "loss": 1.6144,
+      "step": 48030
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1545481481481483e-05,
+      "loss": 1.6962,
+      "step": 48035
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1542518518518522e-05,
+      "loss": 1.5706,
+      "step": 48040
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1539555555555557e-05,
+      "loss": 1.6857,
+      "step": 48045
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1536592592592593e-05,
+      "loss": 1.6497,
+      "step": 48050
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.153362962962963e-05,
+      "loss": 1.5675,
+      "step": 48055
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1530666666666667e-05,
+      "loss": 1.632,
+      "step": 48060
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 2.1527703703703706e-05,
+      "loss": 1.6078,
+      "step": 48065
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.152474074074074e-05,
+      "loss": 1.5795,
+      "step": 48070
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.152177777777778e-05,
+      "loss": 1.6866,
+      "step": 48075
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1518814814814815e-05,
+      "loss": 1.5731,
+      "step": 48080
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1515851851851854e-05,
+      "loss": 1.4527,
+      "step": 48085
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.151288888888889e-05,
+      "loss": 1.5202,
+      "step": 48090
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.150992592592593e-05,
+      "loss": 1.4852,
+      "step": 48095
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1506962962962964e-05,
+      "loss": 1.5316,
+      "step": 48100
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1504000000000003e-05,
+      "loss": 1.5161,
+      "step": 48105
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1501037037037038e-05,
+      "loss": 1.5253,
+      "step": 48110
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1498074074074077e-05,
+      "loss": 1.5583,
+      "step": 48115
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 2.1495111111111113e-05,
+      "loss": 1.5555,
+      "step": 48120
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.149214814814815e-05,
+      "loss": 1.6236,
+      "step": 48125
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1489185185185187e-05,
+      "loss": 1.5313,
+      "step": 48130
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1486222222222226e-05,
+      "loss": 1.5525,
+      "step": 48135
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.148325925925926e-05,
+      "loss": 1.6012,
+      "step": 48140
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.14802962962963e-05,
+      "loss": 1.6264,
+      "step": 48145
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1477333333333335e-05,
+      "loss": 1.5451,
+      "step": 48150
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1474370370370374e-05,
+      "loss": 1.5717,
+      "step": 48155
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.147140740740741e-05,
+      "loss": 1.518,
+      "step": 48160
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1468444444444445e-05,
+      "loss": 1.644,
+      "step": 48165
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.1465481481481484e-05,
+      "loss": 1.6199,
+      "step": 48170
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 2.146251851851852e-05,
+      "loss": 1.6452,
+      "step": 48175
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1459555555555558e-05,
+      "loss": 1.5452,
+      "step": 48180
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1456592592592593e-05,
+      "loss": 1.604,
+      "step": 48185
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1453629629629632e-05,
+      "loss": 1.6413,
+      "step": 48190
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1450666666666668e-05,
+      "loss": 1.6097,
+      "step": 48195
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1447703703703707e-05,
+      "loss": 1.4579,
+      "step": 48200
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1444740740740742e-05,
+      "loss": 1.5101,
+      "step": 48205
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.144177777777778e-05,
+      "loss": 1.5261,
+      "step": 48210
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1438814814814816e-05,
+      "loss": 1.5075,
+      "step": 48215
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.1435851851851855e-05,
+      "loss": 1.5177,
+      "step": 48220
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.143288888888889e-05,
+      "loss": 1.6129,
+      "step": 48225
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 2.142992592592593e-05,
+      "loss": 1.4983,
+      "step": 48230
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1426962962962965e-05,
+      "loss": 1.5483,
+      "step": 48235
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1424000000000004e-05,
+      "loss": 1.6729,
+      "step": 48240
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1421037037037036e-05,
+      "loss": 1.6574,
+      "step": 48245
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1418074074074074e-05,
+      "loss": 1.5554,
+      "step": 48250
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.141511111111111e-05,
+      "loss": 1.5557,
+      "step": 48255
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.141214814814815e-05,
+      "loss": 1.527,
+      "step": 48260
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1409185185185184e-05,
+      "loss": 1.6802,
+      "step": 48265
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1406814814814817e-05,
+      "loss": 1.6135,
+      "step": 48270
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1403851851851853e-05,
+      "loss": 1.6063,
+      "step": 48275
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1400888888888888e-05,
+      "loss": 1.6415,
+      "step": 48280
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1397925925925927e-05,
+      "loss": 1.4916,
+      "step": 48285
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 2.1394962962962962e-05,
+      "loss": 1.6307,
+      "step": 48290
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1392e-05,
+      "loss": 1.4657,
+      "step": 48295
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1389037037037037e-05,
+      "loss": 1.5909,
+      "step": 48300
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1386074074074075e-05,
+      "loss": 1.4674,
+      "step": 48305
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.138311111111111e-05,
+      "loss": 1.6717,
+      "step": 48310
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.138014814814815e-05,
+      "loss": 1.6869,
+      "step": 48315
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1377185185185185e-05,
+      "loss": 1.4619,
+      "step": 48320
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1374222222222224e-05,
+      "loss": 1.7352,
+      "step": 48325
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.137125925925926e-05,
+      "loss": 1.5882,
+      "step": 48330
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1368296296296298e-05,
+      "loss": 1.6455,
+      "step": 48335
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1365333333333334e-05,
+      "loss": 1.6886,
+      "step": 48340
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 2.1362370370370372e-05,
+      "loss": 1.5835,
+      "step": 48345
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1359407407407408e-05,
+      "loss": 1.6421,
+      "step": 48350
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1356444444444447e-05,
+      "loss": 1.706,
+      "step": 48355
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1353481481481482e-05,
+      "loss": 1.4444,
+      "step": 48360
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.135051851851852e-05,
+      "loss": 1.5846,
+      "step": 48365
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1347555555555556e-05,
+      "loss": 1.6485,
+      "step": 48370
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1344592592592595e-05,
+      "loss": 1.6293,
+      "step": 48375
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.134162962962963e-05,
+      "loss": 1.6106,
+      "step": 48380
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.133866666666667e-05,
+      "loss": 1.6035,
+      "step": 48385
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.1335703703703705e-05,
+      "loss": 1.6398,
+      "step": 48390
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.133274074074074e-05,
+      "loss": 1.504,
+      "step": 48395
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 2.132977777777778e-05,
+      "loss": 1.7129,
+      "step": 48400
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1326814814814815e-05,
+      "loss": 1.6101,
+      "step": 48405
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1323851851851853e-05,
+      "loss": 1.5538,
+      "step": 48410
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.132088888888889e-05,
+      "loss": 1.5939,
+      "step": 48415
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1317925925925928e-05,
+      "loss": 1.5957,
+      "step": 48420
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1314962962962963e-05,
+      "loss": 1.6895,
+      "step": 48425
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1312000000000002e-05,
+      "loss": 1.5773,
+      "step": 48430
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1309037037037037e-05,
+      "loss": 1.571,
+      "step": 48435
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1306074074074076e-05,
+      "loss": 1.5118,
+      "step": 48440
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.130311111111111e-05,
+      "loss": 1.5895,
+      "step": 48445
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.130014814814815e-05,
+      "loss": 1.4697,
+      "step": 48450
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 2.1297185185185186e-05,
+      "loss": 1.6358,
+      "step": 48455
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1294222222222225e-05,
+      "loss": 1.4016,
+      "step": 48460
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.129125925925926e-05,
+      "loss": 1.6203,
+      "step": 48465
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.12882962962963e-05,
+      "loss": 1.6866,
+      "step": 48470
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1285333333333334e-05,
+      "loss": 1.4638,
+      "step": 48475
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1282370370370373e-05,
+      "loss": 1.644,
+      "step": 48480
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.127940740740741e-05,
+      "loss": 1.6533,
+      "step": 48485
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1276444444444447e-05,
+      "loss": 1.5509,
+      "step": 48490
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1273481481481483e-05,
+      "loss": 1.5816,
+      "step": 48495
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1270518518518522e-05,
+      "loss": 1.5807,
+      "step": 48500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1267555555555557e-05,
+      "loss": 1.6322,
+      "step": 48505
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.1264592592592593e-05,
+      "loss": 1.5263,
+      "step": 48510
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 2.126162962962963e-05,
+      "loss": 1.5807,
+      "step": 48515
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1258666666666667e-05,
+      "loss": 1.458,
+      "step": 48520
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1255703703703706e-05,
+      "loss": 1.5772,
+      "step": 48525
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.125274074074074e-05,
+      "loss": 1.6866,
+      "step": 48530
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.124977777777778e-05,
+      "loss": 1.6285,
+      "step": 48535
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1246814814814815e-05,
+      "loss": 1.5968,
+      "step": 48540
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1243851851851854e-05,
+      "loss": 1.5838,
+      "step": 48545
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.124088888888889e-05,
+      "loss": 1.57,
+      "step": 48550
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.123792592592593e-05,
+      "loss": 1.5923,
+      "step": 48555
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1234962962962964e-05,
+      "loss": 1.5629,
+      "step": 48560
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1232000000000003e-05,
+      "loss": 1.5648,
+      "step": 48565
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 2.1229037037037038e-05,
+      "loss": 1.6528,
+      "step": 48570
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1226074074074077e-05,
+      "loss": 1.6432,
+      "step": 48575
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1223111111111112e-05,
+      "loss": 1.4514,
+      "step": 48580
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.122014814814815e-05,
+      "loss": 1.5709,
+      "step": 48585
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1217185185185187e-05,
+      "loss": 1.5943,
+      "step": 48590
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1214222222222225e-05,
+      "loss": 1.6851,
+      "step": 48595
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.121125925925926e-05,
+      "loss": 1.6304,
+      "step": 48600
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.12082962962963e-05,
+      "loss": 1.5762,
+      "step": 48605
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1205333333333335e-05,
+      "loss": 1.5765,
+      "step": 48610
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1202370370370374e-05,
+      "loss": 1.4383,
+      "step": 48615
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.119940740740741e-05,
+      "loss": 1.6192,
+      "step": 48620
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 2.1196444444444445e-05,
+      "loss": 1.5757,
+      "step": 48625
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1193481481481484e-05,
+      "loss": 1.7062,
+      "step": 48630
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.119051851851852e-05,
+      "loss": 1.5611,
+      "step": 48635
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1187555555555558e-05,
+      "loss": 1.5267,
+      "step": 48640
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1184592592592593e-05,
+      "loss": 1.7443,
+      "step": 48645
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1181629629629632e-05,
+      "loss": 1.5347,
+      "step": 48650
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1178666666666668e-05,
+      "loss": 1.5754,
+      "step": 48655
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1175703703703706e-05,
+      "loss": 1.6667,
+      "step": 48660
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1172740740740742e-05,
+      "loss": 1.5728,
+      "step": 48665
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.116977777777778e-05,
+      "loss": 1.6377,
+      "step": 48670
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1166814814814816e-05,
+      "loss": 1.5926,
+      "step": 48675
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 2.1163851851851855e-05,
+      "loss": 1.4464,
+      "step": 48680
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.116088888888889e-05,
+      "loss": 1.6894,
+      "step": 48685
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.115792592592593e-05,
+      "loss": 1.5962,
+      "step": 48690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1154962962962965e-05,
+      "loss": 1.6734,
+      "step": 48695
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1152000000000003e-05,
+      "loss": 1.5913,
+      "step": 48700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.114903703703704e-05,
+      "loss": 1.5971,
+      "step": 48705
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1146074074074074e-05,
+      "loss": 1.5631,
+      "step": 48710
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.114311111111111e-05,
+      "loss": 1.6449,
+      "step": 48715
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.114014814814815e-05,
+      "loss": 1.5261,
+      "step": 48720
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1137185185185184e-05,
+      "loss": 1.6033,
+      "step": 48725
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1134222222222223e-05,
+      "loss": 1.7437,
+      "step": 48730
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1131259259259258e-05,
+      "loss": 1.5837,
+      "step": 48735
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 2.1128296296296297e-05,
+      "loss": 1.5545,
+      "step": 48740
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1125333333333333e-05,
+      "loss": 1.553,
+      "step": 48745
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.112237037037037e-05,
+      "loss": 1.64,
+      "step": 48750
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1119407407407407e-05,
+      "loss": 1.588,
+      "step": 48755
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1116444444444446e-05,
+      "loss": 1.6126,
+      "step": 48760
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.111348148148148e-05,
+      "loss": 1.5341,
+      "step": 48765
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.111051851851852e-05,
+      "loss": 1.6896,
+      "step": 48770
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1107555555555555e-05,
+      "loss": 1.5585,
+      "step": 48775
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1104592592592594e-05,
+      "loss": 1.5181,
+      "step": 48780
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.110162962962963e-05,
+      "loss": 1.5537,
+      "step": 48785
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1098666666666665e-05,
+      "loss": 1.6398,
+      "step": 48790
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 2.1095703703703704e-05,
+      "loss": 1.6032,
+      "step": 48795
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.109274074074074e-05,
+      "loss": 1.5996,
+      "step": 48800
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1089777777777778e-05,
+      "loss": 1.586,
+      "step": 48805
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1086814814814814e-05,
+      "loss": 1.6109,
+      "step": 48810
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1083851851851852e-05,
+      "loss": 1.7435,
+      "step": 48815
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1080888888888888e-05,
+      "loss": 1.5501,
+      "step": 48820
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1077925925925927e-05,
+      "loss": 1.4657,
+      "step": 48825
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1074962962962962e-05,
+      "loss": 1.6426,
+      "step": 48830
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1072e-05,
+      "loss": 1.5512,
+      "step": 48835
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1069037037037036e-05,
+      "loss": 1.7229,
+      "step": 48840
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.1066074074074075e-05,
+      "loss": 1.5883,
+      "step": 48845
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 2.106311111111111e-05,
+      "loss": 1.6108,
+      "step": 48850
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.106014814814815e-05,
+      "loss": 1.4672,
+      "step": 48855
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1057185185185185e-05,
+      "loss": 1.634,
+      "step": 48860
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1054222222222224e-05,
+      "loss": 1.6279,
+      "step": 48865
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.105125925925926e-05,
+      "loss": 1.5786,
+      "step": 48870
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1048296296296298e-05,
+      "loss": 1.5631,
+      "step": 48875
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1045333333333333e-05,
+      "loss": 1.5632,
+      "step": 48880
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1042370370370372e-05,
+      "loss": 1.5749,
+      "step": 48885
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1039407407407408e-05,
+      "loss": 1.43,
+      "step": 48890
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1036444444444446e-05,
+      "loss": 1.7026,
+      "step": 48895
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1033481481481482e-05,
+      "loss": 1.4874,
+      "step": 48900
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 2.1030518518518517e-05,
+      "loss": 1.5278,
+      "step": 48905
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1027555555555556e-05,
+      "loss": 1.5622,
+      "step": 48910
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.102459259259259e-05,
+      "loss": 1.4414,
+      "step": 48915
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.102162962962963e-05,
+      "loss": 1.5746,
+      "step": 48920
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1018666666666666e-05,
+      "loss": 1.5149,
+      "step": 48925
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1015703703703705e-05,
+      "loss": 1.5961,
+      "step": 48930
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.101274074074074e-05,
+      "loss": 1.5621,
+      "step": 48935
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.100977777777778e-05,
+      "loss": 1.5935,
+      "step": 48940
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1006814814814814e-05,
+      "loss": 1.5438,
+      "step": 48945
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.1003851851851853e-05,
+      "loss": 1.5941,
+      "step": 48950
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.100088888888889e-05,
+      "loss": 1.6053,
+      "step": 48955
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.0997925925925927e-05,
+      "loss": 1.6134,
+      "step": 48960
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 2.0994962962962963e-05,
+      "loss": 1.4818,
+      "step": 48965
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0992e-05,
+      "loss": 1.5851,
+      "step": 48970
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0989037037037037e-05,
+      "loss": 1.4719,
+      "step": 48975
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0986074074074076e-05,
+      "loss": 1.5611,
+      "step": 48980
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.098311111111111e-05,
+      "loss": 1.5807,
+      "step": 48985
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.098014814814815e-05,
+      "loss": 1.577,
+      "step": 48990
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0977185185185186e-05,
+      "loss": 1.6071,
+      "step": 48995
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0974222222222224e-05,
+      "loss": 1.5249,
+      "step": 49000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.097125925925926e-05,
+      "loss": 1.6473,
+      "step": 49005
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.09682962962963e-05,
+      "loss": 1.6365,
+      "step": 49010
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.0965333333333334e-05,
+      "loss": 1.6197,
+      "step": 49015
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 2.096237037037037e-05,
+      "loss": 1.647,
+      "step": 49020
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.095940740740741e-05,
+      "loss": 1.4746,
+      "step": 49025
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0956444444444444e-05,
+      "loss": 1.5431,
+      "step": 49030
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0953481481481483e-05,
+      "loss": 1.4384,
+      "step": 49035
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0950518518518518e-05,
+      "loss": 1.6573,
+      "step": 49040
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0947555555555557e-05,
+      "loss": 1.6509,
+      "step": 49045
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0944592592592592e-05,
+      "loss": 1.5921,
+      "step": 49050
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.094162962962963e-05,
+      "loss": 1.6706,
+      "step": 49055
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0938666666666667e-05,
+      "loss": 1.5353,
+      "step": 49060
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.0935703703703705e-05,
+      "loss": 1.6225,
+      "step": 49065
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.093274074074074e-05,
+      "loss": 1.588,
+      "step": 49070
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 2.092977777777778e-05,
+      "loss": 1.6211,
+      "step": 49075
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0926814814814815e-05,
+      "loss": 1.6526,
+      "step": 49080
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0923851851851854e-05,
+      "loss": 1.6996,
+      "step": 49085
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.092088888888889e-05,
+      "loss": 1.6127,
+      "step": 49090
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0917925925925928e-05,
+      "loss": 1.555,
+      "step": 49095
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0914962962962964e-05,
+      "loss": 1.5162,
+      "step": 49100
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0912000000000002e-05,
+      "loss": 1.5465,
+      "step": 49105
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0909037037037038e-05,
+      "loss": 1.6422,
+      "step": 49110
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0906074074074077e-05,
+      "loss": 1.5217,
+      "step": 49115
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0903111111111112e-05,
+      "loss": 1.619,
+      "step": 49120
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.090014814814815e-05,
+      "loss": 1.5617,
+      "step": 49125
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 2.0897185185185186e-05,
+      "loss": 1.5746,
+      "step": 49130
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0894222222222222e-05,
+      "loss": 1.5058,
+      "step": 49135
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.089125925925926e-05,
+      "loss": 1.7295,
+      "step": 49140
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0888296296296296e-05,
+      "loss": 1.5349,
+      "step": 49145
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0885333333333335e-05,
+      "loss": 1.6692,
+      "step": 49150
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.088237037037037e-05,
+      "loss": 1.5457,
+      "step": 49155
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.087940740740741e-05,
+      "loss": 1.5076,
+      "step": 49160
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0876444444444445e-05,
+      "loss": 1.616,
+      "step": 49165
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0873481481481483e-05,
+      "loss": 1.6334,
+      "step": 49170
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.087051851851852e-05,
+      "loss": 1.6823,
+      "step": 49175
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0867555555555558e-05,
+      "loss": 1.5324,
+      "step": 49180
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0864592592592593e-05,
+      "loss": 1.5831,
+      "step": 49185
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 2.0861629629629632e-05,
+      "loss": 1.5679,
+      "step": 49190
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0858666666666667e-05,
+      "loss": 1.4764,
+      "step": 49195
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0855703703703706e-05,
+      "loss": 1.5852,
+      "step": 49200
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.085274074074074e-05,
+      "loss": 1.5673,
+      "step": 49205
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.084977777777778e-05,
+      "loss": 1.5226,
+      "step": 49210
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0846814814814816e-05,
+      "loss": 1.6363,
+      "step": 49215
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0843851851851855e-05,
+      "loss": 1.6604,
+      "step": 49220
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.084088888888889e-05,
+      "loss": 1.6934,
+      "step": 49225
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.083792592592593e-05,
+      "loss": 1.5707,
+      "step": 49230
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0834962962962964e-05,
+      "loss": 1.5703,
+      "step": 49235
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.0832000000000003e-05,
+      "loss": 1.6097,
+      "step": 49240
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 2.082903703703704e-05,
+      "loss": 1.6224,
+      "step": 49245
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0826074074074074e-05,
+      "loss": 1.6108,
+      "step": 49250
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0823111111111113e-05,
+      "loss": 1.6177,
+      "step": 49255
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0820148148148148e-05,
+      "loss": 1.532,
+      "step": 49260
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0817185185185187e-05,
+      "loss": 1.5187,
+      "step": 49265
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0814222222222223e-05,
+      "loss": 1.5628,
+      "step": 49270
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.081125925925926e-05,
+      "loss": 1.5729,
+      "step": 49275
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0808296296296297e-05,
+      "loss": 1.639,
+      "step": 49280
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0805333333333336e-05,
+      "loss": 1.5739,
+      "step": 49285
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.080237037037037e-05,
+      "loss": 1.658,
+      "step": 49290
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.079940740740741e-05,
+      "loss": 1.5964,
+      "step": 49295
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 2.0796444444444445e-05,
+      "loss": 1.6005,
+      "step": 49300
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0793481481481484e-05,
+      "loss": 1.7932,
+      "step": 49305
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.079051851851852e-05,
+      "loss": 1.4853,
+      "step": 49310
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.078755555555556e-05,
+      "loss": 1.4995,
+      "step": 49315
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0784592592592594e-05,
+      "loss": 1.5547,
+      "step": 49320
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0781629629629633e-05,
+      "loss": 1.6237,
+      "step": 49325
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0778666666666668e-05,
+      "loss": 1.6151,
+      "step": 49330
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0775703703703707e-05,
+      "loss": 1.6293,
+      "step": 49335
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0772740740740742e-05,
+      "loss": 1.6123,
+      "step": 49340
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.076977777777778e-05,
+      "loss": 1.6295,
+      "step": 49345
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0766814814814817e-05,
+      "loss": 1.6001,
+      "step": 49350
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 2.0763851851851855e-05,
+      "loss": 1.6561,
+      "step": 49355
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.076088888888889e-05,
+      "loss": 1.5899,
+      "step": 49360
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0757925925925926e-05,
+      "loss": 1.6044,
+      "step": 49365
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0754962962962965e-05,
+      "loss": 1.6045,
+      "step": 49370
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0752e-05,
+      "loss": 1.5086,
+      "step": 49375
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.074903703703704e-05,
+      "loss": 1.49,
+      "step": 49380
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0746074074074075e-05,
+      "loss": 1.646,
+      "step": 49385
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0743111111111114e-05,
+      "loss": 1.5894,
+      "step": 49390
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.074014814814815e-05,
+      "loss": 1.5594,
+      "step": 49395
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0737185185185188e-05,
+      "loss": 1.5403,
+      "step": 49400
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0734222222222223e-05,
+      "loss": 1.5321,
+      "step": 49405
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0731259259259262e-05,
+      "loss": 1.6369,
+      "step": 49410
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 2.0728296296296298e-05,
+      "loss": 1.6057,
+      "step": 49415
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0725333333333336e-05,
+      "loss": 1.45,
+      "step": 49420
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0722370370370372e-05,
+      "loss": 1.5284,
+      "step": 49425
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.071940740740741e-05,
+      "loss": 1.7225,
+      "step": 49430
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0716444444444446e-05,
+      "loss": 1.4506,
+      "step": 49435
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0713481481481485e-05,
+      "loss": 1.5763,
+      "step": 49440
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.071051851851852e-05,
+      "loss": 1.6135,
+      "step": 49445
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.070755555555556e-05,
+      "loss": 1.682,
+      "step": 49450
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0704592592592595e-05,
+      "loss": 1.5632,
+      "step": 49455
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0701629629629633e-05,
+      "loss": 1.5185,
+      "step": 49460
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.069866666666667e-05,
+      "loss": 1.5057,
+      "step": 49465
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 2.0695703703703708e-05,
+      "loss": 1.5888,
+      "step": 49470
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0692740740740743e-05,
+      "loss": 1.6223,
+      "step": 49475
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.068977777777778e-05,
+      "loss": 1.7058,
+      "step": 49480
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0686814814814814e-05,
+      "loss": 1.4756,
+      "step": 49485
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0683851851851853e-05,
+      "loss": 1.5278,
+      "step": 49490
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0680888888888888e-05,
+      "loss": 1.6419,
+      "step": 49495
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0677925925925927e-05,
+      "loss": 1.5949,
+      "step": 49500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0674962962962962e-05,
+      "loss": 1.664,
+      "step": 49505
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0672e-05,
+      "loss": 1.5529,
+      "step": 49510
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0669037037037037e-05,
+      "loss": 1.5925,
+      "step": 49515
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.0666074074074072e-05,
+      "loss": 1.5646,
+      "step": 49520
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 2.066311111111111e-05,
+      "loss": 1.5784,
+      "step": 49525
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0660148148148146e-05,
+      "loss": 1.4748,
+      "step": 49530
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0657185185185185e-05,
+      "loss": 1.5894,
+      "step": 49535
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.065422222222222e-05,
+      "loss": 1.5755,
+      "step": 49540
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.065125925925926e-05,
+      "loss": 1.6588,
+      "step": 49545
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0648296296296295e-05,
+      "loss": 1.4702,
+      "step": 49550
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0645333333333334e-05,
+      "loss": 1.5999,
+      "step": 49555
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.064237037037037e-05,
+      "loss": 1.6394,
+      "step": 49560
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0639407407407408e-05,
+      "loss": 1.5283,
+      "step": 49565
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0636444444444443e-05,
+      "loss": 1.7313,
+      "step": 49570
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0633481481481482e-05,
+      "loss": 1.5737,
+      "step": 49575
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 2.0630518518518518e-05,
+      "loss": 1.5782,
+      "step": 49580
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0627555555555557e-05,
+      "loss": 1.6574,
+      "step": 49585
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0624592592592592e-05,
+      "loss": 1.5658,
+      "step": 49590
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.062162962962963e-05,
+      "loss": 1.6031,
+      "step": 49595
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0618666666666666e-05,
+      "loss": 1.5707,
+      "step": 49600
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0615703703703705e-05,
+      "loss": 1.6226,
+      "step": 49605
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.061274074074074e-05,
+      "loss": 1.5638,
+      "step": 49610
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.060977777777778e-05,
+      "loss": 1.5982,
+      "step": 49615
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0606814814814815e-05,
+      "loss": 1.4467,
+      "step": 49620
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0603851851851854e-05,
+      "loss": 1.6394,
+      "step": 49625
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.060088888888889e-05,
+      "loss": 1.5951,
+      "step": 49630
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0597925925925924e-05,
+      "loss": 1.5576,
+      "step": 49635
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 2.0594962962962963e-05,
+      "loss": 1.5621,
+      "step": 49640
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0592e-05,
+      "loss": 1.5921,
+      "step": 49645
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0589037037037038e-05,
+      "loss": 1.4863,
+      "step": 49650
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0586074074074073e-05,
+      "loss": 1.6157,
+      "step": 49655
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0583111111111112e-05,
+      "loss": 1.5637,
+      "step": 49660
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0580148148148147e-05,
+      "loss": 1.611,
+      "step": 49665
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0577185185185186e-05,
+      "loss": 1.6554,
+      "step": 49670
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.057422222222222e-05,
+      "loss": 1.6317,
+      "step": 49675
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.057125925925926e-05,
+      "loss": 1.5502,
+      "step": 49680
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0568296296296296e-05,
+      "loss": 1.517,
+      "step": 49685
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.0565333333333335e-05,
+      "loss": 1.6514,
+      "step": 49690
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 2.056237037037037e-05,
+      "loss": 1.6266,
+      "step": 49695
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.055940740740741e-05,
+      "loss": 1.6582,
+      "step": 49700
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0556444444444444e-05,
+      "loss": 1.5833,
+      "step": 49705
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0553481481481483e-05,
+      "loss": 1.53,
+      "step": 49710
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.055051851851852e-05,
+      "loss": 1.5525,
+      "step": 49715
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0547555555555557e-05,
+      "loss": 1.5935,
+      "step": 49720
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0544592592592593e-05,
+      "loss": 1.517,
+      "step": 49725
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.054162962962963e-05,
+      "loss": 1.5663,
+      "step": 49730
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0538666666666667e-05,
+      "loss": 1.5452,
+      "step": 49735
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0535703703703706e-05,
+      "loss": 1.5779,
+      "step": 49740
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.053274074074074e-05,
+      "loss": 1.6238,
+      "step": 49745
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 2.0529777777777777e-05,
+      "loss": 1.5781,
+      "step": 49750
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0526814814814815e-05,
+      "loss": 1.5738,
+      "step": 49755
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.052385185185185e-05,
+      "loss": 1.6232,
+      "step": 49760
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.052088888888889e-05,
+      "loss": 1.619,
+      "step": 49765
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0517925925925925e-05,
+      "loss": 1.5196,
+      "step": 49770
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0514962962962964e-05,
+      "loss": 1.5067,
+      "step": 49775
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0512e-05,
+      "loss": 1.6335,
+      "step": 49780
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0509037037037038e-05,
+      "loss": 1.5742,
+      "step": 49785
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0506074074074074e-05,
+      "loss": 1.5137,
+      "step": 49790
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0503111111111113e-05,
+      "loss": 1.5472,
+      "step": 49795
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0500148148148148e-05,
+      "loss": 1.6103,
+      "step": 49800
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 2.0497185185185187e-05,
+      "loss": 1.5557,
+      "step": 49805
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0494222222222222e-05,
+      "loss": 1.4856,
+      "step": 49810
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.049125925925926e-05,
+      "loss": 1.5725,
+      "step": 49815
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0488296296296296e-05,
+      "loss": 1.6263,
+      "step": 49820
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0485333333333335e-05,
+      "loss": 1.5305,
+      "step": 49825
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.048237037037037e-05,
+      "loss": 1.5367,
+      "step": 49830
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.047940740740741e-05,
+      "loss": 1.7601,
+      "step": 49835
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0476444444444445e-05,
+      "loss": 1.5781,
+      "step": 49840
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0473481481481484e-05,
+      "loss": 1.5829,
+      "step": 49845
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.047051851851852e-05,
+      "loss": 1.5604,
+      "step": 49850
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0467555555555558e-05,
+      "loss": 1.5173,
+      "step": 49855
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.0464592592592593e-05,
+      "loss": 1.5328,
+      "step": 49860
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 2.046162962962963e-05,
+      "loss": 1.7212,
+      "step": 49865
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0458666666666668e-05,
+      "loss": 1.4366,
+      "step": 49870
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0455703703703703e-05,
+      "loss": 1.6472,
+      "step": 49875
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0452740740740742e-05,
+      "loss": 1.5149,
+      "step": 49880
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0449777777777777e-05,
+      "loss": 1.5404,
+      "step": 49885
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0446814814814816e-05,
+      "loss": 1.5332,
+      "step": 49890
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0443851851851852e-05,
+      "loss": 1.745,
+      "step": 49895
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.044088888888889e-05,
+      "loss": 1.5552,
+      "step": 49900
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0437925925925926e-05,
+      "loss": 1.6491,
+      "step": 49905
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0434962962962965e-05,
+      "loss": 1.4444,
+      "step": 49910
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.0432e-05,
+      "loss": 1.5545,
+      "step": 49915
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 2.042903703703704e-05,
+      "loss": 1.4867,
+      "step": 49920
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0426074074074074e-05,
+      "loss": 1.5529,
+      "step": 49925
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0423111111111113e-05,
+      "loss": 1.5834,
+      "step": 49930
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.042014814814815e-05,
+      "loss": 1.6302,
+      "step": 49935
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0417185185185188e-05,
+      "loss": 1.5661,
+      "step": 49940
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0414222222222223e-05,
+      "loss": 1.5621,
+      "step": 49945
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0411259259259262e-05,
+      "loss": 1.5706,
+      "step": 49950
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0408296296296297e-05,
+      "loss": 1.5657,
+      "step": 49955
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0405333333333336e-05,
+      "loss": 1.7071,
+      "step": 49960
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.040237037037037e-05,
+      "loss": 1.612,
+      "step": 49965
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.039940740740741e-05,
+      "loss": 1.5886,
+      "step": 49970
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 2.0396444444444446e-05,
+      "loss": 1.5698,
+      "step": 49975
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.039348148148148e-05,
+      "loss": 1.6793,
+      "step": 49980
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.039051851851852e-05,
+      "loss": 1.6018,
+      "step": 49985
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0387555555555555e-05,
+      "loss": 1.4914,
+      "step": 49990
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0384592592592594e-05,
+      "loss": 1.5095,
+      "step": 49995
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.038162962962963e-05,
+      "loss": 1.7056,
+      "step": 50000
+    },
+    {
+      "epoch": 8.89,
+      "eval_loss": 1.4441460371017456,
+      "eval_rouge2_fmeasure": 0.1994,
+      "eval_rouge2_precision": 0.2358,
+      "eval_rouge2_recall": 0.1819,
+      "eval_runtime": 36743.4165,
+      "eval_samples_per_second": 0.136,
+      "eval_steps_per_second": 0.068,
+      "step": 50000
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.037866666666667e-05,
+      "loss": 1.5776,
+      "step": 50005
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0375703703703704e-05,
+      "loss": 1.5854,
+      "step": 50010
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0372740740740743e-05,
+      "loss": 1.583,
+      "step": 50015
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0369777777777778e-05,
+      "loss": 1.5779,
+      "step": 50020
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0366814814814817e-05,
+      "loss": 1.6658,
+      "step": 50025
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 2.0363851851851852e-05,
+      "loss": 1.5747,
+      "step": 50030
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.036088888888889e-05,
+      "loss": 1.6231,
+      "step": 50035
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0357925925925927e-05,
+      "loss": 1.6523,
+      "step": 50040
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0354962962962966e-05,
+      "loss": 1.6102,
+      "step": 50045
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0352e-05,
+      "loss": 1.5825,
+      "step": 50050
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.034903703703704e-05,
+      "loss": 1.5891,
+      "step": 50055
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0346074074074075e-05,
+      "loss": 1.4946,
+      "step": 50060
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0343111111111114e-05,
+      "loss": 1.5902,
+      "step": 50065
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.034014814814815e-05,
+      "loss": 1.5364,
+      "step": 50070
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.033718518518519e-05,
+      "loss": 1.6224,
+      "step": 50075
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0334222222222224e-05,
+      "loss": 1.6571,
+      "step": 50080
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0331259259259263e-05,
+      "loss": 1.7396,
+      "step": 50085
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 2.0328296296296298e-05,
+      "loss": 1.612,
+      "step": 50090
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0325333333333333e-05,
+      "loss": 1.4919,
+      "step": 50095
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0322370370370372e-05,
+      "loss": 1.5947,
+      "step": 50100
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0319407407407408e-05,
+      "loss": 1.6084,
+      "step": 50105
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0316444444444447e-05,
+      "loss": 1.6492,
+      "step": 50110
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0313481481481482e-05,
+      "loss": 1.5986,
+      "step": 50115
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.031051851851852e-05,
+      "loss": 1.7051,
+      "step": 50120
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0307555555555556e-05,
+      "loss": 1.6672,
+      "step": 50125
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0304592592592595e-05,
+      "loss": 1.5468,
+      "step": 50130
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.030162962962963e-05,
+      "loss": 1.5186,
+      "step": 50135
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.029866666666667e-05,
+      "loss": 1.5598,
+      "step": 50140
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 2.0295703703703705e-05,
+      "loss": 1.7102,
+      "step": 50145
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0292740740740744e-05,
+      "loss": 1.589,
+      "step": 50150
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.028977777777778e-05,
+      "loss": 1.4843,
+      "step": 50155
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0286814814814818e-05,
+      "loss": 1.5548,
+      "step": 50160
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0283851851851853e-05,
+      "loss": 1.6082,
+      "step": 50165
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0280888888888892e-05,
+      "loss": 1.6247,
+      "step": 50170
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0277925925925927e-05,
+      "loss": 1.5119,
+      "step": 50175
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0274962962962966e-05,
+      "loss": 1.5193,
+      "step": 50180
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0272000000000002e-05,
+      "loss": 1.6745,
+      "step": 50185
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.026903703703704e-05,
+      "loss": 1.5101,
+      "step": 50190
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0266074074074076e-05,
+      "loss": 1.6024,
+      "step": 50195
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 2.0263111111111115e-05,
+      "loss": 1.6457,
+      "step": 50200
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.026014814814815e-05,
+      "loss": 1.6042,
+      "step": 50205
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0257185185185186e-05,
+      "loss": 1.608,
+      "step": 50210
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0254222222222225e-05,
+      "loss": 1.5369,
+      "step": 50215
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.025125925925926e-05,
+      "loss": 1.65,
+      "step": 50220
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.02482962962963e-05,
+      "loss": 1.6281,
+      "step": 50225
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0245333333333334e-05,
+      "loss": 1.6117,
+      "step": 50230
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0242370370370373e-05,
+      "loss": 1.645,
+      "step": 50235
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.023940740740741e-05,
+      "loss": 1.5832,
+      "step": 50240
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0236444444444447e-05,
+      "loss": 1.6381,
+      "step": 50245
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.0233481481481483e-05,
+      "loss": 1.6442,
+      "step": 50250
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 2.023051851851852e-05,
+      "loss": 1.5972,
+      "step": 50255
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0227555555555554e-05,
+      "loss": 1.5896,
+      "step": 50260
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0224592592592592e-05,
+      "loss": 1.635,
+      "step": 50265
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0221629629629628e-05,
+      "loss": 1.4677,
+      "step": 50270
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0218666666666667e-05,
+      "loss": 1.5737,
+      "step": 50275
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0215703703703702e-05,
+      "loss": 1.4144,
+      "step": 50280
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.021274074074074e-05,
+      "loss": 1.5732,
+      "step": 50285
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0209777777777776e-05,
+      "loss": 1.6544,
+      "step": 50290
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0206814814814815e-05,
+      "loss": 1.5883,
+      "step": 50295
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.020385185185185e-05,
+      "loss": 1.5981,
+      "step": 50300
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.020088888888889e-05,
+      "loss": 1.5717,
+      "step": 50305
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0197925925925925e-05,
+      "loss": 1.6268,
+      "step": 50310
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 2.0194962962962964e-05,
+      "loss": 1.6219,
+      "step": 50315
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0192e-05,
+      "loss": 1.6537,
+      "step": 50320
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0189037037037038e-05,
+      "loss": 1.4755,
+      "step": 50325
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0186074074074073e-05,
+      "loss": 1.6602,
+      "step": 50330
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0183111111111112e-05,
+      "loss": 1.6864,
+      "step": 50335
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0180148148148148e-05,
+      "loss": 1.5156,
+      "step": 50340
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0177185185185186e-05,
+      "loss": 1.5972,
+      "step": 50345
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0174222222222222e-05,
+      "loss": 1.6299,
+      "step": 50350
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.017125925925926e-05,
+      "loss": 1.5206,
+      "step": 50355
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0168296296296296e-05,
+      "loss": 1.6595,
+      "step": 50360
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.0165333333333335e-05,
+      "loss": 1.6798,
+      "step": 50365
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 2.016237037037037e-05,
+      "loss": 1.6914,
+      "step": 50370
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0159407407407406e-05,
+      "loss": 1.7021,
+      "step": 50375
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0156444444444445e-05,
+      "loss": 1.5465,
+      "step": 50380
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.015348148148148e-05,
+      "loss": 1.5977,
+      "step": 50385
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.015051851851852e-05,
+      "loss": 1.625,
+      "step": 50390
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0147555555555554e-05,
+      "loss": 1.5877,
+      "step": 50395
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0144592592592593e-05,
+      "loss": 1.6119,
+      "step": 50400
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.014162962962963e-05,
+      "loss": 1.5196,
+      "step": 50405
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0138666666666667e-05,
+      "loss": 1.6991,
+      "step": 50410
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0135703703703703e-05,
+      "loss": 1.544,
+      "step": 50415
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.013274074074074e-05,
+      "loss": 1.6542,
+      "step": 50420
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 2.0129777777777777e-05,
+      "loss": 1.5415,
+      "step": 50425
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0126814814814816e-05,
+      "loss": 1.6447,
+      "step": 50430
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.012385185185185e-05,
+      "loss": 1.544,
+      "step": 50435
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.012088888888889e-05,
+      "loss": 1.7818,
+      "step": 50440
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0117925925925926e-05,
+      "loss": 1.5589,
+      "step": 50445
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0114962962962964e-05,
+      "loss": 1.6556,
+      "step": 50450
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0112e-05,
+      "loss": 1.4949,
+      "step": 50455
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.010903703703704e-05,
+      "loss": 1.5588,
+      "step": 50460
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0106074074074074e-05,
+      "loss": 1.5678,
+      "step": 50465
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0103111111111113e-05,
+      "loss": 1.5118,
+      "step": 50470
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.010014814814815e-05,
+      "loss": 1.6167,
+      "step": 50475
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 2.0097185185185187e-05,
+      "loss": 1.5925,
+      "step": 50480
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0094222222222223e-05,
+      "loss": 1.6209,
+      "step": 50485
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0091259259259258e-05,
+      "loss": 1.4695,
+      "step": 50490
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0088296296296297e-05,
+      "loss": 1.399,
+      "step": 50495
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0085333333333332e-05,
+      "loss": 1.6023,
+      "step": 50500
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.008237037037037e-05,
+      "loss": 1.5268,
+      "step": 50505
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0079407407407407e-05,
+      "loss": 1.6967,
+      "step": 50510
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0076444444444445e-05,
+      "loss": 1.4888,
+      "step": 50515
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.007348148148148e-05,
+      "loss": 1.65,
+      "step": 50520
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.007051851851852e-05,
+      "loss": 1.606,
+      "step": 50525
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0067555555555555e-05,
+      "loss": 1.651,
+      "step": 50530
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.0064592592592594e-05,
+      "loss": 1.6467,
+      "step": 50535
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 2.006162962962963e-05,
+      "loss": 1.6445,
+      "step": 50540
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0058666666666668e-05,
+      "loss": 1.4955,
+      "step": 50545
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0055703703703704e-05,
+      "loss": 1.6046,
+      "step": 50550
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0052740740740742e-05,
+      "loss": 1.5837,
+      "step": 50555
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0049777777777778e-05,
+      "loss": 1.4848,
+      "step": 50560
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0046814814814817e-05,
+      "loss": 1.6505,
+      "step": 50565
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0043851851851852e-05,
+      "loss": 1.4656,
+      "step": 50570
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.004088888888889e-05,
+      "loss": 1.5581,
+      "step": 50575
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0037925925925926e-05,
+      "loss": 1.6902,
+      "step": 50580
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0034962962962965e-05,
+      "loss": 1.5954,
+      "step": 50585
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.0032e-05,
+      "loss": 1.5099,
+      "step": 50590
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 2.002903703703704e-05,
+      "loss": 1.6501,
+      "step": 50595
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0026074074074075e-05,
+      "loss": 1.662,
+      "step": 50600
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.002311111111111e-05,
+      "loss": 1.5935,
+      "step": 50605
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.002014814814815e-05,
+      "loss": 1.6109,
+      "step": 50610
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0017185185185185e-05,
+      "loss": 1.5868,
+      "step": 50615
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0014222222222223e-05,
+      "loss": 1.6229,
+      "step": 50620
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.001125925925926e-05,
+      "loss": 1.5075,
+      "step": 50625
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0008296296296298e-05,
+      "loss": 1.5395,
+      "step": 50630
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0005333333333333e-05,
+      "loss": 1.5758,
+      "step": 50635
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 2.0002370370370372e-05,
+      "loss": 1.4438,
+      "step": 50640
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.9999407407407407e-05,
+      "loss": 1.4391,
+      "step": 50645
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.9996444444444446e-05,
+      "loss": 1.542,
+      "step": 50650
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.999348148148148e-05,
+      "loss": 1.5029,
+      "step": 50655
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.999051851851852e-05,
+      "loss": 1.4636,
+      "step": 50660
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9987555555555556e-05,
+      "loss": 1.5685,
+      "step": 50665
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9984592592592595e-05,
+      "loss": 1.4629,
+      "step": 50670
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.998162962962963e-05,
+      "loss": 1.423,
+      "step": 50675
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.997866666666667e-05,
+      "loss": 1.499,
+      "step": 50680
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9975703703703704e-05,
+      "loss": 1.4359,
+      "step": 50685
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9972740740740743e-05,
+      "loss": 1.5721,
+      "step": 50690
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.996977777777778e-05,
+      "loss": 1.568,
+      "step": 50695
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9966814814814817e-05,
+      "loss": 1.4496,
+      "step": 50700
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.9963851851851853e-05,
+      "loss": 1.5478,
+      "step": 50705
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9960888888888892e-05,
+      "loss": 1.4705,
+      "step": 50710
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9957925925925927e-05,
+      "loss": 1.4499,
+      "step": 50715
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9954962962962963e-05,
+      "loss": 1.5533,
+      "step": 50720
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9952e-05,
+      "loss": 1.3513,
+      "step": 50725
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9949037037037037e-05,
+      "loss": 1.4903,
+      "step": 50730
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9946074074074076e-05,
+      "loss": 1.4191,
+      "step": 50735
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.994311111111111e-05,
+      "loss": 1.4737,
+      "step": 50740
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.994014814814815e-05,
+      "loss": 1.4625,
+      "step": 50745
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9937185185185185e-05,
+      "loss": 1.5971,
+      "step": 50750
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.9934222222222224e-05,
+      "loss": 1.352,
+      "step": 50755
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.993125925925926e-05,
+      "loss": 1.5194,
+      "step": 50760
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.99282962962963e-05,
+      "loss": 1.5412,
+      "step": 50765
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9925333333333334e-05,
+      "loss": 1.5252,
+      "step": 50770
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9922370370370373e-05,
+      "loss": 1.6271,
+      "step": 50775
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9919407407407408e-05,
+      "loss": 1.41,
+      "step": 50780
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9916444444444447e-05,
+      "loss": 1.4527,
+      "step": 50785
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9913481481481482e-05,
+      "loss": 1.5011,
+      "step": 50790
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.991051851851852e-05,
+      "loss": 1.5712,
+      "step": 50795
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9907555555555557e-05,
+      "loss": 1.5773,
+      "step": 50800
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9904592592592595e-05,
+      "loss": 1.4935,
+      "step": 50805
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.990162962962963e-05,
+      "loss": 1.4199,
+      "step": 50810
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.989866666666667e-05,
+      "loss": 1.4975,
+      "step": 50815
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.9895703703703705e-05,
+      "loss": 1.4514,
+      "step": 50820
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9892740740740744e-05,
+      "loss": 1.5839,
+      "step": 50825
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.988977777777778e-05,
+      "loss": 1.4883,
+      "step": 50830
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9886814814814815e-05,
+      "loss": 1.5182,
+      "step": 50835
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9883851851851854e-05,
+      "loss": 1.6113,
+      "step": 50840
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.988088888888889e-05,
+      "loss": 1.4834,
+      "step": 50845
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9877925925925928e-05,
+      "loss": 1.5612,
+      "step": 50850
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9874962962962963e-05,
+      "loss": 1.5499,
+      "step": 50855
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9872000000000002e-05,
+      "loss": 1.4613,
+      "step": 50860
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9869037037037038e-05,
+      "loss": 1.5226,
+      "step": 50865
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9866074074074076e-05,
+      "loss": 1.5097,
+      "step": 50870
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.9863111111111112e-05,
+      "loss": 1.4149,
+      "step": 50875
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.986014814814815e-05,
+      "loss": 1.5027,
+      "step": 50880
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9857185185185186e-05,
+      "loss": 1.4902,
+      "step": 50885
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9854222222222225e-05,
+      "loss": 1.4416,
+      "step": 50890
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.985125925925926e-05,
+      "loss": 1.6142,
+      "step": 50895
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.98482962962963e-05,
+      "loss": 1.5455,
+      "step": 50900
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9845333333333335e-05,
+      "loss": 1.4468,
+      "step": 50905
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9842370370370373e-05,
+      "loss": 1.4418,
+      "step": 50910
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.983940740740741e-05,
+      "loss": 1.4279,
+      "step": 50915
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9836444444444448e-05,
+      "loss": 1.4335,
+      "step": 50920
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9833481481481483e-05,
+      "loss": 1.4673,
+      "step": 50925
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.9830518518518522e-05,
+      "loss": 1.5131,
+      "step": 50930
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9827555555555557e-05,
+      "loss": 1.5448,
+      "step": 50935
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9824592592592596e-05,
+      "loss": 1.4507,
+      "step": 50940
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.982162962962963e-05,
+      "loss": 1.5836,
+      "step": 50945
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9818666666666667e-05,
+      "loss": 1.4678,
+      "step": 50950
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9815703703703706e-05,
+      "loss": 1.4997,
+      "step": 50955
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.981274074074074e-05,
+      "loss": 1.4031,
+      "step": 50960
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.980977777777778e-05,
+      "loss": 1.5196,
+      "step": 50965
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9806814814814816e-05,
+      "loss": 1.5607,
+      "step": 50970
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9803851851851854e-05,
+      "loss": 1.5165,
+      "step": 50975
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.980088888888889e-05,
+      "loss": 1.6401,
+      "step": 50980
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.979792592592593e-05,
+      "loss": 1.5563,
+      "step": 50985
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.9794962962962964e-05,
+      "loss": 1.5452,
+      "step": 50990
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9792000000000003e-05,
+      "loss": 1.4259,
+      "step": 50995
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.978903703703704e-05,
+      "loss": 1.5068,
+      "step": 51000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9786074074074077e-05,
+      "loss": 1.474,
+      "step": 51005
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9783111111111113e-05,
+      "loss": 1.4974,
+      "step": 51010
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.978014814814815e-05,
+      "loss": 1.591,
+      "step": 51015
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9777185185185187e-05,
+      "loss": 1.5843,
+      "step": 51020
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9774222222222226e-05,
+      "loss": 1.4632,
+      "step": 51025
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.977125925925926e-05,
+      "loss": 1.4455,
+      "step": 51030
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9768296296296297e-05,
+      "loss": 1.5999,
+      "step": 51035
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.9765333333333332e-05,
+      "loss": 1.4252,
+      "step": 51040
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.976237037037037e-05,
+      "loss": 1.5969,
+      "step": 51045
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9759407407407406e-05,
+      "loss": 1.4597,
+      "step": 51050
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9756444444444445e-05,
+      "loss": 1.5726,
+      "step": 51055
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.975348148148148e-05,
+      "loss": 1.5343,
+      "step": 51060
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.975051851851852e-05,
+      "loss": 1.4491,
+      "step": 51065
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9747555555555555e-05,
+      "loss": 1.3939,
+      "step": 51070
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9744592592592594e-05,
+      "loss": 1.5018,
+      "step": 51075
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.974162962962963e-05,
+      "loss": 1.5638,
+      "step": 51080
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9738666666666668e-05,
+      "loss": 1.3192,
+      "step": 51085
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9735703703703703e-05,
+      "loss": 1.3618,
+      "step": 51090
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9732740740740742e-05,
+      "loss": 1.4601,
+      "step": 51095
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.9729777777777778e-05,
+      "loss": 1.4781,
+      "step": 51100
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9726814814814813e-05,
+      "loss": 1.5504,
+      "step": 51105
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9723851851851852e-05,
+      "loss": 1.3761,
+      "step": 51110
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9720888888888887e-05,
+      "loss": 1.5334,
+      "step": 51115
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9717925925925926e-05,
+      "loss": 1.4944,
+      "step": 51120
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.971496296296296e-05,
+      "loss": 1.4185,
+      "step": 51125
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9712e-05,
+      "loss": 1.4828,
+      "step": 51130
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9709037037037036e-05,
+      "loss": 1.4456,
+      "step": 51135
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9706074074074075e-05,
+      "loss": 1.4885,
+      "step": 51140
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.970311111111111e-05,
+      "loss": 1.5398,
+      "step": 51145
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.970014814814815e-05,
+      "loss": 1.4476,
+      "step": 51150
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.9697185185185184e-05,
+      "loss": 1.4976,
+      "step": 51155
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9694222222222223e-05,
+      "loss": 1.4727,
+      "step": 51160
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.969125925925926e-05,
+      "loss": 1.5212,
+      "step": 51165
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9688296296296297e-05,
+      "loss": 1.5441,
+      "step": 51170
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9685333333333333e-05,
+      "loss": 1.4892,
+      "step": 51175
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.968237037037037e-05,
+      "loss": 1.4377,
+      "step": 51180
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9679407407407407e-05,
+      "loss": 1.4782,
+      "step": 51185
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9676444444444446e-05,
+      "loss": 1.6596,
+      "step": 51190
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.967348148148148e-05,
+      "loss": 1.5024,
+      "step": 51195
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.967051851851852e-05,
+      "loss": 1.3915,
+      "step": 51200
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9667555555555556e-05,
+      "loss": 1.5399,
+      "step": 51205
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.9664592592592594e-05,
+      "loss": 1.3916,
+      "step": 51210
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.966162962962963e-05,
+      "loss": 1.6543,
+      "step": 51215
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9658666666666665e-05,
+      "loss": 1.4786,
+      "step": 51220
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9655703703703704e-05,
+      "loss": 1.5224,
+      "step": 51225
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.965274074074074e-05,
+      "loss": 1.5361,
+      "step": 51230
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.964977777777778e-05,
+      "loss": 1.4913,
+      "step": 51235
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9646814814814814e-05,
+      "loss": 1.4675,
+      "step": 51240
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9643851851851853e-05,
+      "loss": 1.5758,
+      "step": 51245
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9640888888888888e-05,
+      "loss": 1.5879,
+      "step": 51250
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9637925925925927e-05,
+      "loss": 1.5508,
+      "step": 51255
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9634962962962962e-05,
+      "loss": 1.4384,
+      "step": 51260
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9632e-05,
+      "loss": 1.6381,
+      "step": 51265
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.9629037037037037e-05,
+      "loss": 1.5827,
+      "step": 51270
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9626074074074075e-05,
+      "loss": 1.5454,
+      "step": 51275
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.962311111111111e-05,
+      "loss": 1.5285,
+      "step": 51280
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.962014814814815e-05,
+      "loss": 1.5665,
+      "step": 51285
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9617185185185185e-05,
+      "loss": 1.6334,
+      "step": 51290
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9614222222222224e-05,
+      "loss": 1.4272,
+      "step": 51295
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.961125925925926e-05,
+      "loss": 1.5913,
+      "step": 51300
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9608296296296298e-05,
+      "loss": 1.5112,
+      "step": 51305
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9605333333333334e-05,
+      "loss": 1.4201,
+      "step": 51310
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9602370370370372e-05,
+      "loss": 1.5154,
+      "step": 51315
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9599407407407408e-05,
+      "loss": 1.558,
+      "step": 51320
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.9596444444444447e-05,
+      "loss": 1.44,
+      "step": 51325
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9593481481481482e-05,
+      "loss": 1.5309,
+      "step": 51330
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9590518518518517e-05,
+      "loss": 1.5724,
+      "step": 51335
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9587555555555556e-05,
+      "loss": 1.4948,
+      "step": 51340
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9584592592592592e-05,
+      "loss": 1.4579,
+      "step": 51345
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.958162962962963e-05,
+      "loss": 1.5031,
+      "step": 51350
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9578666666666666e-05,
+      "loss": 1.5334,
+      "step": 51355
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9575703703703705e-05,
+      "loss": 1.6087,
+      "step": 51360
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.957274074074074e-05,
+      "loss": 1.5051,
+      "step": 51365
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.956977777777778e-05,
+      "loss": 1.5643,
+      "step": 51370
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9566814814814815e-05,
+      "loss": 1.5748,
+      "step": 51375
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.9563851851851853e-05,
+      "loss": 1.5208,
+      "step": 51380
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.956088888888889e-05,
+      "loss": 1.3993,
+      "step": 51385
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9557925925925928e-05,
+      "loss": 1.4679,
+      "step": 51390
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9554962962962963e-05,
+      "loss": 1.4178,
+      "step": 51395
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9552000000000002e-05,
+      "loss": 1.5598,
+      "step": 51400
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9549037037037037e-05,
+      "loss": 1.6522,
+      "step": 51405
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9546074074074076e-05,
+      "loss": 1.4572,
+      "step": 51410
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.954311111111111e-05,
+      "loss": 1.4571,
+      "step": 51415
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.954014814814815e-05,
+      "loss": 1.5616,
+      "step": 51420
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9537185185185186e-05,
+      "loss": 1.4767,
+      "step": 51425
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.9534222222222225e-05,
+      "loss": 1.6006,
+      "step": 51430
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.953125925925926e-05,
+      "loss": 1.4874,
+      "step": 51435
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.95282962962963e-05,
+      "loss": 1.4748,
+      "step": 51440
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9525333333333334e-05,
+      "loss": 1.5569,
+      "step": 51445
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.952237037037037e-05,
+      "loss": 1.6453,
+      "step": 51450
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.951940740740741e-05,
+      "loss": 1.4825,
+      "step": 51455
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9516444444444444e-05,
+      "loss": 1.5934,
+      "step": 51460
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9513481481481483e-05,
+      "loss": 1.5303,
+      "step": 51465
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9510518518518518e-05,
+      "loss": 1.4689,
+      "step": 51470
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9507555555555557e-05,
+      "loss": 1.491,
+      "step": 51475
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9504592592592593e-05,
+      "loss": 1.4769,
+      "step": 51480
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.950162962962963e-05,
+      "loss": 1.5491,
+      "step": 51485
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9498666666666667e-05,
+      "loss": 1.5009,
+      "step": 51490
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.9495703703703706e-05,
+      "loss": 1.4224,
+      "step": 51495
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.949274074074074e-05,
+      "loss": 1.6225,
+      "step": 51500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.948977777777778e-05,
+      "loss": 1.5768,
+      "step": 51505
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9486814814814815e-05,
+      "loss": 1.6288,
+      "step": 51510
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9483851851851854e-05,
+      "loss": 1.6132,
+      "step": 51515
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.948088888888889e-05,
+      "loss": 1.5342,
+      "step": 51520
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.947792592592593e-05,
+      "loss": 1.4958,
+      "step": 51525
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9474962962962964e-05,
+      "loss": 1.4537,
+      "step": 51530
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9472000000000003e-05,
+      "loss": 1.4955,
+      "step": 51535
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9469037037037038e-05,
+      "loss": 1.5148,
+      "step": 51540
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9466074074074077e-05,
+      "loss": 1.5618,
+      "step": 51545
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.9463111111111112e-05,
+      "loss": 1.4807,
+      "step": 51550
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.946014814814815e-05,
+      "loss": 1.5138,
+      "step": 51555
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9457185185185187e-05,
+      "loss": 1.5513,
+      "step": 51560
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9454222222222222e-05,
+      "loss": 1.4952,
+      "step": 51565
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.945125925925926e-05,
+      "loss": 1.4795,
+      "step": 51570
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9448296296296296e-05,
+      "loss": 1.5254,
+      "step": 51575
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9445333333333335e-05,
+      "loss": 1.5526,
+      "step": 51580
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.944237037037037e-05,
+      "loss": 1.5332,
+      "step": 51585
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.943940740740741e-05,
+      "loss": 1.5666,
+      "step": 51590
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9436444444444445e-05,
+      "loss": 1.6266,
+      "step": 51595
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.9433481481481484e-05,
+      "loss": 1.4862,
+      "step": 51600
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.943051851851852e-05,
+      "loss": 1.4467,
+      "step": 51605
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9427555555555558e-05,
+      "loss": 1.4954,
+      "step": 51610
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9424592592592593e-05,
+      "loss": 1.401,
+      "step": 51615
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9421629629629632e-05,
+      "loss": 1.5222,
+      "step": 51620
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9418666666666668e-05,
+      "loss": 1.5357,
+      "step": 51625
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9415703703703706e-05,
+      "loss": 1.5662,
+      "step": 51630
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9412740740740742e-05,
+      "loss": 1.4832,
+      "step": 51635
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.940977777777778e-05,
+      "loss": 1.5853,
+      "step": 51640
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9406814814814816e-05,
+      "loss": 1.6265,
+      "step": 51645
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9403851851851855e-05,
+      "loss": 1.4231,
+      "step": 51650
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.940088888888889e-05,
+      "loss": 1.5203,
+      "step": 51655
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.939792592592593e-05,
+      "loss": 1.5043,
+      "step": 51660
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.9394962962962965e-05,
+      "loss": 1.4131,
+      "step": 51665
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9392000000000003e-05,
+      "loss": 1.476,
+      "step": 51670
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.938903703703704e-05,
+      "loss": 1.5579,
+      "step": 51675
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9386074074074074e-05,
+      "loss": 1.5691,
+      "step": 51680
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9383111111111113e-05,
+      "loss": 1.4477,
+      "step": 51685
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.938014814814815e-05,
+      "loss": 1.5441,
+      "step": 51690
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9377185185185187e-05,
+      "loss": 1.4531,
+      "step": 51695
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9374222222222223e-05,
+      "loss": 1.532,
+      "step": 51700
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.937125925925926e-05,
+      "loss": 1.4588,
+      "step": 51705
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9368296296296297e-05,
+      "loss": 1.4216,
+      "step": 51710
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.9365333333333336e-05,
+      "loss": 1.5469,
+      "step": 51715
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.936237037037037e-05,
+      "loss": 1.6113,
+      "step": 51720
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.935940740740741e-05,
+      "loss": 1.5982,
+      "step": 51725
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9356444444444446e-05,
+      "loss": 1.5446,
+      "step": 51730
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9353481481481484e-05,
+      "loss": 1.5316,
+      "step": 51735
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.935051851851852e-05,
+      "loss": 1.5195,
+      "step": 51740
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.934755555555556e-05,
+      "loss": 1.5165,
+      "step": 51745
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9344592592592594e-05,
+      "loss": 1.4771,
+      "step": 51750
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9341629629629633e-05,
+      "loss": 1.5637,
+      "step": 51755
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.933866666666667e-05,
+      "loss": 1.6402,
+      "step": 51760
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9335703703703707e-05,
+      "loss": 1.5338,
+      "step": 51765
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.9332740740740743e-05,
+      "loss": 1.4786,
+      "step": 51770
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.932977777777778e-05,
+      "loss": 1.4935,
+      "step": 51775
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9326814814814817e-05,
+      "loss": 1.5779,
+      "step": 51780
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9323851851851856e-05,
+      "loss": 1.4192,
+      "step": 51785
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.932088888888889e-05,
+      "loss": 1.4549,
+      "step": 51790
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9317925925925927e-05,
+      "loss": 1.4662,
+      "step": 51795
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9314962962962965e-05,
+      "loss": 1.441,
+      "step": 51800
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9312e-05,
+      "loss": 1.5527,
+      "step": 51805
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9309037037037036e-05,
+      "loss": 1.5808,
+      "step": 51810
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9306074074074075e-05,
+      "loss": 1.4767,
+      "step": 51815
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.930311111111111e-05,
+      "loss": 1.4949,
+      "step": 51820
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.930014814814815e-05,
+      "loss": 1.5042,
+      "step": 51825
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.9297185185185185e-05,
+      "loss": 1.4731,
+      "step": 51830
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9294222222222224e-05,
+      "loss": 1.5389,
+      "step": 51835
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.929125925925926e-05,
+      "loss": 1.6636,
+      "step": 51840
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9288296296296294e-05,
+      "loss": 1.614,
+      "step": 51845
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9285333333333333e-05,
+      "loss": 1.5786,
+      "step": 51850
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.928237037037037e-05,
+      "loss": 1.4842,
+      "step": 51855
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9279407407407407e-05,
+      "loss": 1.6326,
+      "step": 51860
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9276444444444443e-05,
+      "loss": 1.5457,
+      "step": 51865
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9273481481481482e-05,
+      "loss": 1.5046,
+      "step": 51870
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9270518518518517e-05,
+      "loss": 1.5181,
+      "step": 51875
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.9267555555555556e-05,
+      "loss": 1.5673,
+      "step": 51880
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.926459259259259e-05,
+      "loss": 1.4592,
+      "step": 51885
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.926162962962963e-05,
+      "loss": 1.6332,
+      "step": 51890
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9258666666666666e-05,
+      "loss": 1.4966,
+      "step": 51895
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9255703703703705e-05,
+      "loss": 1.5761,
+      "step": 51900
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.925274074074074e-05,
+      "loss": 1.5466,
+      "step": 51905
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.924977777777778e-05,
+      "loss": 1.4861,
+      "step": 51910
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9246814814814814e-05,
+      "loss": 1.4751,
+      "step": 51915
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9243851851851853e-05,
+      "loss": 1.4734,
+      "step": 51920
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.924088888888889e-05,
+      "loss": 1.4942,
+      "step": 51925
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9237925925925927e-05,
+      "loss": 1.3831,
+      "step": 51930
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9234962962962963e-05,
+      "loss": 1.566,
+      "step": 51935
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9232e-05,
+      "loss": 1.5691,
+      "step": 51940
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.9229037037037037e-05,
+      "loss": 1.4648,
+      "step": 51945
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9226074074074076e-05,
+      "loss": 1.6115,
+      "step": 51950
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.922311111111111e-05,
+      "loss": 1.6449,
+      "step": 51955
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9220148148148147e-05,
+      "loss": 1.3774,
+      "step": 51960
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9217185185185185e-05,
+      "loss": 1.4848,
+      "step": 51965
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.921422222222222e-05,
+      "loss": 1.5612,
+      "step": 51970
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.921125925925926e-05,
+      "loss": 1.6708,
+      "step": 51975
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9208296296296295e-05,
+      "loss": 1.6228,
+      "step": 51980
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9205333333333334e-05,
+      "loss": 1.5239,
+      "step": 51985
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.920237037037037e-05,
+      "loss": 1.4633,
+      "step": 51990
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9199407407407408e-05,
+      "loss": 1.4982,
+      "step": 51995
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.9196444444444444e-05,
+      "loss": 1.5373,
+      "step": 52000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9193481481481483e-05,
+      "loss": 1.5612,
+      "step": 52005
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9190518518518518e-05,
+      "loss": 1.5117,
+      "step": 52010
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9187555555555557e-05,
+      "loss": 1.5714,
+      "step": 52015
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9184592592592592e-05,
+      "loss": 1.4357,
+      "step": 52020
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.918162962962963e-05,
+      "loss": 1.4625,
+      "step": 52025
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9178666666666666e-05,
+      "loss": 1.4898,
+      "step": 52030
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9175703703703705e-05,
+      "loss": 1.4659,
+      "step": 52035
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.917274074074074e-05,
+      "loss": 1.5357,
+      "step": 52040
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.916977777777778e-05,
+      "loss": 1.5178,
+      "step": 52045
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9166814814814815e-05,
+      "loss": 1.4517,
+      "step": 52050
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.9163851851851854e-05,
+      "loss": 1.5612,
+      "step": 52055
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.916088888888889e-05,
+      "loss": 1.5205,
+      "step": 52060
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9157925925925928e-05,
+      "loss": 1.4157,
+      "step": 52065
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9154962962962963e-05,
+      "loss": 1.5638,
+      "step": 52070
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9152e-05,
+      "loss": 1.426,
+      "step": 52075
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9149037037037038e-05,
+      "loss": 1.5309,
+      "step": 52080
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9146074074074073e-05,
+      "loss": 1.5612,
+      "step": 52085
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9143111111111112e-05,
+      "loss": 1.6217,
+      "step": 52090
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9140148148148147e-05,
+      "loss": 1.6371,
+      "step": 52095
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9137185185185186e-05,
+      "loss": 1.4951,
+      "step": 52100
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.913422222222222e-05,
+      "loss": 1.5469,
+      "step": 52105
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.913125925925926e-05,
+      "loss": 1.5426,
+      "step": 52110
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.9128296296296296e-05,
+      "loss": 1.5917,
+      "step": 52115
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9125333333333335e-05,
+      "loss": 1.5659,
+      "step": 52120
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.912237037037037e-05,
+      "loss": 1.5738,
+      "step": 52125
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.911940740740741e-05,
+      "loss": 1.5553,
+      "step": 52130
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9116444444444444e-05,
+      "loss": 1.5395,
+      "step": 52135
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9113481481481483e-05,
+      "loss": 1.4256,
+      "step": 52140
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.911051851851852e-05,
+      "loss": 1.3629,
+      "step": 52145
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9107555555555558e-05,
+      "loss": 1.5947,
+      "step": 52150
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9104592592592593e-05,
+      "loss": 1.4103,
+      "step": 52155
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9101629629629632e-05,
+      "loss": 1.651,
+      "step": 52160
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9098666666666667e-05,
+      "loss": 1.5684,
+      "step": 52165
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.9095703703703706e-05,
+      "loss": 1.5609,
+      "step": 52170
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.909274074074074e-05,
+      "loss": 1.645,
+      "step": 52175
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.908977777777778e-05,
+      "loss": 1.617,
+      "step": 52180
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9086814814814816e-05,
+      "loss": 1.5013,
+      "step": 52185
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.908385185185185e-05,
+      "loss": 1.5077,
+      "step": 52190
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.908088888888889e-05,
+      "loss": 1.4899,
+      "step": 52195
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9077925925925925e-05,
+      "loss": 1.6514,
+      "step": 52200
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9074962962962964e-05,
+      "loss": 1.5563,
+      "step": 52205
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9072e-05,
+      "loss": 1.5662,
+      "step": 52210
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.906903703703704e-05,
+      "loss": 1.5416,
+      "step": 52215
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9066074074074074e-05,
+      "loss": 1.4595,
+      "step": 52220
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.9063111111111113e-05,
+      "loss": 1.4963,
+      "step": 52225
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9060148148148148e-05,
+      "loss": 1.5882,
+      "step": 52230
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9057185185185187e-05,
+      "loss": 1.5429,
+      "step": 52235
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9054222222222222e-05,
+      "loss": 1.5742,
+      "step": 52240
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.905125925925926e-05,
+      "loss": 1.5464,
+      "step": 52245
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9048296296296297e-05,
+      "loss": 1.4868,
+      "step": 52250
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9045333333333336e-05,
+      "loss": 1.5181,
+      "step": 52255
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.904237037037037e-05,
+      "loss": 1.5181,
+      "step": 52260
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.903940740740741e-05,
+      "loss": 1.5412,
+      "step": 52265
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9036444444444445e-05,
+      "loss": 1.5758,
+      "step": 52270
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.9033481481481484e-05,
+      "loss": 1.5296,
+      "step": 52275
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.903051851851852e-05,
+      "loss": 1.564,
+      "step": 52280
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9027555555555558e-05,
+      "loss": 1.552,
+      "step": 52285
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9024592592592594e-05,
+      "loss": 1.5129,
+      "step": 52290
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9021629629629633e-05,
+      "loss": 1.4322,
+      "step": 52295
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9018666666666668e-05,
+      "loss": 1.5483,
+      "step": 52300
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9015703703703703e-05,
+      "loss": 1.3346,
+      "step": 52305
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9012740740740742e-05,
+      "loss": 1.5272,
+      "step": 52310
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9009777777777778e-05,
+      "loss": 1.5165,
+      "step": 52315
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9006814814814816e-05,
+      "loss": 1.6013,
+      "step": 52320
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.9003851851851852e-05,
+      "loss": 1.5328,
+      "step": 52325
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.900088888888889e-05,
+      "loss": 1.4681,
+      "step": 52330
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.8997925925925926e-05,
+      "loss": 1.5509,
+      "step": 52335
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.8994962962962965e-05,
+      "loss": 1.5472,
+      "step": 52340
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8992e-05,
+      "loss": 1.4159,
+      "step": 52345
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.898903703703704e-05,
+      "loss": 1.4428,
+      "step": 52350
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8986074074074075e-05,
+      "loss": 1.4519,
+      "step": 52355
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8983111111111114e-05,
+      "loss": 1.5064,
+      "step": 52360
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.898014814814815e-05,
+      "loss": 1.5235,
+      "step": 52365
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8977185185185188e-05,
+      "loss": 1.5598,
+      "step": 52370
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8974222222222223e-05,
+      "loss": 1.5391,
+      "step": 52375
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8971259259259262e-05,
+      "loss": 1.4634,
+      "step": 52380
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8968296296296297e-05,
+      "loss": 1.5141,
+      "step": 52385
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8965333333333336e-05,
+      "loss": 1.4912,
+      "step": 52390
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.8962370370370372e-05,
+      "loss": 1.6172,
+      "step": 52395
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.895940740740741e-05,
+      "loss": 1.5,
+      "step": 52400
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8956444444444446e-05,
+      "loss": 1.4518,
+      "step": 52405
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8953481481481485e-05,
+      "loss": 1.4996,
+      "step": 52410
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.895051851851852e-05,
+      "loss": 1.6303,
+      "step": 52415
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8947555555555556e-05,
+      "loss": 1.5302,
+      "step": 52420
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8944592592592594e-05,
+      "loss": 1.5104,
+      "step": 52425
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.894162962962963e-05,
+      "loss": 1.462,
+      "step": 52430
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.893866666666667e-05,
+      "loss": 1.5999,
+      "step": 52435
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8935703703703704e-05,
+      "loss": 1.5093,
+      "step": 52440
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.8932740740740743e-05,
+      "loss": 1.4872,
+      "step": 52445
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.892977777777778e-05,
+      "loss": 1.4335,
+      "step": 52450
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8926814814814817e-05,
+      "loss": 1.3709,
+      "step": 52455
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8923851851851853e-05,
+      "loss": 1.4729,
+      "step": 52460
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.892088888888889e-05,
+      "loss": 1.5973,
+      "step": 52465
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8917925925925927e-05,
+      "loss": 1.5374,
+      "step": 52470
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8914962962962966e-05,
+      "loss": 1.5687,
+      "step": 52475
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8912e-05,
+      "loss": 1.4101,
+      "step": 52480
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.890903703703704e-05,
+      "loss": 1.5153,
+      "step": 52485
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8906074074074075e-05,
+      "loss": 1.5439,
+      "step": 52490
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.8903111111111114e-05,
+      "loss": 1.5347,
+      "step": 52495
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.890014814814815e-05,
+      "loss": 1.5496,
+      "step": 52500
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.889718518518519e-05,
+      "loss": 1.5483,
+      "step": 52505
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8894222222222224e-05,
+      "loss": 1.5495,
+      "step": 52510
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8891259259259263e-05,
+      "loss": 1.4581,
+      "step": 52515
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8888296296296298e-05,
+      "loss": 1.3663,
+      "step": 52520
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8885333333333337e-05,
+      "loss": 1.3726,
+      "step": 52525
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8882370370370372e-05,
+      "loss": 1.4894,
+      "step": 52530
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8879407407407408e-05,
+      "loss": 1.67,
+      "step": 52535
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8876444444444447e-05,
+      "loss": 1.5658,
+      "step": 52540
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8873481481481482e-05,
+      "loss": 1.5714,
+      "step": 52545
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.887051851851852e-05,
+      "loss": 1.4833,
+      "step": 52550
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8867555555555556e-05,
+      "loss": 1.4873,
+      "step": 52555
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.8864592592592595e-05,
+      "loss": 1.4935,
+      "step": 52560
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.886162962962963e-05,
+      "loss": 1.4882,
+      "step": 52565
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.885866666666667e-05,
+      "loss": 1.4436,
+      "step": 52570
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8855703703703705e-05,
+      "loss": 1.5769,
+      "step": 52575
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.885274074074074e-05,
+      "loss": 1.5633,
+      "step": 52580
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8849777777777776e-05,
+      "loss": 1.5372,
+      "step": 52585
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8846814814814815e-05,
+      "loss": 1.5765,
+      "step": 52590
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.884385185185185e-05,
+      "loss": 1.5579,
+      "step": 52595
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.884088888888889e-05,
+      "loss": 1.4854,
+      "step": 52600
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8837925925925924e-05,
+      "loss": 1.5647,
+      "step": 52605
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8834962962962963e-05,
+      "loss": 1.5512,
+      "step": 52610
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8832e-05,
+      "loss": 1.4442,
+      "step": 52615
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.8829037037037037e-05,
+      "loss": 1.5132,
+      "step": 52620
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8826074074074073e-05,
+      "loss": 1.4249,
+      "step": 52625
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.882311111111111e-05,
+      "loss": 1.377,
+      "step": 52630
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8820148148148147e-05,
+      "loss": 1.5298,
+      "step": 52635
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8817185185185186e-05,
+      "loss": 1.5412,
+      "step": 52640
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.881422222222222e-05,
+      "loss": 1.4337,
+      "step": 52645
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.881125925925926e-05,
+      "loss": 1.6483,
+      "step": 52650
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8808296296296296e-05,
+      "loss": 1.4188,
+      "step": 52655
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8805333333333334e-05,
+      "loss": 1.547,
+      "step": 52660
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.880237037037037e-05,
+      "loss": 1.5614,
+      "step": 52665
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.879940740740741e-05,
+      "loss": 1.5169,
+      "step": 52670
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.8796444444444444e-05,
+      "loss": 1.5795,
+      "step": 52675
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8793481481481483e-05,
+      "loss": 1.6319,
+      "step": 52680
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.879051851851852e-05,
+      "loss": 1.5922,
+      "step": 52685
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8787555555555554e-05,
+      "loss": 1.5194,
+      "step": 52690
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8784592592592593e-05,
+      "loss": 1.4209,
+      "step": 52695
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8781629629629628e-05,
+      "loss": 1.4611,
+      "step": 52700
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8778666666666667e-05,
+      "loss": 1.5642,
+      "step": 52705
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8775703703703702e-05,
+      "loss": 1.5813,
+      "step": 52710
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.877274074074074e-05,
+      "loss": 1.4425,
+      "step": 52715
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8769777777777777e-05,
+      "loss": 1.4816,
+      "step": 52720
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.8766814814814815e-05,
+      "loss": 1.5387,
+      "step": 52725
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.876385185185185e-05,
+      "loss": 1.6155,
+      "step": 52730
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.876088888888889e-05,
+      "loss": 1.6089,
+      "step": 52735
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8757925925925925e-05,
+      "loss": 1.4769,
+      "step": 52740
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8754962962962964e-05,
+      "loss": 1.5734,
+      "step": 52745
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8752e-05,
+      "loss": 1.458,
+      "step": 52750
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8749037037037038e-05,
+      "loss": 1.5676,
+      "step": 52755
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8746074074074074e-05,
+      "loss": 1.4887,
+      "step": 52760
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8743111111111112e-05,
+      "loss": 1.4885,
+      "step": 52765
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8740148148148148e-05,
+      "loss": 1.5568,
+      "step": 52770
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8737185185185187e-05,
+      "loss": 1.466,
+      "step": 52775
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8734222222222222e-05,
+      "loss": 1.5664,
+      "step": 52780
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.873125925925926e-05,
+      "loss": 1.4928,
+      "step": 52785
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.8728296296296296e-05,
+      "loss": 1.4886,
+      "step": 52790
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8725333333333335e-05,
+      "loss": 1.4927,
+      "step": 52795
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8722962962962965e-05,
+      "loss": 1.4207,
+      "step": 52800
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.872e-05,
+      "loss": 1.6124,
+      "step": 52805
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.871703703703704e-05,
+      "loss": 1.4225,
+      "step": 52810
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8714074074074075e-05,
+      "loss": 1.5204,
+      "step": 52815
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8711111111111113e-05,
+      "loss": 1.5392,
+      "step": 52820
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.870814814814815e-05,
+      "loss": 1.5742,
+      "step": 52825
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8705185185185188e-05,
+      "loss": 1.5549,
+      "step": 52830
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8702222222222223e-05,
+      "loss": 1.5418,
+      "step": 52835
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8699259259259262e-05,
+      "loss": 1.5664,
+      "step": 52840
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.8696296296296297e-05,
+      "loss": 1.4793,
+      "step": 52845
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8693333333333336e-05,
+      "loss": 1.6126,
+      "step": 52850
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.869037037037037e-05,
+      "loss": 1.4944,
+      "step": 52855
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.868740740740741e-05,
+      "loss": 1.3641,
+      "step": 52860
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8684444444444446e-05,
+      "loss": 1.4965,
+      "step": 52865
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8681481481481485e-05,
+      "loss": 1.488,
+      "step": 52870
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.867851851851852e-05,
+      "loss": 1.5778,
+      "step": 52875
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8675555555555556e-05,
+      "loss": 1.6234,
+      "step": 52880
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8672592592592594e-05,
+      "loss": 1.5703,
+      "step": 52885
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.866962962962963e-05,
+      "loss": 1.4391,
+      "step": 52890
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 1.5961,
+      "step": 52895
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8663703703703704e-05,
+      "loss": 1.4264,
+      "step": 52900
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8660740740740743e-05,
+      "loss": 1.6054,
+      "step": 52905
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.865777777777778e-05,
+      "loss": 1.626,
+      "step": 52910
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8654814814814817e-05,
+      "loss": 1.5561,
+      "step": 52915
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8651851851851853e-05,
+      "loss": 1.4536,
+      "step": 52920
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.864888888888889e-05,
+      "loss": 1.5781,
+      "step": 52925
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8645925925925927e-05,
+      "loss": 1.5282,
+      "step": 52930
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8642962962962966e-05,
+      "loss": 1.6391,
+      "step": 52935
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.864e-05,
+      "loss": 1.6071,
+      "step": 52940
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.863703703703704e-05,
+      "loss": 1.6976,
+      "step": 52945
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8634074074074075e-05,
+      "loss": 1.5425,
+      "step": 52950
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8631111111111114e-05,
+      "loss": 1.5432,
+      "step": 52955
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.862814814814815e-05,
+      "loss": 1.5149,
+      "step": 52960
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.862518518518519e-05,
+      "loss": 1.5467,
+      "step": 52965
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8622222222222224e-05,
+      "loss": 1.5827,
+      "step": 52970
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8619259259259263e-05,
+      "loss": 1.4332,
+      "step": 52975
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8616296296296298e-05,
+      "loss": 1.5034,
+      "step": 52980
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8613333333333337e-05,
+      "loss": 1.4552,
+      "step": 52985
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8610370370370372e-05,
+      "loss": 1.4742,
+      "step": 52990
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8607407407407408e-05,
+      "loss": 1.5992,
+      "step": 52995
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8604444444444447e-05,
+      "loss": 1.4624,
+      "step": 53000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8601481481481482e-05,
+      "loss": 1.5355,
+      "step": 53005
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.859851851851852e-05,
+      "loss": 1.5451,
+      "step": 53010
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.8595555555555556e-05,
+      "loss": 1.3983,
+      "step": 53015
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8592592592592595e-05,
+      "loss": 1.5402,
+      "step": 53020
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.858962962962963e-05,
+      "loss": 1.5319,
+      "step": 53025
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.858666666666667e-05,
+      "loss": 1.4729,
+      "step": 53030
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8583703703703705e-05,
+      "loss": 1.6524,
+      "step": 53035
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8580740740740744e-05,
+      "loss": 1.5144,
+      "step": 53040
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8577777777777776e-05,
+      "loss": 1.5958,
+      "step": 53045
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8574814814814815e-05,
+      "loss": 1.418,
+      "step": 53050
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.857185185185185e-05,
+      "loss": 1.4525,
+      "step": 53055
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.856888888888889e-05,
+      "loss": 1.5126,
+      "step": 53060
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8565925925925924e-05,
+      "loss": 1.5664,
+      "step": 53065
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.8562962962962963e-05,
+      "loss": 1.4276,
+      "step": 53070
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.856e-05,
+      "loss": 1.4984,
+      "step": 53075
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8557037037037037e-05,
+      "loss": 1.6379,
+      "step": 53080
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8554074074074073e-05,
+      "loss": 1.6601,
+      "step": 53085
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.855111111111111e-05,
+      "loss": 1.6103,
+      "step": 53090
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8548148148148147e-05,
+      "loss": 1.5087,
+      "step": 53095
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8545185185185186e-05,
+      "loss": 1.529,
+      "step": 53100
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.854222222222222e-05,
+      "loss": 1.5469,
+      "step": 53105
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.853925925925926e-05,
+      "loss": 1.6424,
+      "step": 53110
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8536296296296296e-05,
+      "loss": 1.3707,
+      "step": 53115
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.8533333333333334e-05,
+      "loss": 1.4565,
+      "step": 53120
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.853037037037037e-05,
+      "loss": 1.5717,
+      "step": 53125
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.852740740740741e-05,
+      "loss": 1.5266,
+      "step": 53130
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8524444444444444e-05,
+      "loss": 1.6055,
+      "step": 53135
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8521481481481483e-05,
+      "loss": 1.5165,
+      "step": 53140
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 1.5523,
+      "step": 53145
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8515555555555557e-05,
+      "loss": 1.5659,
+      "step": 53150
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8512592592592593e-05,
+      "loss": 1.5603,
+      "step": 53155
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8509629629629628e-05,
+      "loss": 1.5647,
+      "step": 53160
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8506666666666667e-05,
+      "loss": 1.4642,
+      "step": 53165
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8503703703703702e-05,
+      "loss": 1.6391,
+      "step": 53170
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.850074074074074e-05,
+      "loss": 1.5329,
+      "step": 53175
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.8497777777777776e-05,
+      "loss": 1.4352,
+      "step": 53180
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8494814814814815e-05,
+      "loss": 1.6342,
+      "step": 53185
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.849185185185185e-05,
+      "loss": 1.4883,
+      "step": 53190
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.848888888888889e-05,
+      "loss": 1.5331,
+      "step": 53195
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8485925925925925e-05,
+      "loss": 1.4351,
+      "step": 53200
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8482962962962964e-05,
+      "loss": 1.5567,
+      "step": 53205
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.848e-05,
+      "loss": 1.5401,
+      "step": 53210
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8477037037037038e-05,
+      "loss": 1.5085,
+      "step": 53215
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8474074074074074e-05,
+      "loss": 1.4168,
+      "step": 53220
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8471111111111112e-05,
+      "loss": 1.568,
+      "step": 53225
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8468148148148148e-05,
+      "loss": 1.4346,
+      "step": 53230
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8465185185185187e-05,
+      "loss": 1.5355,
+      "step": 53235
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.8462222222222222e-05,
+      "loss": 1.5787,
+      "step": 53240
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.845925925925926e-05,
+      "loss": 1.554,
+      "step": 53245
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8456296296296296e-05,
+      "loss": 1.4237,
+      "step": 53250
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8453333333333335e-05,
+      "loss": 1.5452,
+      "step": 53255
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.845037037037037e-05,
+      "loss": 1.4849,
+      "step": 53260
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.844740740740741e-05,
+      "loss": 1.5303,
+      "step": 53265
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8444444444444445e-05,
+      "loss": 1.6177,
+      "step": 53270
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.844148148148148e-05,
+      "loss": 1.5961,
+      "step": 53275
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.843851851851852e-05,
+      "loss": 1.4775,
+      "step": 53280
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8435555555555554e-05,
+      "loss": 1.4317,
+      "step": 53285
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.8432592592592593e-05,
+      "loss": 1.6531,
+      "step": 53290
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.842962962962963e-05,
+      "loss": 1.5542,
+      "step": 53295
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8426666666666668e-05,
+      "loss": 1.7048,
+      "step": 53300
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8423703703703703e-05,
+      "loss": 1.6273,
+      "step": 53305
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8420740740740742e-05,
+      "loss": 1.4796,
+      "step": 53310
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8417777777777777e-05,
+      "loss": 1.5795,
+      "step": 53315
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8414814814814816e-05,
+      "loss": 1.5544,
+      "step": 53320
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.841185185185185e-05,
+      "loss": 1.561,
+      "step": 53325
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.840888888888889e-05,
+      "loss": 1.5616,
+      "step": 53330
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8405925925925926e-05,
+      "loss": 1.5686,
+      "step": 53335
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.8402962962962965e-05,
+      "loss": 1.5136,
+      "step": 53340
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.84e-05,
+      "loss": 1.4563,
+      "step": 53345
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.839703703703704e-05,
+      "loss": 1.5079,
+      "step": 53350
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8394074074074074e-05,
+      "loss": 1.5759,
+      "step": 53355
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8391111111111113e-05,
+      "loss": 1.3618,
+      "step": 53360
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.838814814814815e-05,
+      "loss": 1.5627,
+      "step": 53365
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8385185185185187e-05,
+      "loss": 1.4832,
+      "step": 53370
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8382222222222223e-05,
+      "loss": 1.5932,
+      "step": 53375
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.837925925925926e-05,
+      "loss": 1.4519,
+      "step": 53380
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8376296296296297e-05,
+      "loss": 1.52,
+      "step": 53385
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8373333333333332e-05,
+      "loss": 1.4891,
+      "step": 53390
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 1.5784,
+      "step": 53395
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8367407407407407e-05,
+      "loss": 1.4897,
+      "step": 53400
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.8364444444444446e-05,
+      "loss": 1.4513,
+      "step": 53405
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.836148148148148e-05,
+      "loss": 1.5744,
+      "step": 53410
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.835851851851852e-05,
+      "loss": 1.5402,
+      "step": 53415
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8355555555555555e-05,
+      "loss": 1.4784,
+      "step": 53420
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8352592592592594e-05,
+      "loss": 1.5556,
+      "step": 53425
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.834962962962963e-05,
+      "loss": 1.5255,
+      "step": 53430
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.834666666666667e-05,
+      "loss": 1.7009,
+      "step": 53435
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8343703703703704e-05,
+      "loss": 1.6039,
+      "step": 53440
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8340740740740743e-05,
+      "loss": 1.6701,
+      "step": 53445
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8337777777777778e-05,
+      "loss": 1.4897,
+      "step": 53450
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8334814814814817e-05,
+      "loss": 1.4417,
+      "step": 53455
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.8331851851851852e-05,
+      "loss": 1.5649,
+      "step": 53460
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.832888888888889e-05,
+      "loss": 1.5815,
+      "step": 53465
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8325925925925927e-05,
+      "loss": 1.5448,
+      "step": 53470
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8322962962962965e-05,
+      "loss": 1.4661,
+      "step": 53475
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.832e-05,
+      "loss": 1.464,
+      "step": 53480
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.831703703703704e-05,
+      "loss": 1.4425,
+      "step": 53485
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8314074074074075e-05,
+      "loss": 1.6214,
+      "step": 53490
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8311111111111114e-05,
+      "loss": 1.4337,
+      "step": 53495
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.830814814814815e-05,
+      "loss": 1.4801,
+      "step": 53500
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8305185185185185e-05,
+      "loss": 1.5248,
+      "step": 53505
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8302222222222224e-05,
+      "loss": 1.5553,
+      "step": 53510
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.829925925925926e-05,
+      "loss": 1.4498,
+      "step": 53515
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.8296296296296298e-05,
+      "loss": 1.538,
+      "step": 53520
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8293333333333333e-05,
+      "loss": 1.471,
+      "step": 53525
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8290370370370372e-05,
+      "loss": 1.5087,
+      "step": 53530
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8287407407407407e-05,
+      "loss": 1.4605,
+      "step": 53535
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8284444444444446e-05,
+      "loss": 1.4719,
+      "step": 53540
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8281481481481482e-05,
+      "loss": 1.5288,
+      "step": 53545
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.827851851851852e-05,
+      "loss": 1.4695,
+      "step": 53550
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8275555555555556e-05,
+      "loss": 1.4701,
+      "step": 53555
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8272592592592595e-05,
+      "loss": 1.5227,
+      "step": 53560
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.826962962962963e-05,
+      "loss": 1.5101,
+      "step": 53565
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.826666666666667e-05,
+      "loss": 1.4663,
+      "step": 53570
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.8263703703703705e-05,
+      "loss": 1.6422,
+      "step": 53575
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8260740740740743e-05,
+      "loss": 1.6376,
+      "step": 53580
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.825777777777778e-05,
+      "loss": 1.5237,
+      "step": 53585
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8254814814814818e-05,
+      "loss": 1.4501,
+      "step": 53590
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8251851851851853e-05,
+      "loss": 1.5463,
+      "step": 53595
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8248888888888892e-05,
+      "loss": 1.5614,
+      "step": 53600
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8245925925925927e-05,
+      "loss": 1.5523,
+      "step": 53605
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8242962962962966e-05,
+      "loss": 1.4974,
+      "step": 53610
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.824e-05,
+      "loss": 1.5042,
+      "step": 53615
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8237037037037037e-05,
+      "loss": 1.5366,
+      "step": 53620
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.8234074074074076e-05,
+      "loss": 1.5085,
+      "step": 53625
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.823111111111111e-05,
+      "loss": 1.4552,
+      "step": 53630
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.822814814814815e-05,
+      "loss": 1.5201,
+      "step": 53635
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8225185185185185e-05,
+      "loss": 1.5323,
+      "step": 53640
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8222222222222224e-05,
+      "loss": 1.5352,
+      "step": 53645
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.821925925925926e-05,
+      "loss": 1.4549,
+      "step": 53650
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.82162962962963e-05,
+      "loss": 1.4525,
+      "step": 53655
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8213333333333334e-05,
+      "loss": 1.4004,
+      "step": 53660
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8210370370370373e-05,
+      "loss": 1.6647,
+      "step": 53665
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8207407407407408e-05,
+      "loss": 1.4208,
+      "step": 53670
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8204444444444447e-05,
+      "loss": 1.4308,
+      "step": 53675
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8201481481481483e-05,
+      "loss": 1.5211,
+      "step": 53680
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.819851851851852e-05,
+      "loss": 1.621,
+      "step": 53685
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.8195555555555557e-05,
+      "loss": 1.5202,
+      "step": 53690
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8192592592592596e-05,
+      "loss": 1.4863,
+      "step": 53695
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.818962962962963e-05,
+      "loss": 1.5894,
+      "step": 53700
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.818666666666667e-05,
+      "loss": 1.5244,
+      "step": 53705
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8183703703703705e-05,
+      "loss": 1.5203,
+      "step": 53710
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8180740740740744e-05,
+      "loss": 1.5583,
+      "step": 53715
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.817777777777778e-05,
+      "loss": 1.5182,
+      "step": 53720
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.817481481481482e-05,
+      "loss": 1.5578,
+      "step": 53725
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8171851851851854e-05,
+      "loss": 1.5833,
+      "step": 53730
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.816888888888889e-05,
+      "loss": 1.5828,
+      "step": 53735
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8165925925925928e-05,
+      "loss": 1.4147,
+      "step": 53740
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.8162962962962963e-05,
+      "loss": 1.4974,
+      "step": 53745
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 1.4519,
+      "step": 53750
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8157037037037038e-05,
+      "loss": 1.5565,
+      "step": 53755
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8154074074074077e-05,
+      "loss": 1.4758,
+      "step": 53760
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8151111111111112e-05,
+      "loss": 1.5015,
+      "step": 53765
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 1.5439,
+      "step": 53770
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8145185185185186e-05,
+      "loss": 1.4211,
+      "step": 53775
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8142222222222225e-05,
+      "loss": 1.5991,
+      "step": 53780
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.813925925925926e-05,
+      "loss": 1.5673,
+      "step": 53785
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.81362962962963e-05,
+      "loss": 1.5168,
+      "step": 53790
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8133333333333335e-05,
+      "loss": 1.6075,
+      "step": 53795
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.8130370370370374e-05,
+      "loss": 1.5046,
+      "step": 53800
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8128e-05,
+      "loss": 1.536,
+      "step": 53805
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.812503703703704e-05,
+      "loss": 1.6326,
+      "step": 53810
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8122074074074074e-05,
+      "loss": 1.6135,
+      "step": 53815
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8119111111111113e-05,
+      "loss": 1.5267,
+      "step": 53820
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.811614814814815e-05,
+      "loss": 1.4672,
+      "step": 53825
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8113185185185187e-05,
+      "loss": 1.423,
+      "step": 53830
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8110222222222223e-05,
+      "loss": 1.4972,
+      "step": 53835
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.810725925925926e-05,
+      "loss": 1.5916,
+      "step": 53840
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8104296296296297e-05,
+      "loss": 1.4925,
+      "step": 53845
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.8101333333333332e-05,
+      "loss": 1.4897,
+      "step": 53850
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.809837037037037e-05,
+      "loss": 1.5223,
+      "step": 53855
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8095407407407407e-05,
+      "loss": 1.5351,
+      "step": 53860
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8092444444444445e-05,
+      "loss": 1.5373,
+      "step": 53865
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.808948148148148e-05,
+      "loss": 1.5604,
+      "step": 53870
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.808651851851852e-05,
+      "loss": 1.4046,
+      "step": 53875
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8083555555555555e-05,
+      "loss": 1.5217,
+      "step": 53880
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8080592592592594e-05,
+      "loss": 1.5286,
+      "step": 53885
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.807762962962963e-05,
+      "loss": 1.5288,
+      "step": 53890
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8074666666666668e-05,
+      "loss": 1.4257,
+      "step": 53895
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8071703703703704e-05,
+      "loss": 1.5695,
+      "step": 53900
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8068740740740742e-05,
+      "loss": 1.5515,
+      "step": 53905
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8065777777777778e-05,
+      "loss": 1.4966,
+      "step": 53910
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.8062814814814817e-05,
+      "loss": 1.5215,
+      "step": 53915
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8059851851851852e-05,
+      "loss": 1.5538,
+      "step": 53920
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.805688888888889e-05,
+      "loss": 1.5105,
+      "step": 53925
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8053925925925926e-05,
+      "loss": 1.4481,
+      "step": 53930
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8050962962962965e-05,
+      "loss": 1.4785,
+      "step": 53935
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8048e-05,
+      "loss": 1.5378,
+      "step": 53940
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.804503703703704e-05,
+      "loss": 1.5608,
+      "step": 53945
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8042074074074075e-05,
+      "loss": 1.4849,
+      "step": 53950
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8039111111111114e-05,
+      "loss": 1.6106,
+      "step": 53955
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.803614814814815e-05,
+      "loss": 1.4693,
+      "step": 53960
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8033185185185185e-05,
+      "loss": 1.4601,
+      "step": 53965
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.8030222222222223e-05,
+      "loss": 1.567,
+      "step": 53970
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.802725925925926e-05,
+      "loss": 1.5337,
+      "step": 53975
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8024296296296298e-05,
+      "loss": 1.5243,
+      "step": 53980
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8021333333333333e-05,
+      "loss": 1.4769,
+      "step": 53985
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8018370370370372e-05,
+      "loss": 1.5179,
+      "step": 53990
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8015407407407407e-05,
+      "loss": 1.4569,
+      "step": 53995
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8012444444444446e-05,
+      "loss": 1.5211,
+      "step": 54000
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.800948148148148e-05,
+      "loss": 1.5762,
+      "step": 54005
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.800651851851852e-05,
+      "loss": 1.5771,
+      "step": 54010
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8003555555555556e-05,
+      "loss": 1.4852,
+      "step": 54015
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.8000592592592595e-05,
+      "loss": 1.5645,
+      "step": 54020
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.799762962962963e-05,
+      "loss": 1.5083,
+      "step": 54025
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.799466666666667e-05,
+      "loss": 1.4824,
+      "step": 54030
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7991703703703704e-05,
+      "loss": 1.6246,
+      "step": 54035
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7988740740740743e-05,
+      "loss": 1.4044,
+      "step": 54040
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.798577777777778e-05,
+      "loss": 1.5695,
+      "step": 54045
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7982814814814817e-05,
+      "loss": 1.5076,
+      "step": 54050
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7979851851851853e-05,
+      "loss": 1.4996,
+      "step": 54055
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7976888888888892e-05,
+      "loss": 1.553,
+      "step": 54060
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7973925925925927e-05,
+      "loss": 1.4756,
+      "step": 54065
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7970962962962966e-05,
+      "loss": 1.5169,
+      "step": 54070
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7968e-05,
+      "loss": 1.6612,
+      "step": 54075
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.7965037037037037e-05,
+      "loss": 1.4332,
+      "step": 54080
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7962074074074076e-05,
+      "loss": 1.5928,
+      "step": 54085
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.795911111111111e-05,
+      "loss": 1.5339,
+      "step": 54090
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.795614814814815e-05,
+      "loss": 1.4681,
+      "step": 54095
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7953185185185185e-05,
+      "loss": 1.4833,
+      "step": 54100
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7950222222222224e-05,
+      "loss": 1.5484,
+      "step": 54105
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.794725925925926e-05,
+      "loss": 1.4835,
+      "step": 54110
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.79442962962963e-05,
+      "loss": 1.5865,
+      "step": 54115
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7941333333333334e-05,
+      "loss": 1.4044,
+      "step": 54120
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7938370370370373e-05,
+      "loss": 1.5587,
+      "step": 54125
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7935407407407408e-05,
+      "loss": 1.4233,
+      "step": 54130
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7932444444444447e-05,
+      "loss": 1.6024,
+      "step": 54135
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.7929481481481482e-05,
+      "loss": 1.4617,
+      "step": 54140
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.792651851851852e-05,
+      "loss": 1.4568,
+      "step": 54145
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7923555555555557e-05,
+      "loss": 1.5017,
+      "step": 54150
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7920592592592595e-05,
+      "loss": 1.5826,
+      "step": 54155
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.791762962962963e-05,
+      "loss": 1.5808,
+      "step": 54160
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.791466666666667e-05,
+      "loss": 1.4591,
+      "step": 54165
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7911703703703705e-05,
+      "loss": 1.5615,
+      "step": 54170
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7908740740740744e-05,
+      "loss": 1.4873,
+      "step": 54175
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.790577777777778e-05,
+      "loss": 1.5246,
+      "step": 54180
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7902814814814818e-05,
+      "loss": 1.4016,
+      "step": 54185
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.7899851851851854e-05,
+      "loss": 1.6419,
+      "step": 54190
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.789688888888889e-05,
+      "loss": 1.4813,
+      "step": 54195
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7893925925925928e-05,
+      "loss": 1.5284,
+      "step": 54200
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7890962962962963e-05,
+      "loss": 1.4954,
+      "step": 54205
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7888000000000002e-05,
+      "loss": 1.4813,
+      "step": 54210
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7885037037037038e-05,
+      "loss": 1.5154,
+      "step": 54215
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7882074074074076e-05,
+      "loss": 1.5826,
+      "step": 54220
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7879111111111112e-05,
+      "loss": 1.5873,
+      "step": 54225
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.787614814814815e-05,
+      "loss": 1.546,
+      "step": 54230
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7873185185185186e-05,
+      "loss": 1.5245,
+      "step": 54235
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7870222222222225e-05,
+      "loss": 1.4937,
+      "step": 54240
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.786725925925926e-05,
+      "loss": 1.608,
+      "step": 54245
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.78642962962963e-05,
+      "loss": 1.5129,
+      "step": 54250
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7861333333333335e-05,
+      "loss": 1.5883,
+      "step": 54255
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7858370370370373e-05,
+      "loss": 1.5027,
+      "step": 54260
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.785540740740741e-05,
+      "loss": 1.6329,
+      "step": 54265
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7852444444444448e-05,
+      "loss": 1.4551,
+      "step": 54270
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7849481481481483e-05,
+      "loss": 1.5784,
+      "step": 54275
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7846518518518522e-05,
+      "loss": 1.4644,
+      "step": 54280
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7843555555555554e-05,
+      "loss": 1.5849,
+      "step": 54285
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7840592592592593e-05,
+      "loss": 1.5397,
+      "step": 54290
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7837629629629628e-05,
+      "loss": 1.5748,
+      "step": 54295
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7834666666666667e-05,
+      "loss": 1.5966,
+      "step": 54300
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7831703703703703e-05,
+      "loss": 1.4973,
+      "step": 54305
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.782874074074074e-05,
+      "loss": 1.5531,
+      "step": 54310
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7825777777777777e-05,
+      "loss": 1.5604,
+      "step": 54315
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7822814814814816e-05,
+      "loss": 1.4931,
+      "step": 54320
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.781985185185185e-05,
+      "loss": 1.6302,
+      "step": 54325
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.781688888888889e-05,
+      "loss": 1.5227,
+      "step": 54330
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7813925925925925e-05,
+      "loss": 1.5484,
+      "step": 54335
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7810962962962964e-05,
+      "loss": 1.5015,
+      "step": 54340
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7808e-05,
+      "loss": 1.7295,
+      "step": 54345
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.780503703703704e-05,
+      "loss": 1.6659,
+      "step": 54350
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7802074074074074e-05,
+      "loss": 1.435,
+      "step": 54355
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.779911111111111e-05,
+      "loss": 1.4806,
+      "step": 54360
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.7796148148148148e-05,
+      "loss": 1.4258,
+      "step": 54365
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7793185185185184e-05,
+      "loss": 1.6189,
+      "step": 54370
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7790222222222222e-05,
+      "loss": 1.4716,
+      "step": 54375
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7787259259259258e-05,
+      "loss": 1.4184,
+      "step": 54380
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7784296296296297e-05,
+      "loss": 1.5484,
+      "step": 54385
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7781333333333332e-05,
+      "loss": 1.4738,
+      "step": 54390
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.777837037037037e-05,
+      "loss": 1.5376,
+      "step": 54395
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7775407407407406e-05,
+      "loss": 1.5581,
+      "step": 54400
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7772444444444445e-05,
+      "loss": 1.4833,
+      "step": 54405
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.776948148148148e-05,
+      "loss": 1.6388,
+      "step": 54410
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.776651851851852e-05,
+      "loss": 1.5379,
+      "step": 54415
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.7763555555555555e-05,
+      "loss": 1.4686,
+      "step": 54420
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7760592592592594e-05,
+      "loss": 1.5835,
+      "step": 54425
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.775762962962963e-05,
+      "loss": 1.5437,
+      "step": 54430
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7754666666666668e-05,
+      "loss": 1.4519,
+      "step": 54435
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7751703703703703e-05,
+      "loss": 1.4027,
+      "step": 54440
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7748740740740742e-05,
+      "loss": 1.4697,
+      "step": 54445
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7745777777777778e-05,
+      "loss": 1.5065,
+      "step": 54450
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7742814814814816e-05,
+      "loss": 1.5559,
+      "step": 54455
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7739851851851852e-05,
+      "loss": 1.6508,
+      "step": 54460
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.773688888888889e-05,
+      "loss": 1.5425,
+      "step": 54465
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.7733925925925926e-05,
+      "loss": 1.3459,
+      "step": 54470
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.773096296296296e-05,
+      "loss": 1.5579,
+      "step": 54475
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7728e-05,
+      "loss": 1.5081,
+      "step": 54480
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7725037037037036e-05,
+      "loss": 1.5215,
+      "step": 54485
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7722074074074075e-05,
+      "loss": 1.5532,
+      "step": 54490
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.771911111111111e-05,
+      "loss": 1.5331,
+      "step": 54495
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.771614814814815e-05,
+      "loss": 1.4741,
+      "step": 54500
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7713185185185184e-05,
+      "loss": 1.5276,
+      "step": 54505
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7710222222222223e-05,
+      "loss": 1.6184,
+      "step": 54510
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.770725925925926e-05,
+      "loss": 1.5541,
+      "step": 54515
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7704296296296297e-05,
+      "loss": 1.5962,
+      "step": 54520
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.7701333333333333e-05,
+      "loss": 1.5272,
+      "step": 54525
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.769837037037037e-05,
+      "loss": 1.5855,
+      "step": 54530
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7695407407407407e-05,
+      "loss": 1.4996,
+      "step": 54535
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7692444444444446e-05,
+      "loss": 1.4205,
+      "step": 54540
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.768948148148148e-05,
+      "loss": 1.4455,
+      "step": 54545
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.768651851851852e-05,
+      "loss": 1.5921,
+      "step": 54550
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7683555555555556e-05,
+      "loss": 1.4926,
+      "step": 54555
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7680592592592594e-05,
+      "loss": 1.5612,
+      "step": 54560
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.767762962962963e-05,
+      "loss": 1.5093,
+      "step": 54565
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.767466666666667e-05,
+      "loss": 1.5879,
+      "step": 54570
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7671703703703704e-05,
+      "loss": 1.4768,
+      "step": 54575
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7668740740740743e-05,
+      "loss": 1.5834,
+      "step": 54580
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.766577777777778e-05,
+      "loss": 1.5731,
+      "step": 54585
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.7662814814814814e-05,
+      "loss": 1.4477,
+      "step": 54590
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7659851851851853e-05,
+      "loss": 1.5977,
+      "step": 54595
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7656888888888888e-05,
+      "loss": 1.4965,
+      "step": 54600
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7653925925925927e-05,
+      "loss": 1.5664,
+      "step": 54605
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7650962962962962e-05,
+      "loss": 1.5256,
+      "step": 54610
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7648e-05,
+      "loss": 1.546,
+      "step": 54615
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7645037037037037e-05,
+      "loss": 1.525,
+      "step": 54620
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7642074074074075e-05,
+      "loss": 1.6668,
+      "step": 54625
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.763911111111111e-05,
+      "loss": 1.5191,
+      "step": 54630
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.763614814814815e-05,
+      "loss": 1.4967,
+      "step": 54635
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7633185185185185e-05,
+      "loss": 1.5297,
+      "step": 54640
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.7630222222222224e-05,
+      "loss": 1.4105,
+      "step": 54645
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.762725925925926e-05,
+      "loss": 1.4492,
+      "step": 54650
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7624296296296298e-05,
+      "loss": 1.5065,
+      "step": 54655
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7621333333333334e-05,
+      "loss": 1.6214,
+      "step": 54660
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7618370370370372e-05,
+      "loss": 1.6602,
+      "step": 54665
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7615407407407408e-05,
+      "loss": 1.47,
+      "step": 54670
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7612444444444447e-05,
+      "loss": 1.5095,
+      "step": 54675
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7609481481481482e-05,
+      "loss": 1.5392,
+      "step": 54680
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.760651851851852e-05,
+      "loss": 1.4964,
+      "step": 54685
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7603555555555556e-05,
+      "loss": 1.5416,
+      "step": 54690
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.7600592592592595e-05,
+      "loss": 1.5021,
+      "step": 54695
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.759762962962963e-05,
+      "loss": 1.4246,
+      "step": 54700
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7594666666666666e-05,
+      "loss": 1.4804,
+      "step": 54705
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7591703703703705e-05,
+      "loss": 1.6877,
+      "step": 54710
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.758874074074074e-05,
+      "loss": 1.6503,
+      "step": 54715
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.758577777777778e-05,
+      "loss": 1.5562,
+      "step": 54720
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7582814814814815e-05,
+      "loss": 1.5744,
+      "step": 54725
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7579851851851853e-05,
+      "loss": 1.495,
+      "step": 54730
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.757688888888889e-05,
+      "loss": 1.5209,
+      "step": 54735
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7573925925925928e-05,
+      "loss": 1.5309,
+      "step": 54740
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7570962962962963e-05,
+      "loss": 1.3946,
+      "step": 54745
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7568000000000002e-05,
+      "loss": 1.5188,
+      "step": 54750
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.7565037037037037e-05,
+      "loss": 1.4692,
+      "step": 54755
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7562074074074076e-05,
+      "loss": 1.5176,
+      "step": 54760
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.755911111111111e-05,
+      "loss": 1.4949,
+      "step": 54765
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.755614814814815e-05,
+      "loss": 1.4614,
+      "step": 54770
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7553185185185186e-05,
+      "loss": 1.6215,
+      "step": 54775
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7550222222222225e-05,
+      "loss": 1.6982,
+      "step": 54780
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.754725925925926e-05,
+      "loss": 1.5741,
+      "step": 54785
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.75442962962963e-05,
+      "loss": 1.7009,
+      "step": 54790
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7541333333333334e-05,
+      "loss": 1.5616,
+      "step": 54795
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7538370370370373e-05,
+      "loss": 1.6574,
+      "step": 54800
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.753540740740741e-05,
+      "loss": 1.5594,
+      "step": 54805
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7532444444444447e-05,
+      "loss": 1.6708,
+      "step": 54810
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.7529481481481483e-05,
+      "loss": 1.5441,
+      "step": 54815
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7526518518518518e-05,
+      "loss": 1.4444,
+      "step": 54820
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7523555555555557e-05,
+      "loss": 1.5511,
+      "step": 54825
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7520592592592593e-05,
+      "loss": 1.497,
+      "step": 54830
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.751762962962963e-05,
+      "loss": 1.5138,
+      "step": 54835
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7514666666666667e-05,
+      "loss": 1.5818,
+      "step": 54840
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7511703703703706e-05,
+      "loss": 1.4784,
+      "step": 54845
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.750874074074074e-05,
+      "loss": 1.6055,
+      "step": 54850
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.750577777777778e-05,
+      "loss": 1.574,
+      "step": 54855
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7502814814814815e-05,
+      "loss": 1.4169,
+      "step": 54860
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.7499851851851854e-05,
+      "loss": 1.5426,
+      "step": 54865
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.749688888888889e-05,
+      "loss": 1.5598,
+      "step": 54870
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.749392592592593e-05,
+      "loss": 1.5255,
+      "step": 54875
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7490962962962964e-05,
+      "loss": 1.4475,
+      "step": 54880
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7488000000000003e-05,
+      "loss": 1.6035,
+      "step": 54885
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7485037037037038e-05,
+      "loss": 1.4921,
+      "step": 54890
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7482074074074077e-05,
+      "loss": 1.6154,
+      "step": 54895
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7479111111111112e-05,
+      "loss": 1.42,
+      "step": 54900
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.747614814814815e-05,
+      "loss": 1.4745,
+      "step": 54905
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7473185185185187e-05,
+      "loss": 1.5056,
+      "step": 54910
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.7470222222222225e-05,
+      "loss": 1.6073,
+      "step": 54915
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.746725925925926e-05,
+      "loss": 1.5133,
+      "step": 54920
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.74642962962963e-05,
+      "loss": 1.5787,
+      "step": 54925
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7461333333333335e-05,
+      "loss": 1.4685,
+      "step": 54930
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.745837037037037e-05,
+      "loss": 1.4727,
+      "step": 54935
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.745540740740741e-05,
+      "loss": 1.6172,
+      "step": 54940
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7452444444444445e-05,
+      "loss": 1.4875,
+      "step": 54945
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7449481481481484e-05,
+      "loss": 1.5996,
+      "step": 54950
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.744651851851852e-05,
+      "loss": 1.5735,
+      "step": 54955
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7443555555555558e-05,
+      "loss": 1.5212,
+      "step": 54960
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7440592592592593e-05,
+      "loss": 1.4293,
+      "step": 54965
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7437629629629632e-05,
+      "loss": 1.5809,
+      "step": 54970
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7434666666666668e-05,
+      "loss": 1.4453,
+      "step": 54975
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.7431703703703706e-05,
+      "loss": 1.4591,
+      "step": 54980
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7428740740740742e-05,
+      "loss": 1.4652,
+      "step": 54985
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.742577777777778e-05,
+      "loss": 1.5661,
+      "step": 54990
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7422814814814816e-05,
+      "loss": 1.4299,
+      "step": 54995
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7419851851851855e-05,
+      "loss": 1.6477,
+      "step": 55000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.741688888888889e-05,
+      "loss": 1.4675,
+      "step": 55005
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.741392592592593e-05,
+      "loss": 1.5621,
+      "step": 55010
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7410962962962965e-05,
+      "loss": 1.5241,
+      "step": 55015
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7408000000000003e-05,
+      "loss": 1.636,
+      "step": 55020
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.740503703703704e-05,
+      "loss": 1.5965,
+      "step": 55025
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7402074074074078e-05,
+      "loss": 1.6819,
+      "step": 55030
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7399111111111113e-05,
+      "loss": 1.5361,
+      "step": 55035
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.7396148148148152e-05,
+      "loss": 1.564,
+      "step": 55040
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7393185185185187e-05,
+      "loss": 1.4741,
+      "step": 55045
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7390222222222223e-05,
+      "loss": 1.5171,
+      "step": 55050
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.738725925925926e-05,
+      "loss": 1.5494,
+      "step": 55055
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7384296296296297e-05,
+      "loss": 1.5259,
+      "step": 55060
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7381333333333332e-05,
+      "loss": 1.5216,
+      "step": 55065
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.737837037037037e-05,
+      "loss": 1.5086,
+      "step": 55070
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7375407407407407e-05,
+      "loss": 1.3518,
+      "step": 55075
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7372444444444446e-05,
+      "loss": 1.4738,
+      "step": 55080
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.736948148148148e-05,
+      "loss": 1.5763,
+      "step": 55085
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7366518518518516e-05,
+      "loss": 1.606,
+      "step": 55090
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.7363555555555555e-05,
+      "loss": 1.4934,
+      "step": 55095
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.736059259259259e-05,
+      "loss": 1.5381,
+      "step": 55100
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.735762962962963e-05,
+      "loss": 1.4844,
+      "step": 55105
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7354666666666665e-05,
+      "loss": 1.5998,
+      "step": 55110
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7351703703703704e-05,
+      "loss": 1.4209,
+      "step": 55115
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.734874074074074e-05,
+      "loss": 1.4938,
+      "step": 55120
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7345777777777778e-05,
+      "loss": 1.5149,
+      "step": 55125
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7342814814814813e-05,
+      "loss": 1.5591,
+      "step": 55130
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7339851851851852e-05,
+      "loss": 1.4319,
+      "step": 55135
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7336888888888888e-05,
+      "loss": 1.5793,
+      "step": 55140
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7333925925925927e-05,
+      "loss": 1.4591,
+      "step": 55145
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.7330962962962962e-05,
+      "loss": 1.6139,
+      "step": 55150
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7328e-05,
+      "loss": 1.5746,
+      "step": 55155
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7325037037037036e-05,
+      "loss": 1.3921,
+      "step": 55160
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7322074074074075e-05,
+      "loss": 1.5366,
+      "step": 55165
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.731911111111111e-05,
+      "loss": 1.5098,
+      "step": 55170
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.731614814814815e-05,
+      "loss": 1.5341,
+      "step": 55175
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7313185185185185e-05,
+      "loss": 1.5865,
+      "step": 55180
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7310222222222224e-05,
+      "loss": 1.5945,
+      "step": 55185
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.730725925925926e-05,
+      "loss": 1.4708,
+      "step": 55190
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7304296296296298e-05,
+      "loss": 1.6234,
+      "step": 55195
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.7301333333333333e-05,
+      "loss": 1.5696,
+      "step": 55200
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.729837037037037e-05,
+      "loss": 1.4576,
+      "step": 55205
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7295407407407408e-05,
+      "loss": 1.4692,
+      "step": 55210
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7292444444444443e-05,
+      "loss": 1.3785,
+      "step": 55215
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7289481481481482e-05,
+      "loss": 1.3889,
+      "step": 55220
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7286518518518517e-05,
+      "loss": 1.529,
+      "step": 55225
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7283555555555556e-05,
+      "loss": 1.6058,
+      "step": 55230
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.728059259259259e-05,
+      "loss": 1.54,
+      "step": 55235
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.727762962962963e-05,
+      "loss": 1.5099,
+      "step": 55240
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7274666666666666e-05,
+      "loss": 1.522,
+      "step": 55245
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7271703703703705e-05,
+      "loss": 1.5237,
+      "step": 55250
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.726874074074074e-05,
+      "loss": 1.5013,
+      "step": 55255
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.726577777777778e-05,
+      "loss": 1.5111,
+      "step": 55260
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7262814814814814e-05,
+      "loss": 1.5267,
+      "step": 55265
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7259851851851853e-05,
+      "loss": 1.4933,
+      "step": 55270
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.725688888888889e-05,
+      "loss": 1.5465,
+      "step": 55275
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7253925925925927e-05,
+      "loss": 1.595,
+      "step": 55280
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7250962962962963e-05,
+      "loss": 1.4557,
+      "step": 55285
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7248e-05,
+      "loss": 1.5768,
+      "step": 55290
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7245037037037037e-05,
+      "loss": 1.6103,
+      "step": 55295
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7242074074074076e-05,
+      "loss": 1.5136,
+      "step": 55300
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.723911111111111e-05,
+      "loss": 1.5994,
+      "step": 55305
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.723614814814815e-05,
+      "loss": 1.4701,
+      "step": 55310
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.7233185185185186e-05,
+      "loss": 1.5971,
+      "step": 55315
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.723022222222222e-05,
+      "loss": 1.6812,
+      "step": 55320
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.722725925925926e-05,
+      "loss": 1.5619,
+      "step": 55325
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7224296296296295e-05,
+      "loss": 1.5565,
+      "step": 55330
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7221333333333334e-05,
+      "loss": 1.5861,
+      "step": 55335
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.721837037037037e-05,
+      "loss": 1.4722,
+      "step": 55340
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7215407407407408e-05,
+      "loss": 1.5776,
+      "step": 55345
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7212444444444444e-05,
+      "loss": 1.5992,
+      "step": 55350
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7209481481481483e-05,
+      "loss": 1.4286,
+      "step": 55355
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7206518518518518e-05,
+      "loss": 1.5741,
+      "step": 55360
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7203555555555557e-05,
+      "loss": 1.6209,
+      "step": 55365
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.7200592592592592e-05,
+      "loss": 1.4816,
+      "step": 55370
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.719762962962963e-05,
+      "loss": 1.4475,
+      "step": 55375
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7194666666666666e-05,
+      "loss": 1.4962,
+      "step": 55380
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7191703703703705e-05,
+      "loss": 1.3798,
+      "step": 55385
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.718874074074074e-05,
+      "loss": 1.7109,
+      "step": 55390
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.718577777777778e-05,
+      "loss": 1.3526,
+      "step": 55395
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7182814814814815e-05,
+      "loss": 1.5241,
+      "step": 55400
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7179851851851854e-05,
+      "loss": 1.4618,
+      "step": 55405
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.717688888888889e-05,
+      "loss": 1.5312,
+      "step": 55410
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7173925925925928e-05,
+      "loss": 1.5948,
+      "step": 55415
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7170962962962964e-05,
+      "loss": 1.4865,
+      "step": 55420
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7168000000000002e-05,
+      "loss": 1.4777,
+      "step": 55425
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.7165037037037038e-05,
+      "loss": 1.4915,
+      "step": 55430
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7162074074074073e-05,
+      "loss": 1.5157,
+      "step": 55435
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7159111111111112e-05,
+      "loss": 1.5839,
+      "step": 55440
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7156148148148147e-05,
+      "loss": 1.5197,
+      "step": 55445
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7153185185185186e-05,
+      "loss": 1.459,
+      "step": 55450
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7150222222222222e-05,
+      "loss": 1.5412,
+      "step": 55455
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.714725925925926e-05,
+      "loss": 1.5439,
+      "step": 55460
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7144296296296296e-05,
+      "loss": 1.4996,
+      "step": 55465
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7141333333333335e-05,
+      "loss": 1.556,
+      "step": 55470
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.713837037037037e-05,
+      "loss": 1.4095,
+      "step": 55475
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.713540740740741e-05,
+      "loss": 1.608,
+      "step": 55480
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7132444444444444e-05,
+      "loss": 1.6704,
+      "step": 55485
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.7129481481481483e-05,
+      "loss": 1.5824,
+      "step": 55490
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.712651851851852e-05,
+      "loss": 1.502,
+      "step": 55495
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7123555555555558e-05,
+      "loss": 1.5069,
+      "step": 55500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7120592592592593e-05,
+      "loss": 1.5466,
+      "step": 55505
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7117629629629632e-05,
+      "loss": 1.4501,
+      "step": 55510
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7114666666666667e-05,
+      "loss": 1.5647,
+      "step": 55515
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7111703703703706e-05,
+      "loss": 1.5654,
+      "step": 55520
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.710874074074074e-05,
+      "loss": 1.6235,
+      "step": 55525
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.710577777777778e-05,
+      "loss": 1.4631,
+      "step": 55530
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7102814814814816e-05,
+      "loss": 1.6643,
+      "step": 55535
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.7099851851851855e-05,
+      "loss": 1.5965,
+      "step": 55540
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.709688888888889e-05,
+      "loss": 1.597,
+      "step": 55545
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7093925925925925e-05,
+      "loss": 1.5985,
+      "step": 55550
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7090962962962964e-05,
+      "loss": 1.6827,
+      "step": 55555
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7088e-05,
+      "loss": 1.5294,
+      "step": 55560
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.708503703703704e-05,
+      "loss": 1.6238,
+      "step": 55565
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7082074074074074e-05,
+      "loss": 1.4813,
+      "step": 55570
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7079111111111113e-05,
+      "loss": 1.4501,
+      "step": 55575
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7076148148148148e-05,
+      "loss": 1.5369,
+      "step": 55580
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7073185185185187e-05,
+      "loss": 1.5457,
+      "step": 55585
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7070222222222222e-05,
+      "loss": 1.5048,
+      "step": 55590
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.706725925925926e-05,
+      "loss": 1.4843,
+      "step": 55595
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.7064296296296297e-05,
+      "loss": 1.5527,
+      "step": 55600
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7061333333333336e-05,
+      "loss": 1.4307,
+      "step": 55605
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.705837037037037e-05,
+      "loss": 1.6305,
+      "step": 55610
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.705540740740741e-05,
+      "loss": 1.4328,
+      "step": 55615
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7052444444444445e-05,
+      "loss": 1.4845,
+      "step": 55620
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7049481481481484e-05,
+      "loss": 1.4451,
+      "step": 55625
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.704651851851852e-05,
+      "loss": 1.5021,
+      "step": 55630
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.704355555555556e-05,
+      "loss": 1.592,
+      "step": 55635
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7040592592592594e-05,
+      "loss": 1.4752,
+      "step": 55640
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7037629629629633e-05,
+      "loss": 1.4744,
+      "step": 55645
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7034666666666668e-05,
+      "loss": 1.6049,
+      "step": 55650
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.7031703703703707e-05,
+      "loss": 1.5754,
+      "step": 55655
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7028740740740742e-05,
+      "loss": 1.5678,
+      "step": 55660
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7025777777777778e-05,
+      "loss": 1.5917,
+      "step": 55665
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7022814814814817e-05,
+      "loss": 1.569,
+      "step": 55670
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7019851851851852e-05,
+      "loss": 1.5961,
+      "step": 55675
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.701688888888889e-05,
+      "loss": 1.4696,
+      "step": 55680
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7013925925925926e-05,
+      "loss": 1.5225,
+      "step": 55685
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7010962962962965e-05,
+      "loss": 1.4792,
+      "step": 55690
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7008e-05,
+      "loss": 1.5202,
+      "step": 55695
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.700503703703704e-05,
+      "loss": 1.5892,
+      "step": 55700
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.7002074074074075e-05,
+      "loss": 1.5454,
+      "step": 55705
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.6999111111111114e-05,
+      "loss": 1.5368,
+      "step": 55710
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.699614814814815e-05,
+      "loss": 1.543,
+      "step": 55715
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6993185185185188e-05,
+      "loss": 1.4906,
+      "step": 55720
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6990222222222223e-05,
+      "loss": 1.5221,
+      "step": 55725
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6987259259259262e-05,
+      "loss": 1.5514,
+      "step": 55730
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6984296296296297e-05,
+      "loss": 1.5906,
+      "step": 55735
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6981333333333336e-05,
+      "loss": 1.5805,
+      "step": 55740
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6978370370370372e-05,
+      "loss": 1.462,
+      "step": 55745
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.697540740740741e-05,
+      "loss": 1.5429,
+      "step": 55750
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6972444444444446e-05,
+      "loss": 1.5489,
+      "step": 55755
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.6969481481481485e-05,
+      "loss": 1.6264,
+      "step": 55760
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.696651851851852e-05,
+      "loss": 1.6514,
+      "step": 55765
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.696355555555556e-05,
+      "loss": 1.6208,
+      "step": 55770
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6960592592592595e-05,
+      "loss": 1.7229,
+      "step": 55775
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.695762962962963e-05,
+      "loss": 1.4408,
+      "step": 55780
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.695466666666667e-05,
+      "loss": 1.6531,
+      "step": 55785
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6951703703703704e-05,
+      "loss": 1.6879,
+      "step": 55790
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6948740740740743e-05,
+      "loss": 1.5688,
+      "step": 55795
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.694577777777778e-05,
+      "loss": 1.5016,
+      "step": 55800
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6942814814814817e-05,
+      "loss": 1.445,
+      "step": 55805
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6939851851851853e-05,
+      "loss": 1.5127,
+      "step": 55810
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.693688888888889e-05,
+      "loss": 1.4937,
+      "step": 55815
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6933925925925927e-05,
+      "loss": 1.6443,
+      "step": 55820
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.6930962962962966e-05,
+      "loss": 1.5872,
+      "step": 55825
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6928e-05,
+      "loss": 1.4814,
+      "step": 55830
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6925037037037037e-05,
+      "loss": 1.5755,
+      "step": 55835
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6922074074074072e-05,
+      "loss": 1.5642,
+      "step": 55840
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.691911111111111e-05,
+      "loss": 1.4533,
+      "step": 55845
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6916148148148146e-05,
+      "loss": 1.5246,
+      "step": 55850
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6913185185185185e-05,
+      "loss": 1.511,
+      "step": 55855
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.691022222222222e-05,
+      "loss": 1.4451,
+      "step": 55860
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.690725925925926e-05,
+      "loss": 1.4933,
+      "step": 55865
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6904296296296295e-05,
+      "loss": 1.5376,
+      "step": 55870
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.6901333333333334e-05,
+      "loss": 1.5438,
+      "step": 55875
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.689837037037037e-05,
+      "loss": 1.5084,
+      "step": 55880
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6895407407407408e-05,
+      "loss": 1.6436,
+      "step": 55885
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6892444444444443e-05,
+      "loss": 1.5424,
+      "step": 55890
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6889481481481482e-05,
+      "loss": 1.5215,
+      "step": 55895
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6886518518518518e-05,
+      "loss": 1.5333,
+      "step": 55900
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6883555555555556e-05,
+      "loss": 1.5025,
+      "step": 55905
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6880592592592592e-05,
+      "loss": 1.6064,
+      "step": 55910
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.687762962962963e-05,
+      "loss": 1.456,
+      "step": 55915
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6874666666666666e-05,
+      "loss": 1.5671,
+      "step": 55920
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6871703703703705e-05,
+      "loss": 1.4753,
+      "step": 55925
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.686874074074074e-05,
+      "loss": 1.5083,
+      "step": 55930
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.686577777777778e-05,
+      "loss": 1.5953,
+      "step": 55935
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.6862814814814815e-05,
+      "loss": 1.5671,
+      "step": 55940
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.685985185185185e-05,
+      "loss": 1.537,
+      "step": 55945
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.685688888888889e-05,
+      "loss": 1.5699,
+      "step": 55950
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6853925925925924e-05,
+      "loss": 1.5137,
+      "step": 55955
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6850962962962963e-05,
+      "loss": 1.5493,
+      "step": 55960
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6848e-05,
+      "loss": 1.5812,
+      "step": 55965
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6845037037037037e-05,
+      "loss": 1.5359,
+      "step": 55970
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6842074074074073e-05,
+      "loss": 1.5635,
+      "step": 55975
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.683911111111111e-05,
+      "loss": 1.5251,
+      "step": 55980
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6836148148148147e-05,
+      "loss": 1.5709,
+      "step": 55985
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.6833185185185186e-05,
+      "loss": 1.553,
+      "step": 55990
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.683022222222222e-05,
+      "loss": 1.5106,
+      "step": 55995
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.682725925925926e-05,
+      "loss": 1.5495,
+      "step": 56000
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6824296296296296e-05,
+      "loss": 1.5,
+      "step": 56005
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6821333333333334e-05,
+      "loss": 1.4419,
+      "step": 56010
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.681837037037037e-05,
+      "loss": 1.711,
+      "step": 56015
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.681540740740741e-05,
+      "loss": 1.5766,
+      "step": 56020
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6812444444444444e-05,
+      "loss": 1.4804,
+      "step": 56025
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6809481481481483e-05,
+      "loss": 1.5387,
+      "step": 56030
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.680651851851852e-05,
+      "loss": 1.5965,
+      "step": 56035
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6803555555555557e-05,
+      "loss": 1.5914,
+      "step": 56040
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.6800592592592593e-05,
+      "loss": 1.5473,
+      "step": 56045
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.679762962962963e-05,
+      "loss": 1.5701,
+      "step": 56050
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6794666666666667e-05,
+      "loss": 1.5801,
+      "step": 56055
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6791703703703702e-05,
+      "loss": 1.575,
+      "step": 56060
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.678874074074074e-05,
+      "loss": 1.6211,
+      "step": 56065
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6785777777777777e-05,
+      "loss": 1.5513,
+      "step": 56070
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6782814814814815e-05,
+      "loss": 1.5964,
+      "step": 56075
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.677985185185185e-05,
+      "loss": 1.542,
+      "step": 56080
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.677688888888889e-05,
+      "loss": 1.5463,
+      "step": 56085
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6773925925925925e-05,
+      "loss": 1.4798,
+      "step": 56090
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6770962962962964e-05,
+      "loss": 1.5966,
+      "step": 56095
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6768e-05,
+      "loss": 1.6393,
+      "step": 56100
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.6765037037037038e-05,
+      "loss": 1.501,
+      "step": 56105
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6762074074074074e-05,
+      "loss": 1.5032,
+      "step": 56110
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6759111111111112e-05,
+      "loss": 1.5574,
+      "step": 56115
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6756148148148148e-05,
+      "loss": 1.49,
+      "step": 56120
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6753185185185187e-05,
+      "loss": 1.554,
+      "step": 56125
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6750222222222222e-05,
+      "loss": 1.5002,
+      "step": 56130
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.674725925925926e-05,
+      "loss": 1.5316,
+      "step": 56135
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6744296296296296e-05,
+      "loss": 1.4794,
+      "step": 56140
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6741333333333335e-05,
+      "loss": 1.5091,
+      "step": 56145
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.673837037037037e-05,
+      "loss": 1.4438,
+      "step": 56150
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.673540740740741e-05,
+      "loss": 1.5243,
+      "step": 56155
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6732444444444445e-05,
+      "loss": 1.5675,
+      "step": 56160
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.6729481481481484e-05,
+      "loss": 1.4815,
+      "step": 56165
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.672651851851852e-05,
+      "loss": 1.616,
+      "step": 56170
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6723555555555555e-05,
+      "loss": 1.6924,
+      "step": 56175
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6720592592592593e-05,
+      "loss": 1.5124,
+      "step": 56180
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.671762962962963e-05,
+      "loss": 1.5835,
+      "step": 56185
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6714666666666668e-05,
+      "loss": 1.6238,
+      "step": 56190
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6711703703703703e-05,
+      "loss": 1.5555,
+      "step": 56195
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6708740740740742e-05,
+      "loss": 1.5174,
+      "step": 56200
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6705777777777777e-05,
+      "loss": 1.5372,
+      "step": 56205
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.6702814814814816e-05,
+      "loss": 1.6737,
+      "step": 56210
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.669985185185185e-05,
+      "loss": 1.5666,
+      "step": 56215
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.669688888888889e-05,
+      "loss": 1.5074,
+      "step": 56220
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6693925925925926e-05,
+      "loss": 1.5528,
+      "step": 56225
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6690962962962965e-05,
+      "loss": 1.5752,
+      "step": 56230
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6688e-05,
+      "loss": 1.4894,
+      "step": 56235
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.668503703703704e-05,
+      "loss": 1.5343,
+      "step": 56240
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6682074074074074e-05,
+      "loss": 1.6339,
+      "step": 56245
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6679111111111113e-05,
+      "loss": 1.5617,
+      "step": 56250
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.667614814814815e-05,
+      "loss": 1.4568,
+      "step": 56255
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6673185185185187e-05,
+      "loss": 1.4813,
+      "step": 56260
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6670222222222223e-05,
+      "loss": 1.464,
+      "step": 56265
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6667259259259262e-05,
+      "loss": 1.444,
+      "step": 56270
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.6664296296296297e-05,
+      "loss": 1.5838,
+      "step": 56275
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6661333333333336e-05,
+      "loss": 1.3857,
+      "step": 56280
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.665837037037037e-05,
+      "loss": 1.4634,
+      "step": 56285
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6655407407407407e-05,
+      "loss": 1.4352,
+      "step": 56290
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6652444444444446e-05,
+      "loss": 1.3871,
+      "step": 56295
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.664948148148148e-05,
+      "loss": 1.3942,
+      "step": 56300
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.664651851851852e-05,
+      "loss": 1.4356,
+      "step": 56305
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6643555555555555e-05,
+      "loss": 1.4767,
+      "step": 56310
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6640592592592594e-05,
+      "loss": 1.4146,
+      "step": 56315
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.663762962962963e-05,
+      "loss": 1.4985,
+      "step": 56320
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.663466666666667e-05,
+      "loss": 1.4298,
+      "step": 56325
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.6631703703703704e-05,
+      "loss": 1.548,
+      "step": 56330
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6628740740740743e-05,
+      "loss": 1.354,
+      "step": 56335
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6625777777777778e-05,
+      "loss": 1.4091,
+      "step": 56340
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6622814814814817e-05,
+      "loss": 1.4469,
+      "step": 56345
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6619851851851852e-05,
+      "loss": 1.435,
+      "step": 56350
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.661688888888889e-05,
+      "loss": 1.4372,
+      "step": 56355
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6613925925925927e-05,
+      "loss": 1.3682,
+      "step": 56360
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6610962962962965e-05,
+      "loss": 1.6016,
+      "step": 56365
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6608e-05,
+      "loss": 1.3731,
+      "step": 56370
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.660503703703704e-05,
+      "loss": 1.4114,
+      "step": 56375
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6602074074074075e-05,
+      "loss": 1.5372,
+      "step": 56380
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.6599111111111114e-05,
+      "loss": 1.5105,
+      "step": 56385
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.659614814814815e-05,
+      "loss": 1.3513,
+      "step": 56390
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6593185185185188e-05,
+      "loss": 1.411,
+      "step": 56395
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6590222222222224e-05,
+      "loss": 1.374,
+      "step": 56400
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.658725925925926e-05,
+      "loss": 1.413,
+      "step": 56405
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6584296296296298e-05,
+      "loss": 1.4381,
+      "step": 56410
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6581333333333333e-05,
+      "loss": 1.4425,
+      "step": 56415
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6578370370370372e-05,
+      "loss": 1.3954,
+      "step": 56420
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6575407407407408e-05,
+      "loss": 1.4891,
+      "step": 56425
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6572444444444446e-05,
+      "loss": 1.5278,
+      "step": 56430
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6569481481481482e-05,
+      "loss": 1.515,
+      "step": 56435
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.656651851851852e-05,
+      "loss": 1.4845,
+      "step": 56440
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.6563555555555556e-05,
+      "loss": 1.497,
+      "step": 56445
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6560592592592595e-05,
+      "loss": 1.4527,
+      "step": 56450
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.655762962962963e-05,
+      "loss": 1.4164,
+      "step": 56455
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.655466666666667e-05,
+      "loss": 1.4828,
+      "step": 56460
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6551703703703705e-05,
+      "loss": 1.5218,
+      "step": 56465
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6548740740740743e-05,
+      "loss": 1.4109,
+      "step": 56470
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.654577777777778e-05,
+      "loss": 1.3946,
+      "step": 56475
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6542814814814818e-05,
+      "loss": 1.4448,
+      "step": 56480
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6539851851851853e-05,
+      "loss": 1.446,
+      "step": 56485
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6536888888888892e-05,
+      "loss": 1.5763,
+      "step": 56490
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6533925925925927e-05,
+      "loss": 1.4697,
+      "step": 56495
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.6530962962962966e-05,
+      "loss": 1.434,
+      "step": 56500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6528e-05,
+      "loss": 1.5345,
+      "step": 56505
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.652503703703704e-05,
+      "loss": 1.475,
+      "step": 56510
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6522074074074076e-05,
+      "loss": 1.5443,
+      "step": 56515
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.651911111111111e-05,
+      "loss": 1.4144,
+      "step": 56520
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.651614814814815e-05,
+      "loss": 1.3183,
+      "step": 56525
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6513185185185186e-05,
+      "loss": 1.5795,
+      "step": 56530
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6510222222222224e-05,
+      "loss": 1.4622,
+      "step": 56535
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.650725925925926e-05,
+      "loss": 1.4535,
+      "step": 56540
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.65042962962963e-05,
+      "loss": 1.3773,
+      "step": 56545
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6501333333333334e-05,
+      "loss": 1.3614,
+      "step": 56550
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.6498370370370373e-05,
+      "loss": 1.5734,
+      "step": 56555
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.649540740740741e-05,
+      "loss": 1.361,
+      "step": 56560
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6492444444444447e-05,
+      "loss": 1.4434,
+      "step": 56565
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6489481481481483e-05,
+      "loss": 1.3869,
+      "step": 56570
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.648651851851852e-05,
+      "loss": 1.4083,
+      "step": 56575
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6483555555555557e-05,
+      "loss": 1.5777,
+      "step": 56580
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6480592592592596e-05,
+      "loss": 1.4525,
+      "step": 56585
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.647762962962963e-05,
+      "loss": 1.5079,
+      "step": 56590
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.647466666666667e-05,
+      "loss": 1.4864,
+      "step": 56595
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6471703703703705e-05,
+      "loss": 1.4319,
+      "step": 56600
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6468740740740744e-05,
+      "loss": 1.3583,
+      "step": 56605
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6465777777777776e-05,
+      "loss": 1.549,
+      "step": 56610
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.6462814814814815e-05,
+      "loss": 1.376,
+      "step": 56615
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.645985185185185e-05,
+      "loss": 1.4434,
+      "step": 56620
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.645688888888889e-05,
+      "loss": 1.4769,
+      "step": 56625
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6453925925925925e-05,
+      "loss": 1.3302,
+      "step": 56630
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6450962962962964e-05,
+      "loss": 1.5486,
+      "step": 56635
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6448e-05,
+      "loss": 1.4111,
+      "step": 56640
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6445037037037038e-05,
+      "loss": 1.4691,
+      "step": 56645
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6442074074074073e-05,
+      "loss": 1.3895,
+      "step": 56650
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6439111111111112e-05,
+      "loss": 1.4762,
+      "step": 56655
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6436148148148148e-05,
+      "loss": 1.5681,
+      "step": 56660
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6433185185185186e-05,
+      "loss": 1.3444,
+      "step": 56665
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.6430222222222222e-05,
+      "loss": 1.5443,
+      "step": 56670
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6427259259259257e-05,
+      "loss": 1.4395,
+      "step": 56675
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6424296296296296e-05,
+      "loss": 1.3889,
+      "step": 56680
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.642133333333333e-05,
+      "loss": 1.3609,
+      "step": 56685
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.641837037037037e-05,
+      "loss": 1.4472,
+      "step": 56690
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6415407407407406e-05,
+      "loss": 1.3483,
+      "step": 56695
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6412444444444445e-05,
+      "loss": 1.458,
+      "step": 56700
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.640948148148148e-05,
+      "loss": 1.5122,
+      "step": 56705
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.640651851851852e-05,
+      "loss": 1.5652,
+      "step": 56710
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6403555555555554e-05,
+      "loss": 1.4863,
+      "step": 56715
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.6400592592592593e-05,
+      "loss": 1.422,
+      "step": 56720
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.639762962962963e-05,
+      "loss": 1.3409,
+      "step": 56725
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6394666666666667e-05,
+      "loss": 1.3657,
+      "step": 56730
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6391703703703703e-05,
+      "loss": 1.497,
+      "step": 56735
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.638874074074074e-05,
+      "loss": 1.4964,
+      "step": 56740
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6385777777777777e-05,
+      "loss": 1.4076,
+      "step": 56745
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6382814814814816e-05,
+      "loss": 1.4491,
+      "step": 56750
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.637985185185185e-05,
+      "loss": 1.4611,
+      "step": 56755
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.637688888888889e-05,
+      "loss": 1.4653,
+      "step": 56760
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6373925925925926e-05,
+      "loss": 1.5266,
+      "step": 56765
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6370962962962964e-05,
+      "loss": 1.4873,
+      "step": 56770
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.6368e-05,
+      "loss": 1.4298,
+      "step": 56775
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.636503703703704e-05,
+      "loss": 1.5629,
+      "step": 56780
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.636266666666667e-05,
+      "loss": 1.4198,
+      "step": 56785
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6359703703703704e-05,
+      "loss": 1.4296,
+      "step": 56790
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6356740740740743e-05,
+      "loss": 1.3766,
+      "step": 56795
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6353777777777778e-05,
+      "loss": 1.5183,
+      "step": 56800
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6350814814814817e-05,
+      "loss": 1.4686,
+      "step": 56805
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6347851851851852e-05,
+      "loss": 1.5341,
+      "step": 56810
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.634488888888889e-05,
+      "loss": 1.4001,
+      "step": 56815
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6341925925925927e-05,
+      "loss": 1.4328,
+      "step": 56820
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6338962962962965e-05,
+      "loss": 1.3983,
+      "step": 56825
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6336e-05,
+      "loss": 1.4608,
+      "step": 56830
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.633303703703704e-05,
+      "loss": 1.4269,
+      "step": 56835
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.6330074074074075e-05,
+      "loss": 1.523,
+      "step": 56840
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6327111111111114e-05,
+      "loss": 1.4866,
+      "step": 56845
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.632414814814815e-05,
+      "loss": 1.5254,
+      "step": 56850
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6321185185185188e-05,
+      "loss": 1.4886,
+      "step": 56855
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6318222222222224e-05,
+      "loss": 1.4051,
+      "step": 56860
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.631525925925926e-05,
+      "loss": 1.479,
+      "step": 56865
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6312296296296298e-05,
+      "loss": 1.4943,
+      "step": 56870
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6309333333333333e-05,
+      "loss": 1.3982,
+      "step": 56875
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6306370370370372e-05,
+      "loss": 1.5515,
+      "step": 56880
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6303407407407408e-05,
+      "loss": 1.3565,
+      "step": 56885
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6300444444444446e-05,
+      "loss": 1.4249,
+      "step": 56890
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.6297481481481482e-05,
+      "loss": 1.4261,
+      "step": 56895
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.629451851851852e-05,
+      "loss": 1.3856,
+      "step": 56900
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6291555555555556e-05,
+      "loss": 1.5375,
+      "step": 56905
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6288592592592595e-05,
+      "loss": 1.5177,
+      "step": 56910
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.628562962962963e-05,
+      "loss": 1.5217,
+      "step": 56915
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.628266666666667e-05,
+      "loss": 1.522,
+      "step": 56920
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6279703703703705e-05,
+      "loss": 1.3629,
+      "step": 56925
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6276740740740743e-05,
+      "loss": 1.5598,
+      "step": 56930
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.627377777777778e-05,
+      "loss": 1.502,
+      "step": 56935
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6270814814814818e-05,
+      "loss": 1.4898,
+      "step": 56940
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6267851851851853e-05,
+      "loss": 1.4654,
+      "step": 56945
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.6264888888888892e-05,
+      "loss": 1.3799,
+      "step": 56950
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6261925925925927e-05,
+      "loss": 1.3897,
+      "step": 56955
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6258962962962966e-05,
+      "loss": 1.4804,
+      "step": 56960
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6256e-05,
+      "loss": 1.441,
+      "step": 56965
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.625303703703704e-05,
+      "loss": 1.6212,
+      "step": 56970
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6250074074074076e-05,
+      "loss": 1.3991,
+      "step": 56975
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.624711111111111e-05,
+      "loss": 1.4992,
+      "step": 56980
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.624414814814815e-05,
+      "loss": 1.4879,
+      "step": 56985
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6241185185185186e-05,
+      "loss": 1.5002,
+      "step": 56990
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.6238222222222224e-05,
+      "loss": 1.4677,
+      "step": 56995
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.623525925925926e-05,
+      "loss": 1.4447,
+      "step": 57000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.62322962962963e-05,
+      "loss": 1.4389,
+      "step": 57005
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6229333333333334e-05,
+      "loss": 1.5065,
+      "step": 57010
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6226370370370373e-05,
+      "loss": 1.4818,
+      "step": 57015
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6223407407407408e-05,
+      "loss": 1.4784,
+      "step": 57020
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6220444444444447e-05,
+      "loss": 1.4065,
+      "step": 57025
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6217481481481483e-05,
+      "loss": 1.4242,
+      "step": 57030
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.621451851851852e-05,
+      "loss": 1.4451,
+      "step": 57035
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6211555555555557e-05,
+      "loss": 1.6388,
+      "step": 57040
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6208592592592596e-05,
+      "loss": 1.4881,
+      "step": 57045
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.620562962962963e-05,
+      "loss": 1.386,
+      "step": 57050
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.620266666666667e-05,
+      "loss": 1.4108,
+      "step": 57055
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6199703703703705e-05,
+      "loss": 1.3407,
+      "step": 57060
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.6196740740740744e-05,
+      "loss": 1.5643,
+      "step": 57065
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.619377777777778e-05,
+      "loss": 1.4984,
+      "step": 57070
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6190814814814815e-05,
+      "loss": 1.5101,
+      "step": 57075
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.618785185185185e-05,
+      "loss": 1.5068,
+      "step": 57080
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.618488888888889e-05,
+      "loss": 1.2716,
+      "step": 57085
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6181925925925925e-05,
+      "loss": 1.3864,
+      "step": 57090
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6178962962962964e-05,
+      "loss": 1.5048,
+      "step": 57095
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6176e-05,
+      "loss": 1.3704,
+      "step": 57100
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6173037037037038e-05,
+      "loss": 1.5049,
+      "step": 57105
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6170074074074073e-05,
+      "loss": 1.4995,
+      "step": 57110
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6167111111111112e-05,
+      "loss": 1.4458,
+      "step": 57115
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.6164148148148147e-05,
+      "loss": 1.4959,
+      "step": 57120
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6161185185185186e-05,
+      "loss": 1.5256,
+      "step": 57125
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6158222222222222e-05,
+      "loss": 1.4021,
+      "step": 57130
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.615525925925926e-05,
+      "loss": 1.5498,
+      "step": 57135
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6152296296296296e-05,
+      "loss": 1.4001,
+      "step": 57140
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.614933333333333e-05,
+      "loss": 1.3844,
+      "step": 57145
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.614637037037037e-05,
+      "loss": 1.4142,
+      "step": 57150
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6143407407407406e-05,
+      "loss": 1.5698,
+      "step": 57155
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6140444444444444e-05,
+      "loss": 1.5131,
+      "step": 57160
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.613748148148148e-05,
+      "loss": 1.4638,
+      "step": 57165
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.613451851851852e-05,
+      "loss": 1.47,
+      "step": 57170
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.6131555555555554e-05,
+      "loss": 1.4337,
+      "step": 57175
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6128592592592593e-05,
+      "loss": 1.4232,
+      "step": 57180
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.612562962962963e-05,
+      "loss": 1.4953,
+      "step": 57185
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6122666666666667e-05,
+      "loss": 1.4315,
+      "step": 57190
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6119703703703703e-05,
+      "loss": 1.5302,
+      "step": 57195
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.611674074074074e-05,
+      "loss": 1.516,
+      "step": 57200
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6113777777777777e-05,
+      "loss": 1.485,
+      "step": 57205
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6110814814814816e-05,
+      "loss": 1.4658,
+      "step": 57210
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.610785185185185e-05,
+      "loss": 1.4462,
+      "step": 57215
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.610488888888889e-05,
+      "loss": 1.4209,
+      "step": 57220
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6101925925925925e-05,
+      "loss": 1.4613,
+      "step": 57225
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1.6098962962962964e-05,
+      "loss": 1.4842,
+      "step": 57230
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6096e-05,
+      "loss": 1.4258,
+      "step": 57235
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.609303703703704e-05,
+      "loss": 1.4375,
+      "step": 57240
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6090074074074074e-05,
+      "loss": 1.5497,
+      "step": 57245
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6087111111111113e-05,
+      "loss": 1.4033,
+      "step": 57250
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6084148148148148e-05,
+      "loss": 1.3823,
+      "step": 57255
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6081185185185184e-05,
+      "loss": 1.4156,
+      "step": 57260
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6078222222222222e-05,
+      "loss": 1.5244,
+      "step": 57265
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6075259259259258e-05,
+      "loss": 1.4995,
+      "step": 57270
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6072296296296297e-05,
+      "loss": 1.55,
+      "step": 57275
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6069333333333332e-05,
+      "loss": 1.5527,
+      "step": 57280
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.606637037037037e-05,
+      "loss": 1.5155,
+      "step": 57285
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.6063407407407406e-05,
+      "loss": 1.5184,
+      "step": 57290
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6060444444444445e-05,
+      "loss": 1.3142,
+      "step": 57295
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.605748148148148e-05,
+      "loss": 1.5419,
+      "step": 57300
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.605451851851852e-05,
+      "loss": 1.4331,
+      "step": 57305
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6051555555555555e-05,
+      "loss": 1.4807,
+      "step": 57310
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6048592592592594e-05,
+      "loss": 1.5338,
+      "step": 57315
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.604562962962963e-05,
+      "loss": 1.624,
+      "step": 57320
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6042666666666668e-05,
+      "loss": 1.4465,
+      "step": 57325
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6039703703703703e-05,
+      "loss": 1.5088,
+      "step": 57330
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6036740740740742e-05,
+      "loss": 1.5624,
+      "step": 57335
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6033777777777778e-05,
+      "loss": 1.4521,
+      "step": 57340
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 1.6030814814814817e-05,
+      "loss": 1.4668,
+      "step": 57345
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6027851851851852e-05,
+      "loss": 1.4755,
+      "step": 57350
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.602488888888889e-05,
+      "loss": 1.4893,
+      "step": 57355
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6021925925925926e-05,
+      "loss": 1.4304,
+      "step": 57360
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6018962962962965e-05,
+      "loss": 1.3858,
+      "step": 57365
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6016e-05,
+      "loss": 1.4237,
+      "step": 57370
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6013037037037036e-05,
+      "loss": 1.5444,
+      "step": 57375
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6010074074074075e-05,
+      "loss": 1.4403,
+      "step": 57380
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.600711111111111e-05,
+      "loss": 1.4722,
+      "step": 57385
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.600414814814815e-05,
+      "loss": 1.4666,
+      "step": 57390
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.6001185185185184e-05,
+      "loss": 1.4422,
+      "step": 57395
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.5998222222222223e-05,
+      "loss": 1.4633,
+      "step": 57400
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.599525925925926e-05,
+      "loss": 1.5391,
+      "step": 57405
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5992296296296298e-05,
+      "loss": 1.4014,
+      "step": 57410
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5989333333333333e-05,
+      "loss": 1.5381,
+      "step": 57415
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5986370370370372e-05,
+      "loss": 1.4941,
+      "step": 57420
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5983407407407407e-05,
+      "loss": 1.5359,
+      "step": 57425
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5980444444444446e-05,
+      "loss": 1.5125,
+      "step": 57430
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.597748148148148e-05,
+      "loss": 1.4809,
+      "step": 57435
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.597451851851852e-05,
+      "loss": 1.4194,
+      "step": 57440
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5971555555555556e-05,
+      "loss": 1.42,
+      "step": 57445
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.5968592592592595e-05,
+      "loss": 1.4757,
+      "step": 57450
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 1.596562962962963e-05,
+      "loss": 1.476,
+      "step": 57455
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.596266666666667e-05,
+      "loss": 1.5578,
+      "step": 57460
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5959703703703704e-05,
+      "loss": 1.5182,
+      "step": 57465
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5956740740740743e-05,
+      "loss": 1.4282,
+      "step": 57470
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.595377777777778e-05,
+      "loss": 1.4785,
+      "step": 57475
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5950814814814817e-05,
+      "loss": 1.5029,
+      "step": 57480
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5947851851851853e-05,
+      "loss": 1.498,
+      "step": 57485
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5944888888888888e-05,
+      "loss": 1.4752,
+      "step": 57490
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5941925925925927e-05,
+      "loss": 1.3562,
+      "step": 57495
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5938962962962962e-05,
+      "loss": 1.5276,
+      "step": 57500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5936e-05,
+      "loss": 1.4798,
+      "step": 57505
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5933037037037037e-05,
+      "loss": 1.4945,
+      "step": 57510
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.5930074074074076e-05,
+      "loss": 1.4772,
+      "step": 57515
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.592711111111111e-05,
+      "loss": 1.3629,
+      "step": 57520
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.592414814814815e-05,
+      "loss": 1.4873,
+      "step": 57525
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5921185185185185e-05,
+      "loss": 1.4423,
+      "step": 57530
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5918222222222224e-05,
+      "loss": 1.3348,
+      "step": 57535
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.591525925925926e-05,
+      "loss": 1.5268,
+      "step": 57540
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5912296296296298e-05,
+      "loss": 1.4804,
+      "step": 57545
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5909333333333334e-05,
+      "loss": 1.4794,
+      "step": 57550
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5906370370370373e-05,
+      "loss": 1.5295,
+      "step": 57555
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5903407407407408e-05,
+      "loss": 1.5312,
+      "step": 57560
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5900444444444447e-05,
+      "loss": 1.5624,
+      "step": 57565
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 1.5897481481481482e-05,
+      "loss": 1.44,
+      "step": 57570
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.589451851851852e-05,
+      "loss": 1.4336,
+      "step": 57575
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5891555555555556e-05,
+      "loss": 1.5577,
+      "step": 57580
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5888592592592595e-05,
+      "loss": 1.4874,
+      "step": 57585
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.588562962962963e-05,
+      "loss": 1.4328,
+      "step": 57590
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.588266666666667e-05,
+      "loss": 1.5161,
+      "step": 57595
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5879703703703705e-05,
+      "loss": 1.5601,
+      "step": 57600
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.587674074074074e-05,
+      "loss": 1.5182,
+      "step": 57605
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.587377777777778e-05,
+      "loss": 1.4069,
+      "step": 57610
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5870814814814815e-05,
+      "loss": 1.453,
+      "step": 57615
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.5867851851851853e-05,
+      "loss": 1.4956,
+      "step": 57620
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 1.586488888888889e-05,
+      "loss": 1.4002,
+      "step": 57625
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5861925925925928e-05,
+      "loss": 1.4868,
+      "step": 57630
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5858962962962963e-05,
+      "loss": 1.4464,
+      "step": 57635
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5856000000000002e-05,
+      "loss": 1.4464,
+      "step": 57640
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5853037037037037e-05,
+      "loss": 1.4961,
+      "step": 57645
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5850074074074076e-05,
+      "loss": 1.4678,
+      "step": 57650
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5847111111111112e-05,
+      "loss": 1.4334,
+      "step": 57655
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.584414814814815e-05,
+      "loss": 1.5021,
+      "step": 57660
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5841185185185186e-05,
+      "loss": 1.487,
+      "step": 57665
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.5838222222222225e-05,
+      "loss": 1.4619,
+      "step": 57670
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.583525925925926e-05,
+      "loss": 1.394,
+      "step": 57675
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.58322962962963e-05,
+      "loss": 1.4126,
+      "step": 57680
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5829333333333334e-05,
+      "loss": 1.3816,
+      "step": 57685
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5826370370370373e-05,
+      "loss": 1.4895,
+      "step": 57690
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.582340740740741e-05,
+      "loss": 1.5423,
+      "step": 57695
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5820444444444448e-05,
+      "loss": 1.4877,
+      "step": 57700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5817481481481483e-05,
+      "loss": 1.3639,
+      "step": 57705
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5814518518518522e-05,
+      "loss": 1.479,
+      "step": 57710
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5811555555555557e-05,
+      "loss": 1.3829,
+      "step": 57715
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5808592592592593e-05,
+      "loss": 1.4367,
+      "step": 57720
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.580562962962963e-05,
+      "loss": 1.4307,
+      "step": 57725
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5802666666666667e-05,
+      "loss": 1.5144,
+      "step": 57730
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.5799703703703706e-05,
+      "loss": 1.3496,
+      "step": 57735
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 1.579674074074074e-05,
+      "loss": 1.4094,
+      "step": 57740
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.579377777777778e-05,
+      "loss": 1.4717,
+      "step": 57745
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5790814814814815e-05,
+      "loss": 1.4042,
+      "step": 57750
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5787851851851854e-05,
+      "loss": 1.4701,
+      "step": 57755
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.578488888888889e-05,
+      "loss": 1.4628,
+      "step": 57760
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.578192592592593e-05,
+      "loss": 1.4795,
+      "step": 57765
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5778962962962964e-05,
+      "loss": 1.5003,
+      "step": 57770
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5776000000000003e-05,
+      "loss": 1.4691,
+      "step": 57775
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5773037037037038e-05,
+      "loss": 1.4658,
+      "step": 57780
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5770074074074077e-05,
+      "loss": 1.4067,
+      "step": 57785
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5767111111111112e-05,
+      "loss": 1.3834,
+      "step": 57790
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.576414814814815e-05,
+      "loss": 1.4586,
+      "step": 57795
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5761185185185187e-05,
+      "loss": 1.5833,
+      "step": 57800
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5758222222222226e-05,
+      "loss": 1.5523,
+      "step": 57805
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.575525925925926e-05,
+      "loss": 1.3894,
+      "step": 57810
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.57522962962963e-05,
+      "loss": 1.4667,
+      "step": 57815
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5749333333333335e-05,
+      "loss": 1.5777,
+      "step": 57820
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5746370370370374e-05,
+      "loss": 1.4126,
+      "step": 57825
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.574340740740741e-05,
+      "loss": 1.4299,
+      "step": 57830
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5740444444444445e-05,
+      "loss": 1.5142,
+      "step": 57835
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5737481481481484e-05,
+      "loss": 1.4349,
+      "step": 57840
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.573451851851852e-05,
+      "loss": 1.5233,
+      "step": 57845
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 1.5731555555555555e-05,
+      "loss": 1.4105,
+      "step": 57850
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5728592592592593e-05,
+      "loss": 1.3908,
+      "step": 57855
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.572562962962963e-05,
+      "loss": 1.3775,
+      "step": 57860
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5722666666666668e-05,
+      "loss": 1.4297,
+      "step": 57865
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5719703703703703e-05,
+      "loss": 1.4513,
+      "step": 57870
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.571674074074074e-05,
+      "loss": 1.5963,
+      "step": 57875
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5713777777777777e-05,
+      "loss": 1.5029,
+      "step": 57880
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5710814814814813e-05,
+      "loss": 1.4527,
+      "step": 57885
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.570785185185185e-05,
+      "loss": 1.4051,
+      "step": 57890
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5704888888888887e-05,
+      "loss": 1.5059,
+      "step": 57895
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.5701925925925926e-05,
+      "loss": 1.4013,
+      "step": 57900
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.569896296296296e-05,
+      "loss": 1.471,
+      "step": 57905
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5696e-05,
+      "loss": 1.4712,
+      "step": 57910
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5693037037037036e-05,
+      "loss": 1.5096,
+      "step": 57915
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5690074074074074e-05,
+      "loss": 1.4099,
+      "step": 57920
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.568711111111111e-05,
+      "loss": 1.399,
+      "step": 57925
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.568414814814815e-05,
+      "loss": 1.4619,
+      "step": 57930
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5681185185185184e-05,
+      "loss": 1.4629,
+      "step": 57935
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5678222222222223e-05,
+      "loss": 1.3985,
+      "step": 57940
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.567525925925926e-05,
+      "loss": 1.3746,
+      "step": 57945
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5672296296296297e-05,
+      "loss": 1.5952,
+      "step": 57950
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5669333333333333e-05,
+      "loss": 1.4213,
+      "step": 57955
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.566637037037037e-05,
+      "loss": 1.4808,
+      "step": 57960
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 1.5663407407407407e-05,
+      "loss": 1.4624,
+      "step": 57965
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5660444444444446e-05,
+      "loss": 1.4431,
+      "step": 57970
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.565748148148148e-05,
+      "loss": 1.5327,
+      "step": 57975
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.565451851851852e-05,
+      "loss": 1.5175,
+      "step": 57980
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5651555555555555e-05,
+      "loss": 1.406,
+      "step": 57985
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.564859259259259e-05,
+      "loss": 1.5648,
+      "step": 57990
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.564562962962963e-05,
+      "loss": 1.5232,
+      "step": 57995
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5642666666666665e-05,
+      "loss": 1.439,
+      "step": 58000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5639703703703704e-05,
+      "loss": 1.5279,
+      "step": 58005
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.563674074074074e-05,
+      "loss": 1.466,
+      "step": 58010
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5633777777777778e-05,
+      "loss": 1.6846,
+      "step": 58015
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.5630814814814814e-05,
+      "loss": 1.5833,
+      "step": 58020
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5627851851851852e-05,
+      "loss": 1.3396,
+      "step": 58025
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5624888888888888e-05,
+      "loss": 1.4698,
+      "step": 58030
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5621925925925927e-05,
+      "loss": 1.4349,
+      "step": 58035
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5618962962962962e-05,
+      "loss": 1.5199,
+      "step": 58040
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5616e-05,
+      "loss": 1.4957,
+      "step": 58045
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5613037037037036e-05,
+      "loss": 1.4374,
+      "step": 58050
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5610074074074075e-05,
+      "loss": 1.5204,
+      "step": 58055
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.560711111111111e-05,
+      "loss": 1.55,
+      "step": 58060
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.560414814814815e-05,
+      "loss": 1.4052,
+      "step": 58065
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5601185185185185e-05,
+      "loss": 1.6104,
+      "step": 58070
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 1.5598222222222224e-05,
+      "loss": 1.4262,
+      "step": 58075
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.559525925925926e-05,
+      "loss": 1.4609,
+      "step": 58080
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5592296296296298e-05,
+      "loss": 1.384,
+      "step": 58085
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5589333333333333e-05,
+      "loss": 1.4125,
+      "step": 58090
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5586370370370372e-05,
+      "loss": 1.2276,
+      "step": 58095
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5583407407407408e-05,
+      "loss": 1.5046,
+      "step": 58100
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5580444444444443e-05,
+      "loss": 1.5841,
+      "step": 58105
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5577481481481482e-05,
+      "loss": 1.3791,
+      "step": 58110
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5574518518518517e-05,
+      "loss": 1.382,
+      "step": 58115
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.5571555555555556e-05,
+      "loss": 1.4543,
+      "step": 58120
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.556859259259259e-05,
+      "loss": 1.5362,
+      "step": 58125
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 1.556562962962963e-05,
+      "loss": 1.5292,
+      "step": 58130
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5562666666666666e-05,
+      "loss": 1.4859,
+      "step": 58135
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5559703703703705e-05,
+      "loss": 1.5177,
+      "step": 58140
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.555674074074074e-05,
+      "loss": 1.4574,
+      "step": 58145
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.555377777777778e-05,
+      "loss": 1.4358,
+      "step": 58150
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5550814814814814e-05,
+      "loss": 1.515,
+      "step": 58155
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5547851851851853e-05,
+      "loss": 1.4314,
+      "step": 58160
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.554488888888889e-05,
+      "loss": 1.4331,
+      "step": 58165
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5541925925925927e-05,
+      "loss": 1.4103,
+      "step": 58170
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5538962962962963e-05,
+      "loss": 1.5063,
+      "step": 58175
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5536e-05,
+      "loss": 1.4246,
+      "step": 58180
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5533037037037037e-05,
+      "loss": 1.4631,
+      "step": 58185
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.5530074074074076e-05,
+      "loss": 1.4796,
+      "step": 58190
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.552711111111111e-05,
+      "loss": 1.4138,
+      "step": 58195
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.552414814814815e-05,
+      "loss": 1.4656,
+      "step": 58200
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5521185185185186e-05,
+      "loss": 1.4755,
+      "step": 58205
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5518222222222224e-05,
+      "loss": 1.5233,
+      "step": 58210
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.551525925925926e-05,
+      "loss": 1.4286,
+      "step": 58215
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5512296296296295e-05,
+      "loss": 1.5411,
+      "step": 58220
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5509333333333334e-05,
+      "loss": 1.3792,
+      "step": 58225
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.550637037037037e-05,
+      "loss": 1.4245,
+      "step": 58230
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.550340740740741e-05,
+      "loss": 1.4686,
+      "step": 58235
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5500444444444444e-05,
+      "loss": 1.4294,
+      "step": 58240
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 1.5497481481481483e-05,
+      "loss": 1.4766,
+      "step": 58245
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5494518518518518e-05,
+      "loss": 1.4996,
+      "step": 58250
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5491555555555557e-05,
+      "loss": 1.3917,
+      "step": 58255
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5488592592592592e-05,
+      "loss": 1.4286,
+      "step": 58260
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.548562962962963e-05,
+      "loss": 1.4053,
+      "step": 58265
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5482666666666667e-05,
+      "loss": 1.441,
+      "step": 58270
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5479703703703705e-05,
+      "loss": 1.4097,
+      "step": 58275
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.547674074074074e-05,
+      "loss": 1.4893,
+      "step": 58280
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.547377777777778e-05,
+      "loss": 1.5554,
+      "step": 58285
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5470814814814815e-05,
+      "loss": 1.3852,
+      "step": 58290
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.5467851851851854e-05,
+      "loss": 1.5069,
+      "step": 58295
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.546488888888889e-05,
+      "loss": 1.432,
+      "step": 58300
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5461925925925928e-05,
+      "loss": 1.555,
+      "step": 58305
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5458962962962964e-05,
+      "loss": 1.472,
+      "step": 58310
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5456000000000002e-05,
+      "loss": 1.3646,
+      "step": 58315
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5453037037037038e-05,
+      "loss": 1.4823,
+      "step": 58320
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5450074074074077e-05,
+      "loss": 1.5113,
+      "step": 58325
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5447111111111112e-05,
+      "loss": 1.5456,
+      "step": 58330
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5444148148148148e-05,
+      "loss": 1.5348,
+      "step": 58335
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5441185185185186e-05,
+      "loss": 1.4874,
+      "step": 58340
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5438222222222222e-05,
+      "loss": 1.4757,
+      "step": 58345
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.543525925925926e-05,
+      "loss": 1.4871,
+      "step": 58350
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 1.5432296296296296e-05,
+      "loss": 1.4621,
+      "step": 58355
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5429333333333335e-05,
+      "loss": 1.4622,
+      "step": 58360
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.542637037037037e-05,
+      "loss": 1.4975,
+      "step": 58365
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.542340740740741e-05,
+      "loss": 1.4183,
+      "step": 58370
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5420444444444445e-05,
+      "loss": 1.4316,
+      "step": 58375
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5417481481481483e-05,
+      "loss": 1.5553,
+      "step": 58380
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.541451851851852e-05,
+      "loss": 1.4925,
+      "step": 58385
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5411555555555558e-05,
+      "loss": 1.4499,
+      "step": 58390
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5408592592592593e-05,
+      "loss": 1.5471,
+      "step": 58395
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5405629629629632e-05,
+      "loss": 1.5105,
+      "step": 58400
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5402666666666667e-05,
+      "loss": 1.4681,
+      "step": 58405
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.5399703703703706e-05,
+      "loss": 1.5636,
+      "step": 58410
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.539674074074074e-05,
+      "loss": 1.4542,
+      "step": 58415
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.539377777777778e-05,
+      "loss": 1.4576,
+      "step": 58420
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5390814814814816e-05,
+      "loss": 1.51,
+      "step": 58425
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5387851851851855e-05,
+      "loss": 1.5478,
+      "step": 58430
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.538488888888889e-05,
+      "loss": 1.4927,
+      "step": 58435
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.538192592592593e-05,
+      "loss": 1.5992,
+      "step": 58440
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5378962962962964e-05,
+      "loss": 1.4584,
+      "step": 58445
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5376e-05,
+      "loss": 1.548,
+      "step": 58450
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.537303703703704e-05,
+      "loss": 1.4723,
+      "step": 58455
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5370074074074074e-05,
+      "loss": 1.5418,
+      "step": 58460
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.5367111111111113e-05,
+      "loss": 1.4756,
+      "step": 58465
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 1.536414814814815e-05,
+      "loss": 1.3878,
+      "step": 58470
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5361185185185187e-05,
+      "loss": 1.4383,
+      "step": 58475
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5358222222222223e-05,
+      "loss": 1.4408,
+      "step": 58480
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.535525925925926e-05,
+      "loss": 1.5634,
+      "step": 58485
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5352296296296297e-05,
+      "loss": 1.4814,
+      "step": 58490
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5349333333333336e-05,
+      "loss": 1.5613,
+      "step": 58495
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.534637037037037e-05,
+      "loss": 1.5315,
+      "step": 58500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.534340740740741e-05,
+      "loss": 1.4643,
+      "step": 58505
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5340444444444445e-05,
+      "loss": 1.4133,
+      "step": 58510
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.5337481481481484e-05,
+      "loss": 1.4707,
+      "step": 58515
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.533451851851852e-05,
+      "loss": 1.5165,
+      "step": 58520
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.533155555555556e-05,
+      "loss": 1.4408,
+      "step": 58525
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5328592592592594e-05,
+      "loss": 1.4326,
+      "step": 58530
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5325629629629633e-05,
+      "loss": 1.5168,
+      "step": 58535
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5322666666666668e-05,
+      "loss": 1.4176,
+      "step": 58540
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5319703703703707e-05,
+      "loss": 1.4434,
+      "step": 58545
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5316740740740742e-05,
+      "loss": 1.4942,
+      "step": 58550
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.531377777777778e-05,
+      "loss": 1.476,
+      "step": 58555
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5310814814814817e-05,
+      "loss": 1.4115,
+      "step": 58560
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5307851851851852e-05,
+      "loss": 1.492,
+      "step": 58565
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.530488888888889e-05,
+      "loss": 1.5896,
+      "step": 58570
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5301925925925926e-05,
+      "loss": 1.4734,
+      "step": 58575
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 1.5298962962962965e-05,
+      "loss": 1.545,
+      "step": 58580
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5296e-05,
+      "loss": 1.4105,
+      "step": 58585
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.529303703703704e-05,
+      "loss": 1.415,
+      "step": 58590
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5290074074074075e-05,
+      "loss": 1.3821,
+      "step": 58595
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5287111111111114e-05,
+      "loss": 1.4481,
+      "step": 58600
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.528414814814815e-05,
+      "loss": 1.5852,
+      "step": 58605
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5281185185185188e-05,
+      "loss": 1.4181,
+      "step": 58610
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5278222222222223e-05,
+      "loss": 1.6013,
+      "step": 58615
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5275259259259262e-05,
+      "loss": 1.6393,
+      "step": 58620
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5272296296296294e-05,
+      "loss": 1.5386,
+      "step": 58625
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5269333333333333e-05,
+      "loss": 1.5359,
+      "step": 58630
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.526637037037037e-05,
+      "loss": 1.4954,
+      "step": 58635
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 1.5263407407407407e-05,
+      "loss": 1.4417,
+      "step": 58640
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5260444444444443e-05,
+      "loss": 1.5191,
+      "step": 58645
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5257481481481482e-05,
+      "loss": 1.4004,
+      "step": 58650
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5254518518518519e-05,
+      "loss": 1.5465,
+      "step": 58655
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5251555555555556e-05,
+      "loss": 1.486,
+      "step": 58660
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5248592592592591e-05,
+      "loss": 1.5559,
+      "step": 58665
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5245629629629628e-05,
+      "loss": 1.5072,
+      "step": 58670
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5242666666666666e-05,
+      "loss": 1.5274,
+      "step": 58675
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5239703703703703e-05,
+      "loss": 1.4976,
+      "step": 58680
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.523674074074074e-05,
+      "loss": 1.4645,
+      "step": 58685
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5233777777777777e-05,
+      "loss": 1.5053,
+      "step": 58690
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.5230814814814814e-05,
+      "loss": 1.4699,
+      "step": 58695
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5227851851851851e-05,
+      "loss": 1.4412,
+      "step": 58700
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5224888888888888e-05,
+      "loss": 1.4507,
+      "step": 58705
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5221925925925925e-05,
+      "loss": 1.3676,
+      "step": 58710
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5218962962962963e-05,
+      "loss": 1.5188,
+      "step": 58715
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5216e-05,
+      "loss": 1.401,
+      "step": 58720
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5213037037037037e-05,
+      "loss": 1.556,
+      "step": 58725
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5210074074074074e-05,
+      "loss": 1.5186,
+      "step": 58730
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5207111111111111e-05,
+      "loss": 1.5115,
+      "step": 58735
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5204148148148148e-05,
+      "loss": 1.4199,
+      "step": 58740
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5201185185185185e-05,
+      "loss": 1.3656,
+      "step": 58745
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 1.5198222222222222e-05,
+      "loss": 1.6178,
+      "step": 58750
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.519525925925926e-05,
+      "loss": 1.6593,
+      "step": 58755
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5192296296296297e-05,
+      "loss": 1.5589,
+      "step": 58760
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5189333333333334e-05,
+      "loss": 1.4319,
+      "step": 58765
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5186370370370371e-05,
+      "loss": 1.5043,
+      "step": 58770
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5183407407407408e-05,
+      "loss": 1.4125,
+      "step": 58775
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5180444444444444e-05,
+      "loss": 1.6009,
+      "step": 58780
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.517748148148148e-05,
+      "loss": 1.5148,
+      "step": 58785
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5174518518518518e-05,
+      "loss": 1.3878,
+      "step": 58790
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5171555555555555e-05,
+      "loss": 1.3321,
+      "step": 58795
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5168592592592592e-05,
+      "loss": 1.4368,
+      "step": 58800
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.5165629629629629e-05,
+      "loss": 1.4802,
+      "step": 58805
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5162666666666666e-05,
+      "loss": 1.4354,
+      "step": 58810
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5159703703703703e-05,
+      "loss": 1.4217,
+      "step": 58815
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.515674074074074e-05,
+      "loss": 1.604,
+      "step": 58820
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5153777777777778e-05,
+      "loss": 1.5944,
+      "step": 58825
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5150814814814815e-05,
+      "loss": 1.5013,
+      "step": 58830
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5147851851851852e-05,
+      "loss": 1.5318,
+      "step": 58835
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5144888888888889e-05,
+      "loss": 1.4149,
+      "step": 58840
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5141925925925926e-05,
+      "loss": 1.5226,
+      "step": 58845
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5138962962962963e-05,
+      "loss": 1.4291,
+      "step": 58850
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5136e-05,
+      "loss": 1.3898,
+      "step": 58855
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5133037037037038e-05,
+      "loss": 1.5512,
+      "step": 58860
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 1.5130074074074075e-05,
+      "loss": 1.5051,
+      "step": 58865
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5127111111111112e-05,
+      "loss": 1.5904,
+      "step": 58870
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5124148148148149e-05,
+      "loss": 1.5698,
+      "step": 58875
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5121185185185186e-05,
+      "loss": 1.4048,
+      "step": 58880
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5118222222222223e-05,
+      "loss": 1.5495,
+      "step": 58885
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.511525925925926e-05,
+      "loss": 1.4548,
+      "step": 58890
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5112296296296297e-05,
+      "loss": 1.4213,
+      "step": 58895
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5109333333333333e-05,
+      "loss": 1.4604,
+      "step": 58900
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.510637037037037e-05,
+      "loss": 1.5712,
+      "step": 58905
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5103407407407407e-05,
+      "loss": 1.3939,
+      "step": 58910
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5100444444444444e-05,
+      "loss": 1.4122,
+      "step": 58915
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.5097481481481481e-05,
+      "loss": 1.4492,
+      "step": 58920
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5094518518518519e-05,
+      "loss": 1.5293,
+      "step": 58925
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5091555555555556e-05,
+      "loss": 1.5234,
+      "step": 58930
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5088592592592593e-05,
+      "loss": 1.4366,
+      "step": 58935
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.508562962962963e-05,
+      "loss": 1.4986,
+      "step": 58940
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5082666666666667e-05,
+      "loss": 1.4934,
+      "step": 58945
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5079703703703704e-05,
+      "loss": 1.4953,
+      "step": 58950
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5076740740740741e-05,
+      "loss": 1.4866,
+      "step": 58955
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5073777777777778e-05,
+      "loss": 1.5742,
+      "step": 58960
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5070814814814816e-05,
+      "loss": 1.4206,
+      "step": 58965
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.5067851851851853e-05,
+      "loss": 1.401,
+      "step": 58970
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 1.506488888888889e-05,
+      "loss": 1.344,
+      "step": 58975
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5061925925925927e-05,
+      "loss": 1.4353,
+      "step": 58980
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5058962962962964e-05,
+      "loss": 1.5807,
+      "step": 58985
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5056000000000001e-05,
+      "loss": 1.3981,
+      "step": 58990
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5053037037037038e-05,
+      "loss": 1.3942,
+      "step": 58995
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5050074074074075e-05,
+      "loss": 1.5138,
+      "step": 59000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5047111111111113e-05,
+      "loss": 1.5463,
+      "step": 59005
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.504414814814815e-05,
+      "loss": 1.5116,
+      "step": 59010
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5041185185185185e-05,
+      "loss": 1.4823,
+      "step": 59015
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5038222222222222e-05,
+      "loss": 1.4437,
+      "step": 59020
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.503525925925926e-05,
+      "loss": 1.5179,
+      "step": 59025
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.5032296296296297e-05,
+      "loss": 1.4892,
+      "step": 59030
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5029333333333334e-05,
+      "loss": 1.4386,
+      "step": 59035
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.502637037037037e-05,
+      "loss": 1.5707,
+      "step": 59040
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5023407407407408e-05,
+      "loss": 1.3956,
+      "step": 59045
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5020444444444445e-05,
+      "loss": 1.6347,
+      "step": 59050
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5017481481481482e-05,
+      "loss": 1.5217,
+      "step": 59055
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.501451851851852e-05,
+      "loss": 1.5199,
+      "step": 59060
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5011555555555556e-05,
+      "loss": 1.373,
+      "step": 59065
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5008592592592594e-05,
+      "loss": 1.5163,
+      "step": 59070
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.500562962962963e-05,
+      "loss": 1.5799,
+      "step": 59075
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.5002666666666668e-05,
+      "loss": 1.5092,
+      "step": 59080
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.4999703703703705e-05,
+      "loss": 1.4816,
+      "step": 59085
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 1.4996740740740742e-05,
+      "loss": 1.4873,
+      "step": 59090
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.499377777777778e-05,
+      "loss": 1.4824,
+      "step": 59095
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4990814814814816e-05,
+      "loss": 1.4378,
+      "step": 59100
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4987851851851853e-05,
+      "loss": 1.4483,
+      "step": 59105
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.498488888888889e-05,
+      "loss": 1.5145,
+      "step": 59110
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4981925925925928e-05,
+      "loss": 1.5637,
+      "step": 59115
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4978962962962965e-05,
+      "loss": 1.5308,
+      "step": 59120
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4976000000000002e-05,
+      "loss": 1.521,
+      "step": 59125
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4973037037037037e-05,
+      "loss": 1.3467,
+      "step": 59130
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4970074074074075e-05,
+      "loss": 1.4773,
+      "step": 59135
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4967111111111112e-05,
+      "loss": 1.4491,
+      "step": 59140
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 1.4964148148148149e-05,
+      "loss": 1.4323,
+      "step": 59145
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4961185185185186e-05,
+      "loss": 1.3991,
+      "step": 59150
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4958222222222223e-05,
+      "loss": 1.5583,
+      "step": 59155
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.495525925925926e-05,
+      "loss": 1.444,
+      "step": 59160
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4952296296296297e-05,
+      "loss": 1.4432,
+      "step": 59165
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4949333333333334e-05,
+      "loss": 1.3875,
+      "step": 59170
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4946370370370372e-05,
+      "loss": 1.4553,
+      "step": 59175
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4943407407407409e-05,
+      "loss": 1.498,
+      "step": 59180
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4940444444444446e-05,
+      "loss": 1.3265,
+      "step": 59185
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4937481481481483e-05,
+      "loss": 1.5591,
+      "step": 59190
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.493451851851852e-05,
+      "loss": 1.3483,
+      "step": 59195
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.4931555555555557e-05,
+      "loss": 1.5502,
+      "step": 59200
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4928592592592594e-05,
+      "loss": 1.4743,
+      "step": 59205
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4925629629629631e-05,
+      "loss": 1.4783,
+      "step": 59210
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4922666666666669e-05,
+      "loss": 1.439,
+      "step": 59215
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4919703703703706e-05,
+      "loss": 1.3919,
+      "step": 59220
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4916740740740743e-05,
+      "loss": 1.5234,
+      "step": 59225
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.491377777777778e-05,
+      "loss": 1.4989,
+      "step": 59230
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4910814814814817e-05,
+      "loss": 1.4481,
+      "step": 59235
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4907851851851854e-05,
+      "loss": 1.5067,
+      "step": 59240
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.490488888888889e-05,
+      "loss": 1.5316,
+      "step": 59245
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4901925925925927e-05,
+      "loss": 1.5063,
+      "step": 59250
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 1.4898962962962964e-05,
+      "loss": 1.4686,
+      "step": 59255
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4896000000000001e-05,
+      "loss": 1.4907,
+      "step": 59260
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4893037037037038e-05,
+      "loss": 1.449,
+      "step": 59265
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4890074074074075e-05,
+      "loss": 1.4979,
+      "step": 59270
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4887111111111112e-05,
+      "loss": 1.518,
+      "step": 59275
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.488414814814815e-05,
+      "loss": 1.527,
+      "step": 59280
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4881185185185187e-05,
+      "loss": 1.4579,
+      "step": 59285
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4878222222222224e-05,
+      "loss": 1.3996,
+      "step": 59290
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4875259259259261e-05,
+      "loss": 1.4874,
+      "step": 59295
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4872296296296298e-05,
+      "loss": 1.5069,
+      "step": 59300
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4869333333333335e-05,
+      "loss": 1.5242,
+      "step": 59305
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.4866370370370372e-05,
+      "loss": 1.5465,
+      "step": 59310
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.486340740740741e-05,
+      "loss": 1.5863,
+      "step": 59315
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4860444444444447e-05,
+      "loss": 1.5668,
+      "step": 59320
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4857481481481484e-05,
+      "loss": 1.5052,
+      "step": 59325
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.485451851851852e-05,
+      "loss": 1.5289,
+      "step": 59330
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4851555555555558e-05,
+      "loss": 1.5435,
+      "step": 59335
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4848592592592595e-05,
+      "loss": 1.5255,
+      "step": 59340
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4845629629629632e-05,
+      "loss": 1.4009,
+      "step": 59345
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.484266666666667e-05,
+      "loss": 1.4632,
+      "step": 59350
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4839703703703706e-05,
+      "loss": 1.4995,
+      "step": 59355
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4836740740740742e-05,
+      "loss": 1.5939,
+      "step": 59360
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4833777777777779e-05,
+      "loss": 1.5393,
+      "step": 59365
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 1.4830814814814816e-05,
+      "loss": 1.312,
+      "step": 59370
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4827851851851853e-05,
+      "loss": 1.4138,
+      "step": 59375
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.482488888888889e-05,
+      "loss": 1.5072,
+      "step": 59380
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4821925925925928e-05,
+      "loss": 1.5742,
+      "step": 59385
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4818962962962965e-05,
+      "loss": 1.44,
+      "step": 59390
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4816e-05,
+      "loss": 1.4811,
+      "step": 59395
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4813037037037036e-05,
+      "loss": 1.3953,
+      "step": 59400
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4810074074074073e-05,
+      "loss": 1.465,
+      "step": 59405
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.480711111111111e-05,
+      "loss": 1.4889,
+      "step": 59410
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4804148148148147e-05,
+      "loss": 1.5823,
+      "step": 59415
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4801185185185184e-05,
+      "loss": 1.4349,
+      "step": 59420
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.4798222222222221e-05,
+      "loss": 1.4311,
+      "step": 59425
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4795259259259258e-05,
+      "loss": 1.5111,
+      "step": 59430
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4792296296296295e-05,
+      "loss": 1.4303,
+      "step": 59435
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4789333333333333e-05,
+      "loss": 1.434,
+      "step": 59440
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.478637037037037e-05,
+      "loss": 1.4622,
+      "step": 59445
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4783407407407407e-05,
+      "loss": 1.4281,
+      "step": 59450
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4780444444444444e-05,
+      "loss": 1.5744,
+      "step": 59455
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4777481481481481e-05,
+      "loss": 1.5579,
+      "step": 59460
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4774518518518518e-05,
+      "loss": 1.5868,
+      "step": 59465
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4771555555555555e-05,
+      "loss": 1.5097,
+      "step": 59470
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.4768592592592592e-05,
+      "loss": 1.3369,
+      "step": 59475
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 1.476562962962963e-05,
+      "loss": 1.4505,
+      "step": 59480
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4762666666666667e-05,
+      "loss": 1.4633,
+      "step": 59485
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4759703703703704e-05,
+      "loss": 1.5249,
+      "step": 59490
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4756740740740741e-05,
+      "loss": 1.5538,
+      "step": 59495
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4753777777777778e-05,
+      "loss": 1.4517,
+      "step": 59500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4750814814814815e-05,
+      "loss": 1.5316,
+      "step": 59505
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4747851851851852e-05,
+      "loss": 1.4518,
+      "step": 59510
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4744888888888888e-05,
+      "loss": 1.479,
+      "step": 59515
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4741925925925925e-05,
+      "loss": 1.342,
+      "step": 59520
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4738962962962962e-05,
+      "loss": 1.4356,
+      "step": 59525
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4736e-05,
+      "loss": 1.513,
+      "step": 59530
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4733037037037036e-05,
+      "loss": 1.4033,
+      "step": 59535
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.4730074074074073e-05,
+      "loss": 1.4627,
+      "step": 59540
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.472711111111111e-05,
+      "loss": 1.5144,
+      "step": 59545
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4724148148148148e-05,
+      "loss": 1.4642,
+      "step": 59550
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4721185185185185e-05,
+      "loss": 1.6015,
+      "step": 59555
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4718222222222222e-05,
+      "loss": 1.4545,
+      "step": 59560
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4715259259259259e-05,
+      "loss": 1.4569,
+      "step": 59565
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4712296296296296e-05,
+      "loss": 1.5536,
+      "step": 59570
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4709333333333333e-05,
+      "loss": 1.574,
+      "step": 59575
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.470637037037037e-05,
+      "loss": 1.5928,
+      "step": 59580
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4703407407407408e-05,
+      "loss": 1.557,
+      "step": 59585
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4700444444444445e-05,
+      "loss": 1.5245,
+      "step": 59590
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 1.4697481481481482e-05,
+      "loss": 1.5897,
+      "step": 59595
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4694518518518519e-05,
+      "loss": 1.4137,
+      "step": 59600
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4691555555555556e-05,
+      "loss": 1.5963,
+      "step": 59605
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4688592592592593e-05,
+      "loss": 1.4987,
+      "step": 59610
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.468562962962963e-05,
+      "loss": 1.4535,
+      "step": 59615
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4682666666666667e-05,
+      "loss": 1.4947,
+      "step": 59620
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4679703703703705e-05,
+      "loss": 1.5308,
+      "step": 59625
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4676740740740742e-05,
+      "loss": 1.4693,
+      "step": 59630
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4673777777777777e-05,
+      "loss": 1.5759,
+      "step": 59635
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4670814814814814e-05,
+      "loss": 1.489,
+      "step": 59640
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4667851851851851e-05,
+      "loss": 1.4272,
+      "step": 59645
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 1.4664888888888889e-05,
+      "loss": 1.4291,
+      "step": 59650
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4661925925925926e-05,
+      "loss": 1.562,
+      "step": 59655
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4658962962962963e-05,
+      "loss": 1.4417,
+      "step": 59660
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4656e-05,
+      "loss": 1.5057,
+      "step": 59665
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4653037037037037e-05,
+      "loss": 1.5495,
+      "step": 59670
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4650074074074074e-05,
+      "loss": 1.5777,
+      "step": 59675
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4647111111111111e-05,
+      "loss": 1.4937,
+      "step": 59680
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4644148148148148e-05,
+      "loss": 1.5622,
+      "step": 59685
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4641185185185186e-05,
+      "loss": 1.5449,
+      "step": 59690
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4638222222222223e-05,
+      "loss": 1.4458,
+      "step": 59695
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.463525925925926e-05,
+      "loss": 1.5572,
+      "step": 59700
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.4632296296296297e-05,
+      "loss": 1.3597,
+      "step": 59705
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4629333333333334e-05,
+      "loss": 1.5151,
+      "step": 59710
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4626370370370371e-05,
+      "loss": 1.4862,
+      "step": 59715
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4623407407407408e-05,
+      "loss": 1.478,
+      "step": 59720
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4620444444444445e-05,
+      "loss": 1.4345,
+      "step": 59725
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4617481481481483e-05,
+      "loss": 1.5415,
+      "step": 59730
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.461451851851852e-05,
+      "loss": 1.6176,
+      "step": 59735
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4611555555555557e-05,
+      "loss": 1.563,
+      "step": 59740
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4608592592592594e-05,
+      "loss": 1.3982,
+      "step": 59745
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.460562962962963e-05,
+      "loss": 1.4572,
+      "step": 59750
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4602666666666667e-05,
+      "loss": 1.3848,
+      "step": 59755
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.4599703703703704e-05,
+      "loss": 1.4139,
+      "step": 59760
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 1.459674074074074e-05,
+      "loss": 1.5477,
+      "step": 59765
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4593777777777778e-05,
+      "loss": 1.4756,
+      "step": 59770
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4590814814814815e-05,
+      "loss": 1.5747,
+      "step": 59775
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4587851851851852e-05,
+      "loss": 1.5008,
+      "step": 59780
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.458488888888889e-05,
+      "loss": 1.5302,
+      "step": 59785
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4581925925925926e-05,
+      "loss": 1.3975,
+      "step": 59790
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4578962962962964e-05,
+      "loss": 1.4491,
+      "step": 59795
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4576e-05,
+      "loss": 1.4996,
+      "step": 59800
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4573037037037038e-05,
+      "loss": 1.5466,
+      "step": 59805
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4570074074074075e-05,
+      "loss": 1.6001,
+      "step": 59810
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.4567111111111112e-05,
+      "loss": 1.469,
+      "step": 59815
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.456414814814815e-05,
+      "loss": 1.5242,
+      "step": 59820
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4561185185185186e-05,
+      "loss": 1.3997,
+      "step": 59825
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4558222222222223e-05,
+      "loss": 1.3717,
+      "step": 59830
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.455525925925926e-05,
+      "loss": 1.4614,
+      "step": 59835
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4552296296296298e-05,
+      "loss": 1.4632,
+      "step": 59840
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4549333333333335e-05,
+      "loss": 1.5575,
+      "step": 59845
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4546370370370372e-05,
+      "loss": 1.5949,
+      "step": 59850
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4543407407407409e-05,
+      "loss": 1.5025,
+      "step": 59855
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4540444444444446e-05,
+      "loss": 1.5178,
+      "step": 59860
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4537481481481482e-05,
+      "loss": 1.4239,
+      "step": 59865
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4534518518518519e-05,
+      "loss": 1.4941,
+      "step": 59870
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 1.4531555555555556e-05,
+      "loss": 1.4488,
+      "step": 59875
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4528592592592593e-05,
+      "loss": 1.4208,
+      "step": 59880
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.452562962962963e-05,
+      "loss": 1.5486,
+      "step": 59885
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4522666666666667e-05,
+      "loss": 1.5851,
+      "step": 59890
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4519703703703704e-05,
+      "loss": 1.4702,
+      "step": 59895
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4516740740740742e-05,
+      "loss": 1.4877,
+      "step": 59900
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4513777777777779e-05,
+      "loss": 1.5204,
+      "step": 59905
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4510814814814816e-05,
+      "loss": 1.5545,
+      "step": 59910
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4507851851851853e-05,
+      "loss": 1.4911,
+      "step": 59915
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.450488888888889e-05,
+      "loss": 1.4137,
+      "step": 59920
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4501925925925927e-05,
+      "loss": 1.5622,
+      "step": 59925
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.4498962962962964e-05,
+      "loss": 1.4953,
+      "step": 59930
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4496000000000001e-05,
+      "loss": 1.5193,
+      "step": 59935
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4493037037037039e-05,
+      "loss": 1.539,
+      "step": 59940
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4490074074074076e-05,
+      "loss": 1.4207,
+      "step": 59945
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4487111111111113e-05,
+      "loss": 1.5879,
+      "step": 59950
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.448414814814815e-05,
+      "loss": 1.4598,
+      "step": 59955
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4481185185185187e-05,
+      "loss": 1.472,
+      "step": 59960
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4478222222222224e-05,
+      "loss": 1.5657,
+      "step": 59965
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4475259259259261e-05,
+      "loss": 1.4782,
+      "step": 59970
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4472296296296299e-05,
+      "loss": 1.5402,
+      "step": 59975
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4469333333333334e-05,
+      "loss": 1.544,
+      "step": 59980
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4466370370370371e-05,
+      "loss": 1.3854,
+      "step": 59985
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 1.4463407407407408e-05,
+      "loss": 1.4544,
+      "step": 59990
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4460444444444445e-05,
+      "loss": 1.5513,
+      "step": 59995
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4457481481481482e-05,
+      "loss": 1.5374,
+      "step": 60000
+    },
+    {
+      "epoch": 10.67,
+      "eval_loss": 1.3671331405639648,
+      "eval_rouge2_fmeasure": 0.2064,
+      "eval_rouge2_precision": 0.241,
+      "eval_rouge2_recall": 0.1899,
+      "eval_runtime": 37898.643,
+      "eval_samples_per_second": 0.132,
+      "eval_steps_per_second": 0.066,
+      "step": 60000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.445451851851852e-05,
+      "loss": 1.5456,
+      "step": 60005
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4451555555555557e-05,
+      "loss": 1.5704,
+      "step": 60010
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4448592592592594e-05,
+      "loss": 1.4521,
+      "step": 60015
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4445629629629631e-05,
+      "loss": 1.4357,
+      "step": 60020
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4442666666666668e-05,
+      "loss": 1.5525,
+      "step": 60025
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4439703703703705e-05,
+      "loss": 1.5162,
+      "step": 60030
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4436740740740742e-05,
+      "loss": 1.4784,
+      "step": 60035
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.443377777777778e-05,
+      "loss": 1.6106,
+      "step": 60040
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.4430814814814817e-05,
+      "loss": 1.4565,
+      "step": 60045
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4427851851851854e-05,
+      "loss": 1.5233,
+      "step": 60050
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4424888888888891e-05,
+      "loss": 1.5845,
+      "step": 60055
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4421925925925928e-05,
+      "loss": 1.4287,
+      "step": 60060
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4418962962962965e-05,
+      "loss": 1.5147,
+      "step": 60065
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4416000000000002e-05,
+      "loss": 1.5221,
+      "step": 60070
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.441303703703704e-05,
+      "loss": 1.4068,
+      "step": 60075
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4410074074074077e-05,
+      "loss": 1.4833,
+      "step": 60080
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4407111111111114e-05,
+      "loss": 1.6188,
+      "step": 60085
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.440414814814815e-05,
+      "loss": 1.5162,
+      "step": 60090
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4401185185185186e-05,
+      "loss": 1.332,
+      "step": 60095
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 1.4398222222222223e-05,
+      "loss": 1.5525,
+      "step": 60100
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.439525925925926e-05,
+      "loss": 1.3926,
+      "step": 60105
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4392296296296298e-05,
+      "loss": 1.4938,
+      "step": 60110
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4389333333333335e-05,
+      "loss": 1.4017,
+      "step": 60115
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4386370370370372e-05,
+      "loss": 1.4176,
+      "step": 60120
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4383407407407409e-05,
+      "loss": 1.5972,
+      "step": 60125
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4380444444444446e-05,
+      "loss": 1.4271,
+      "step": 60130
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4377481481481483e-05,
+      "loss": 1.4331,
+      "step": 60135
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.437451851851852e-05,
+      "loss": 1.5158,
+      "step": 60140
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4371555555555557e-05,
+      "loss": 1.4903,
+      "step": 60145
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4368592592592595e-05,
+      "loss": 1.512,
+      "step": 60150
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 1.4365629629629632e-05,
+      "loss": 1.407,
+      "step": 60155
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4362666666666669e-05,
+      "loss": 1.5375,
+      "step": 60160
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4359703703703706e-05,
+      "loss": 1.4681,
+      "step": 60165
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.435674074074074e-05,
+      "loss": 1.4373,
+      "step": 60170
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4353777777777777e-05,
+      "loss": 1.5577,
+      "step": 60175
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4350814814814814e-05,
+      "loss": 1.4308,
+      "step": 60180
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4347851851851851e-05,
+      "loss": 1.456,
+      "step": 60185
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4344888888888888e-05,
+      "loss": 1.4482,
+      "step": 60190
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4341925925925925e-05,
+      "loss": 1.5602,
+      "step": 60195
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4338962962962962e-05,
+      "loss": 1.5001,
+      "step": 60200
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4336e-05,
+      "loss": 1.4545,
+      "step": 60205
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4333037037037037e-05,
+      "loss": 1.4525,
+      "step": 60210
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.4330074074074074e-05,
+      "loss": 1.5842,
+      "step": 60215
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4327111111111111e-05,
+      "loss": 1.4506,
+      "step": 60220
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4324148148148148e-05,
+      "loss": 1.5572,
+      "step": 60225
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4321185185185185e-05,
+      "loss": 1.3788,
+      "step": 60230
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4318222222222222e-05,
+      "loss": 1.5026,
+      "step": 60235
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.431525925925926e-05,
+      "loss": 1.5083,
+      "step": 60240
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4312296296296297e-05,
+      "loss": 1.4996,
+      "step": 60245
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4309333333333332e-05,
+      "loss": 1.4328,
+      "step": 60250
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.430637037037037e-05,
+      "loss": 1.4383,
+      "step": 60255
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4303407407407406e-05,
+      "loss": 1.4603,
+      "step": 60260
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.4300444444444443e-05,
+      "loss": 1.6414,
+      "step": 60265
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 1.429748148148148e-05,
+      "loss": 1.5872,
+      "step": 60270
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4294518518518518e-05,
+      "loss": 1.4806,
+      "step": 60275
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4291555555555555e-05,
+      "loss": 1.3273,
+      "step": 60280
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4288592592592592e-05,
+      "loss": 1.4482,
+      "step": 60285
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4285629629629629e-05,
+      "loss": 1.4408,
+      "step": 60290
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4282666666666666e-05,
+      "loss": 1.3322,
+      "step": 60295
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4279703703703703e-05,
+      "loss": 1.5857,
+      "step": 60300
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.427674074074074e-05,
+      "loss": 1.3632,
+      "step": 60305
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4273777777777778e-05,
+      "loss": 1.4963,
+      "step": 60310
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4270814814814815e-05,
+      "loss": 1.5187,
+      "step": 60315
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4267851851851852e-05,
+      "loss": 1.5726,
+      "step": 60320
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.4264888888888889e-05,
+      "loss": 1.4157,
+      "step": 60325
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4261925925925926e-05,
+      "loss": 1.5694,
+      "step": 60330
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4258962962962963e-05,
+      "loss": 1.4503,
+      "step": 60335
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4256e-05,
+      "loss": 1.3732,
+      "step": 60340
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4253037037037038e-05,
+      "loss": 1.571,
+      "step": 60345
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4250074074074075e-05,
+      "loss": 1.496,
+      "step": 60350
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4247111111111112e-05,
+      "loss": 1.4736,
+      "step": 60355
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4244148148148149e-05,
+      "loss": 1.5226,
+      "step": 60360
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4241185185185186e-05,
+      "loss": 1.4895,
+      "step": 60365
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4238222222222221e-05,
+      "loss": 1.5708,
+      "step": 60370
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4235259259259259e-05,
+      "loss": 1.4889,
+      "step": 60375
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 1.4232296296296296e-05,
+      "loss": 1.5046,
+      "step": 60380
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4229333333333333e-05,
+      "loss": 1.5581,
+      "step": 60385
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.422637037037037e-05,
+      "loss": 1.3962,
+      "step": 60390
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4223407407407407e-05,
+      "loss": 1.4948,
+      "step": 60395
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4220444444444444e-05,
+      "loss": 1.4324,
+      "step": 60400
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4217481481481481e-05,
+      "loss": 1.5764,
+      "step": 60405
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4214518518518518e-05,
+      "loss": 1.5,
+      "step": 60410
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4211555555555556e-05,
+      "loss": 1.4142,
+      "step": 60415
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4208592592592593e-05,
+      "loss": 1.4421,
+      "step": 60420
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.420562962962963e-05,
+      "loss": 1.4746,
+      "step": 60425
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4202666666666667e-05,
+      "loss": 1.4443,
+      "step": 60430
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4199703703703704e-05,
+      "loss": 1.4639,
+      "step": 60435
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.4196740740740741e-05,
+      "loss": 1.5154,
+      "step": 60440
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4193777777777778e-05,
+      "loss": 1.4681,
+      "step": 60445
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4190814814814816e-05,
+      "loss": 1.447,
+      "step": 60450
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4187851851851853e-05,
+      "loss": 1.5505,
+      "step": 60455
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.418488888888889e-05,
+      "loss": 1.5202,
+      "step": 60460
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4181925925925927e-05,
+      "loss": 1.5247,
+      "step": 60465
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4178962962962964e-05,
+      "loss": 1.3956,
+      "step": 60470
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4176000000000001e-05,
+      "loss": 1.5164,
+      "step": 60475
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4173037037037038e-05,
+      "loss": 1.5039,
+      "step": 60480
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4170074074074074e-05,
+      "loss": 1.592,
+      "step": 60485
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.416711111111111e-05,
+      "loss": 1.602,
+      "step": 60490
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 1.4164148148148148e-05,
+      "loss": 1.5388,
+      "step": 60495
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4161185185185185e-05,
+      "loss": 1.5217,
+      "step": 60500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4158222222222222e-05,
+      "loss": 1.6331,
+      "step": 60505
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.415525925925926e-05,
+      "loss": 1.3241,
+      "step": 60510
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4152296296296296e-05,
+      "loss": 1.3566,
+      "step": 60515
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4149333333333334e-05,
+      "loss": 1.4882,
+      "step": 60520
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.414637037037037e-05,
+      "loss": 1.4998,
+      "step": 60525
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4143407407407408e-05,
+      "loss": 1.4572,
+      "step": 60530
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4140444444444445e-05,
+      "loss": 1.3891,
+      "step": 60535
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4137481481481482e-05,
+      "loss": 1.402,
+      "step": 60540
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.413451851851852e-05,
+      "loss": 1.4221,
+      "step": 60545
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.4131555555555556e-05,
+      "loss": 1.3952,
+      "step": 60550
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4128592592592594e-05,
+      "loss": 1.5275,
+      "step": 60555
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.412562962962963e-05,
+      "loss": 1.5083,
+      "step": 60560
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4122666666666668e-05,
+      "loss": 1.41,
+      "step": 60565
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4119703703703705e-05,
+      "loss": 1.45,
+      "step": 60570
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4116740740740742e-05,
+      "loss": 1.482,
+      "step": 60575
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4113777777777779e-05,
+      "loss": 1.5258,
+      "step": 60580
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4110814814814816e-05,
+      "loss": 1.4711,
+      "step": 60585
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4107851851851853e-05,
+      "loss": 1.5044,
+      "step": 60590
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.410488888888889e-05,
+      "loss": 1.4309,
+      "step": 60595
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4101925925925926e-05,
+      "loss": 1.5276,
+      "step": 60600
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 1.4098962962962963e-05,
+      "loss": 1.5159,
+      "step": 60605
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4096e-05,
+      "loss": 1.4738,
+      "step": 60610
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4093037037037037e-05,
+      "loss": 1.5394,
+      "step": 60615
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4090074074074074e-05,
+      "loss": 1.4643,
+      "step": 60620
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4087111111111112e-05,
+      "loss": 1.4516,
+      "step": 60625
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4084148148148149e-05,
+      "loss": 1.5585,
+      "step": 60630
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4081185185185186e-05,
+      "loss": 1.3662,
+      "step": 60635
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4078222222222223e-05,
+      "loss": 1.5321,
+      "step": 60640
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.407525925925926e-05,
+      "loss": 1.3711,
+      "step": 60645
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4072296296296297e-05,
+      "loss": 1.642,
+      "step": 60650
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4069333333333334e-05,
+      "loss": 1.3266,
+      "step": 60655
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4066370370370372e-05,
+      "loss": 1.6406,
+      "step": 60660
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 1.4063407407407409e-05,
+      "loss": 1.4431,
+      "step": 60665
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4060444444444446e-05,
+      "loss": 1.4774,
+      "step": 60670
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4057481481481483e-05,
+      "loss": 1.5469,
+      "step": 60675
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.405451851851852e-05,
+      "loss": 1.378,
+      "step": 60680
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4051555555555557e-05,
+      "loss": 1.5359,
+      "step": 60685
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4048592592592594e-05,
+      "loss": 1.475,
+      "step": 60690
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4045629629629631e-05,
+      "loss": 1.4788,
+      "step": 60695
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4042666666666669e-05,
+      "loss": 1.5485,
+      "step": 60700
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4039703703703706e-05,
+      "loss": 1.4579,
+      "step": 60705
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4036740740740743e-05,
+      "loss": 1.395,
+      "step": 60710
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4033777777777778e-05,
+      "loss": 1.4205,
+      "step": 60715
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.4030814814814815e-05,
+      "loss": 1.48,
+      "step": 60720
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4027851851851852e-05,
+      "loss": 1.4741,
+      "step": 60725
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.402488888888889e-05,
+      "loss": 1.4177,
+      "step": 60730
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4021925925925927e-05,
+      "loss": 1.4869,
+      "step": 60735
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4018962962962964e-05,
+      "loss": 1.4335,
+      "step": 60740
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4016000000000001e-05,
+      "loss": 1.4961,
+      "step": 60745
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4013037037037038e-05,
+      "loss": 1.5344,
+      "step": 60750
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4010074074074075e-05,
+      "loss": 1.4651,
+      "step": 60755
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4007111111111112e-05,
+      "loss": 1.5028,
+      "step": 60760
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.400414814814815e-05,
+      "loss": 1.4334,
+      "step": 60765
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.4001185185185187e-05,
+      "loss": 1.4436,
+      "step": 60770
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 1.3998222222222224e-05,
+      "loss": 1.4604,
+      "step": 60775
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3995259259259261e-05,
+      "loss": 1.4823,
+      "step": 60780
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3992296296296298e-05,
+      "loss": 1.5768,
+      "step": 60785
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3989333333333335e-05,
+      "loss": 1.4512,
+      "step": 60790
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3986370370370372e-05,
+      "loss": 1.5093,
+      "step": 60795
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.398340740740741e-05,
+      "loss": 1.5598,
+      "step": 60800
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3980444444444447e-05,
+      "loss": 1.5159,
+      "step": 60805
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3977481481481484e-05,
+      "loss": 1.5411,
+      "step": 60810
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.397451851851852e-05,
+      "loss": 1.4923,
+      "step": 60815
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3971555555555558e-05,
+      "loss": 1.6446,
+      "step": 60820
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.3968592592592595e-05,
+      "loss": 1.5503,
+      "step": 60825
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.396562962962963e-05,
+      "loss": 1.3748,
+      "step": 60830
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3962666666666668e-05,
+      "loss": 1.4955,
+      "step": 60835
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3959703703703705e-05,
+      "loss": 1.506,
+      "step": 60840
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3956740740740742e-05,
+      "loss": 1.4908,
+      "step": 60845
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3953777777777779e-05,
+      "loss": 1.536,
+      "step": 60850
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3950814814814816e-05,
+      "loss": 1.5315,
+      "step": 60855
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3947851851851853e-05,
+      "loss": 1.4839,
+      "step": 60860
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.394488888888889e-05,
+      "loss": 1.5444,
+      "step": 60865
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3941925925925928e-05,
+      "loss": 1.5274,
+      "step": 60870
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3938962962962965e-05,
+      "loss": 1.4289,
+      "step": 60875
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3936000000000002e-05,
+      "loss": 1.5858,
+      "step": 60880
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3933037037037039e-05,
+      "loss": 1.5229,
+      "step": 60885
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 1.3930074074074076e-05,
+      "loss": 1.4989,
+      "step": 60890
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3927111111111113e-05,
+      "loss": 1.5028,
+      "step": 60895
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.392414814814815e-05,
+      "loss": 1.4146,
+      "step": 60900
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3921185185185187e-05,
+      "loss": 1.5247,
+      "step": 60905
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3918814814814815e-05,
+      "loss": 1.4705,
+      "step": 60910
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3915851851851853e-05,
+      "loss": 1.4779,
+      "step": 60915
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.391288888888889e-05,
+      "loss": 1.5891,
+      "step": 60920
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3909925925925927e-05,
+      "loss": 1.4385,
+      "step": 60925
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3906962962962964e-05,
+      "loss": 1.3966,
+      "step": 60930
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3904000000000001e-05,
+      "loss": 1.441,
+      "step": 60935
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3901037037037038e-05,
+      "loss": 1.5257,
+      "step": 60940
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.3898074074074075e-05,
+      "loss": 1.535,
+      "step": 60945
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.389511111111111e-05,
+      "loss": 1.5025,
+      "step": 60950
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3892148148148148e-05,
+      "loss": 1.6283,
+      "step": 60955
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3889185185185185e-05,
+      "loss": 1.4912,
+      "step": 60960
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3886222222222222e-05,
+      "loss": 1.551,
+      "step": 60965
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.388325925925926e-05,
+      "loss": 1.4639,
+      "step": 60970
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3880296296296296e-05,
+      "loss": 1.558,
+      "step": 60975
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3877333333333334e-05,
+      "loss": 1.5029,
+      "step": 60980
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.387437037037037e-05,
+      "loss": 1.5696,
+      "step": 60985
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3871407407407408e-05,
+      "loss": 1.5208,
+      "step": 60990
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3868444444444445e-05,
+      "loss": 1.4529,
+      "step": 60995
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 1.3865481481481482e-05,
+      "loss": 1.477,
+      "step": 61000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3862518518518519e-05,
+      "loss": 1.5563,
+      "step": 61005
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3859555555555556e-05,
+      "loss": 1.4448,
+      "step": 61010
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3856592592592593e-05,
+      "loss": 1.5204,
+      "step": 61015
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.385362962962963e-05,
+      "loss": 1.4065,
+      "step": 61020
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3850666666666668e-05,
+      "loss": 1.4736,
+      "step": 61025
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3847703703703705e-05,
+      "loss": 1.4789,
+      "step": 61030
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3844740740740742e-05,
+      "loss": 1.5453,
+      "step": 61035
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3841777777777779e-05,
+      "loss": 1.4828,
+      "step": 61040
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3838814814814816e-05,
+      "loss": 1.5385,
+      "step": 61045
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.3835851851851853e-05,
+      "loss": 1.4465,
+      "step": 61050
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.383288888888889e-05,
+      "loss": 1.559,
+      "step": 61055
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3829925925925928e-05,
+      "loss": 1.5849,
+      "step": 61060
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3826962962962963e-05,
+      "loss": 1.4584,
+      "step": 61065
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3824e-05,
+      "loss": 1.4701,
+      "step": 61070
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3821037037037037e-05,
+      "loss": 1.4577,
+      "step": 61075
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3818074074074074e-05,
+      "loss": 1.507,
+      "step": 61080
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3815111111111112e-05,
+      "loss": 1.4794,
+      "step": 61085
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3812148148148149e-05,
+      "loss": 1.4863,
+      "step": 61090
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3809185185185186e-05,
+      "loss": 1.4413,
+      "step": 61095
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3806222222222223e-05,
+      "loss": 1.4945,
+      "step": 61100
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.380325925925926e-05,
+      "loss": 1.3642,
+      "step": 61105
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3800296296296297e-05,
+      "loss": 1.4319,
+      "step": 61110
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 1.3797333333333334e-05,
+      "loss": 1.4072,
+      "step": 61115
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3794370370370371e-05,
+      "loss": 1.5702,
+      "step": 61120
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3791407407407409e-05,
+      "loss": 1.4985,
+      "step": 61125
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3788444444444446e-05,
+      "loss": 1.5686,
+      "step": 61130
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3785481481481483e-05,
+      "loss": 1.4985,
+      "step": 61135
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.378251851851852e-05,
+      "loss": 1.5289,
+      "step": 61140
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3779555555555557e-05,
+      "loss": 1.6471,
+      "step": 61145
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3776592592592594e-05,
+      "loss": 1.5163,
+      "step": 61150
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3773629629629631e-05,
+      "loss": 1.5686,
+      "step": 61155
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3770666666666668e-05,
+      "loss": 1.5007,
+      "step": 61160
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3767703703703706e-05,
+      "loss": 1.5711,
+      "step": 61165
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 1.3764740740740743e-05,
+      "loss": 1.4783,
+      "step": 61170
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.376177777777778e-05,
+      "loss": 1.4889,
+      "step": 61175
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3758814814814815e-05,
+      "loss": 1.4393,
+      "step": 61180
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3755851851851852e-05,
+      "loss": 1.4724,
+      "step": 61185
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.375288888888889e-05,
+      "loss": 1.4998,
+      "step": 61190
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3749925925925927e-05,
+      "loss": 1.5213,
+      "step": 61195
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3746962962962964e-05,
+      "loss": 1.4467,
+      "step": 61200
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3744000000000001e-05,
+      "loss": 1.4373,
+      "step": 61205
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3741037037037038e-05,
+      "loss": 1.5972,
+      "step": 61210
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3738074074074075e-05,
+      "loss": 1.5606,
+      "step": 61215
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.3735111111111112e-05,
+      "loss": 1.4841,
+      "step": 61220
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.373214814814815e-05,
+      "loss": 1.3935,
+      "step": 61225
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3729185185185187e-05,
+      "loss": 1.4767,
+      "step": 61230
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3726222222222224e-05,
+      "loss": 1.5002,
+      "step": 61235
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.372325925925926e-05,
+      "loss": 1.4437,
+      "step": 61240
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3720296296296298e-05,
+      "loss": 1.5449,
+      "step": 61245
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3717333333333335e-05,
+      "loss": 1.4299,
+      "step": 61250
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3714370370370372e-05,
+      "loss": 1.4137,
+      "step": 61255
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.371140740740741e-05,
+      "loss": 1.5176,
+      "step": 61260
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3708444444444446e-05,
+      "loss": 1.5481,
+      "step": 61265
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3705481481481484e-05,
+      "loss": 1.4137,
+      "step": 61270
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.370251851851852e-05,
+      "loss": 1.4913,
+      "step": 61275
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 1.3699555555555558e-05,
+      "loss": 1.5252,
+      "step": 61280
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3696592592592595e-05,
+      "loss": 1.4621,
+      "step": 61285
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3693629629629632e-05,
+      "loss": 1.496,
+      "step": 61290
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3690666666666667e-05,
+      "loss": 1.3599,
+      "step": 61295
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3687703703703705e-05,
+      "loss": 1.4028,
+      "step": 61300
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3684740740740742e-05,
+      "loss": 1.4168,
+      "step": 61305
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3681777777777779e-05,
+      "loss": 1.5474,
+      "step": 61310
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3678814814814816e-05,
+      "loss": 1.4723,
+      "step": 61315
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3675851851851853e-05,
+      "loss": 1.4583,
+      "step": 61320
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.367288888888889e-05,
+      "loss": 1.6029,
+      "step": 61325
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3669925925925927e-05,
+      "loss": 1.4726,
+      "step": 61330
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3666962962962965e-05,
+      "loss": 1.4167,
+      "step": 61335
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.3664000000000002e-05,
+      "loss": 1.5281,
+      "step": 61340
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3661037037037039e-05,
+      "loss": 1.3605,
+      "step": 61345
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3658074074074076e-05,
+      "loss": 1.4417,
+      "step": 61350
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3655111111111113e-05,
+      "loss": 1.4952,
+      "step": 61355
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.365214814814815e-05,
+      "loss": 1.5668,
+      "step": 61360
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3649185185185187e-05,
+      "loss": 1.5079,
+      "step": 61365
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3646222222222224e-05,
+      "loss": 1.4569,
+      "step": 61370
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3643259259259262e-05,
+      "loss": 1.4849,
+      "step": 61375
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3640296296296299e-05,
+      "loss": 1.5349,
+      "step": 61380
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3637333333333336e-05,
+      "loss": 1.5154,
+      "step": 61385
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.3634370370370373e-05,
+      "loss": 1.4101,
+      "step": 61390
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 1.363140740740741e-05,
+      "loss": 1.5637,
+      "step": 61395
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3628444444444447e-05,
+      "loss": 1.4237,
+      "step": 61400
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3625481481481484e-05,
+      "loss": 1.4898,
+      "step": 61405
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3622518518518518e-05,
+      "loss": 1.486,
+      "step": 61410
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3619555555555555e-05,
+      "loss": 1.4525,
+      "step": 61415
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3616592592592592e-05,
+      "loss": 1.4803,
+      "step": 61420
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.361362962962963e-05,
+      "loss": 1.5547,
+      "step": 61425
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3610666666666667e-05,
+      "loss": 1.5006,
+      "step": 61430
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3607703703703704e-05,
+      "loss": 1.453,
+      "step": 61435
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.360474074074074e-05,
+      "loss": 1.5266,
+      "step": 61440
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3601777777777778e-05,
+      "loss": 1.4771,
+      "step": 61445
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.3598814814814813e-05,
+      "loss": 1.5533,
+      "step": 61450
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.359585185185185e-05,
+      "loss": 1.5165,
+      "step": 61455
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3592888888888888e-05,
+      "loss": 1.4918,
+      "step": 61460
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3589925925925925e-05,
+      "loss": 1.5114,
+      "step": 61465
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3586962962962962e-05,
+      "loss": 1.5091,
+      "step": 61470
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3583999999999999e-05,
+      "loss": 1.5227,
+      "step": 61475
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3581037037037036e-05,
+      "loss": 1.4602,
+      "step": 61480
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3578074074074073e-05,
+      "loss": 1.4675,
+      "step": 61485
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.357511111111111e-05,
+      "loss": 1.5081,
+      "step": 61490
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3572148148148148e-05,
+      "loss": 1.5707,
+      "step": 61495
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3569185185185185e-05,
+      "loss": 1.5891,
+      "step": 61500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 1.3566222222222222e-05,
+      "loss": 1.5937,
+      "step": 61505
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3563259259259259e-05,
+      "loss": 1.4343,
+      "step": 61510
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3560296296296296e-05,
+      "loss": 1.5453,
+      "step": 61515
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3557333333333333e-05,
+      "loss": 1.4873,
+      "step": 61520
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.355437037037037e-05,
+      "loss": 1.5869,
+      "step": 61525
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3551407407407407e-05,
+      "loss": 1.4988,
+      "step": 61530
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3548444444444445e-05,
+      "loss": 1.5813,
+      "step": 61535
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3545481481481482e-05,
+      "loss": 1.4467,
+      "step": 61540
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3542518518518519e-05,
+      "loss": 1.6053,
+      "step": 61545
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3539555555555556e-05,
+      "loss": 1.509,
+      "step": 61550
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3536592592592593e-05,
+      "loss": 1.473,
+      "step": 61555
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.353362962962963e-05,
+      "loss": 1.3753,
+      "step": 61560
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.3530666666666667e-05,
+      "loss": 1.4439,
+      "step": 61565
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3527703703703703e-05,
+      "loss": 1.4831,
+      "step": 61570
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.352474074074074e-05,
+      "loss": 1.5335,
+      "step": 61575
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3521777777777777e-05,
+      "loss": 1.5248,
+      "step": 61580
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3518814814814814e-05,
+      "loss": 1.5045,
+      "step": 61585
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3515851851851851e-05,
+      "loss": 1.5527,
+      "step": 61590
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3512888888888888e-05,
+      "loss": 1.4771,
+      "step": 61595
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3509925925925926e-05,
+      "loss": 1.5255,
+      "step": 61600
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3506962962962963e-05,
+      "loss": 1.4246,
+      "step": 61605
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3504e-05,
+      "loss": 1.4403,
+      "step": 61610
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3501037037037037e-05,
+      "loss": 1.5184,
+      "step": 61615
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 1.3498074074074074e-05,
+      "loss": 1.4688,
+      "step": 61620
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3495111111111111e-05,
+      "loss": 1.4908,
+      "step": 61625
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3492148148148148e-05,
+      "loss": 1.4648,
+      "step": 61630
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3489185185185185e-05,
+      "loss": 1.4734,
+      "step": 61635
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3486222222222223e-05,
+      "loss": 1.4293,
+      "step": 61640
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.348325925925926e-05,
+      "loss": 1.4557,
+      "step": 61645
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3480296296296297e-05,
+      "loss": 1.5062,
+      "step": 61650
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3477333333333334e-05,
+      "loss": 1.5551,
+      "step": 61655
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3474370370370371e-05,
+      "loss": 1.5217,
+      "step": 61660
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3471407407407408e-05,
+      "loss": 1.5396,
+      "step": 61665
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3468444444444445e-05,
+      "loss": 1.5263,
+      "step": 61670
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 1.3465481481481482e-05,
+      "loss": 1.5292,
+      "step": 61675
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.346251851851852e-05,
+      "loss": 1.4142,
+      "step": 61680
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3459555555555555e-05,
+      "loss": 1.552,
+      "step": 61685
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3456592592592592e-05,
+      "loss": 1.3939,
+      "step": 61690
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.345362962962963e-05,
+      "loss": 1.6079,
+      "step": 61695
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3450666666666666e-05,
+      "loss": 1.4173,
+      "step": 61700
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3447703703703704e-05,
+      "loss": 1.5368,
+      "step": 61705
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.344474074074074e-05,
+      "loss": 1.4957,
+      "step": 61710
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3441777777777778e-05,
+      "loss": 1.4505,
+      "step": 61715
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3438814814814815e-05,
+      "loss": 1.5222,
+      "step": 61720
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.3435851851851852e-05,
+      "loss": 1.5929,
+      "step": 61725
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.343288888888889e-05,
+      "loss": 1.5399,
+      "step": 61730
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3429925925925926e-05,
+      "loss": 1.3692,
+      "step": 61735
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3426962962962963e-05,
+      "loss": 1.4934,
+      "step": 61740
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3424e-05,
+      "loss": 1.6251,
+      "step": 61745
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3421037037037038e-05,
+      "loss": 1.6087,
+      "step": 61750
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3418074074074075e-05,
+      "loss": 1.5132,
+      "step": 61755
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3415111111111112e-05,
+      "loss": 1.4847,
+      "step": 61760
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3412148148148149e-05,
+      "loss": 1.516,
+      "step": 61765
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3409185185185186e-05,
+      "loss": 1.3766,
+      "step": 61770
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3406222222222223e-05,
+      "loss": 1.377,
+      "step": 61775
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.340325925925926e-05,
+      "loss": 1.4542,
+      "step": 61780
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3400296296296298e-05,
+      "loss": 1.6086,
+      "step": 61785
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 1.3397333333333335e-05,
+      "loss": 1.5306,
+      "step": 61790
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3394370370370372e-05,
+      "loss": 1.3323,
+      "step": 61795
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3391407407407407e-05,
+      "loss": 1.3893,
+      "step": 61800
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3388444444444444e-05,
+      "loss": 1.5448,
+      "step": 61805
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3385481481481482e-05,
+      "loss": 1.4792,
+      "step": 61810
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3382518518518519e-05,
+      "loss": 1.4159,
+      "step": 61815
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3379555555555556e-05,
+      "loss": 1.5375,
+      "step": 61820
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3376592592592593e-05,
+      "loss": 1.4492,
+      "step": 61825
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.337362962962963e-05,
+      "loss": 1.4078,
+      "step": 61830
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3370666666666667e-05,
+      "loss": 1.3954,
+      "step": 61835
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3367703703703704e-05,
+      "loss": 1.4495,
+      "step": 61840
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.3364740740740741e-05,
+      "loss": 1.4143,
+      "step": 61845
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3361777777777779e-05,
+      "loss": 1.4637,
+      "step": 61850
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3358814814814816e-05,
+      "loss": 1.4407,
+      "step": 61855
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3355851851851853e-05,
+      "loss": 1.4594,
+      "step": 61860
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.335288888888889e-05,
+      "loss": 1.447,
+      "step": 61865
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3349925925925927e-05,
+      "loss": 1.472,
+      "step": 61870
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3346962962962964e-05,
+      "loss": 1.4909,
+      "step": 61875
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3344000000000001e-05,
+      "loss": 1.4133,
+      "step": 61880
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3341037037037038e-05,
+      "loss": 1.4068,
+      "step": 61885
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3338074074074076e-05,
+      "loss": 1.5331,
+      "step": 61890
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.3335111111111113e-05,
+      "loss": 1.3475,
+      "step": 61895
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 1.333214814814815e-05,
+      "loss": 1.3808,
+      "step": 61900
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3329185185185187e-05,
+      "loss": 1.4218,
+      "step": 61905
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3326222222222224e-05,
+      "loss": 1.3575,
+      "step": 61910
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.332325925925926e-05,
+      "loss": 1.3837,
+      "step": 61915
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3320296296296297e-05,
+      "loss": 1.449,
+      "step": 61920
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3317333333333334e-05,
+      "loss": 1.4727,
+      "step": 61925
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3314370370370371e-05,
+      "loss": 1.4079,
+      "step": 61930
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3311407407407408e-05,
+      "loss": 1.4416,
+      "step": 61935
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3308444444444445e-05,
+      "loss": 1.4915,
+      "step": 61940
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3305481481481482e-05,
+      "loss": 1.4474,
+      "step": 61945
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.330251851851852e-05,
+      "loss": 1.3913,
+      "step": 61950
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.3299555555555557e-05,
+      "loss": 1.4863,
+      "step": 61955
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3296592592592594e-05,
+      "loss": 1.2624,
+      "step": 61960
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.329362962962963e-05,
+      "loss": 1.462,
+      "step": 61965
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3290666666666668e-05,
+      "loss": 1.3583,
+      "step": 61970
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3287703703703705e-05,
+      "loss": 1.3833,
+      "step": 61975
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3284740740740742e-05,
+      "loss": 1.4646,
+      "step": 61980
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.328177777777778e-05,
+      "loss": 1.4493,
+      "step": 61985
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3278814814814816e-05,
+      "loss": 1.3662,
+      "step": 61990
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3275851851851854e-05,
+      "loss": 1.4863,
+      "step": 61995
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.327288888888889e-05,
+      "loss": 1.3113,
+      "step": 62000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3269925925925928e-05,
+      "loss": 1.4307,
+      "step": 62005
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3266962962962965e-05,
+      "loss": 1.4471,
+      "step": 62010
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 1.3264000000000002e-05,
+      "loss": 1.48,
+      "step": 62015
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.326103703703704e-05,
+      "loss": 1.4442,
+      "step": 62020
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3258074074074076e-05,
+      "loss": 1.4229,
+      "step": 62025
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3255111111111112e-05,
+      "loss": 1.3587,
+      "step": 62030
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3252148148148149e-05,
+      "loss": 1.3338,
+      "step": 62035
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3249185185185186e-05,
+      "loss": 1.4004,
+      "step": 62040
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3246222222222223e-05,
+      "loss": 1.4567,
+      "step": 62045
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.324325925925926e-05,
+      "loss": 1.3829,
+      "step": 62050
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3240296296296297e-05,
+      "loss": 1.4231,
+      "step": 62055
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3237333333333335e-05,
+      "loss": 1.3665,
+      "step": 62060
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3234370370370372e-05,
+      "loss": 1.4033,
+      "step": 62065
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 1.3231407407407409e-05,
+      "loss": 1.3873,
+      "step": 62070
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3228444444444446e-05,
+      "loss": 1.4874,
+      "step": 62075
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3225481481481483e-05,
+      "loss": 1.4459,
+      "step": 62080
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.322251851851852e-05,
+      "loss": 1.4785,
+      "step": 62085
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3219555555555557e-05,
+      "loss": 1.4257,
+      "step": 62090
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3216592592592594e-05,
+      "loss": 1.3861,
+      "step": 62095
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3213629629629632e-05,
+      "loss": 1.4343,
+      "step": 62100
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3210666666666669e-05,
+      "loss": 1.4426,
+      "step": 62105
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3207703703703706e-05,
+      "loss": 1.493,
+      "step": 62110
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3204740740740743e-05,
+      "loss": 1.4734,
+      "step": 62115
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.320177777777778e-05,
+      "loss": 1.3785,
+      "step": 62120
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.3198814814814817e-05,
+      "loss": 1.3405,
+      "step": 62125
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3195851851851854e-05,
+      "loss": 1.4027,
+      "step": 62130
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3192888888888891e-05,
+      "loss": 1.3828,
+      "step": 62135
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3189925925925929e-05,
+      "loss": 1.4075,
+      "step": 62140
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3186962962962964e-05,
+      "loss": 1.3446,
+      "step": 62145
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3184000000000001e-05,
+      "loss": 1.3724,
+      "step": 62150
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3181037037037038e-05,
+      "loss": 1.4064,
+      "step": 62155
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3178074074074075e-05,
+      "loss": 1.4636,
+      "step": 62160
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3175111111111113e-05,
+      "loss": 1.3908,
+      "step": 62165
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.317214814814815e-05,
+      "loss": 1.4878,
+      "step": 62170
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3169185185185187e-05,
+      "loss": 1.4843,
+      "step": 62175
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 1.3166222222222224e-05,
+      "loss": 1.378,
+      "step": 62180
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3163259259259258e-05,
+      "loss": 1.4966,
+      "step": 62185
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3160296296296295e-05,
+      "loss": 1.4666,
+      "step": 62190
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3157333333333332e-05,
+      "loss": 1.4291,
+      "step": 62195
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3154370370370369e-05,
+      "loss": 1.5085,
+      "step": 62200
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3151407407407406e-05,
+      "loss": 1.3034,
+      "step": 62205
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3148444444444443e-05,
+      "loss": 1.387,
+      "step": 62210
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.314548148148148e-05,
+      "loss": 1.4746,
+      "step": 62215
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3142518518518518e-05,
+      "loss": 1.3871,
+      "step": 62220
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3139555555555555e-05,
+      "loss": 1.619,
+      "step": 62225
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3136592592592592e-05,
+      "loss": 1.4775,
+      "step": 62230
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3133629629629629e-05,
+      "loss": 1.4864,
+      "step": 62235
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.3130666666666666e-05,
+      "loss": 1.4858,
+      "step": 62240
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3127703703703703e-05,
+      "loss": 1.4496,
+      "step": 62245
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.312474074074074e-05,
+      "loss": 1.566,
+      "step": 62250
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3121777777777777e-05,
+      "loss": 1.365,
+      "step": 62255
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3118814814814815e-05,
+      "loss": 1.411,
+      "step": 62260
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3115851851851852e-05,
+      "loss": 1.4052,
+      "step": 62265
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3112888888888889e-05,
+      "loss": 1.5939,
+      "step": 62270
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3109925925925926e-05,
+      "loss": 1.3822,
+      "step": 62275
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3106962962962963e-05,
+      "loss": 1.4119,
+      "step": 62280
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3104e-05,
+      "loss": 1.4517,
+      "step": 62285
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3101037037037037e-05,
+      "loss": 1.3412,
+      "step": 62290
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 1.3098074074074074e-05,
+      "loss": 1.5226,
+      "step": 62295
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3095111111111112e-05,
+      "loss": 1.4079,
+      "step": 62300
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3092148148148147e-05,
+      "loss": 1.4322,
+      "step": 62305
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3089185185185184e-05,
+      "loss": 1.3837,
+      "step": 62310
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3086222222222221e-05,
+      "loss": 1.5039,
+      "step": 62315
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3083259259259258e-05,
+      "loss": 1.3157,
+      "step": 62320
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3080296296296296e-05,
+      "loss": 1.4661,
+      "step": 62325
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3077333333333333e-05,
+      "loss": 1.2867,
+      "step": 62330
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.307437037037037e-05,
+      "loss": 1.3055,
+      "step": 62335
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3071407407407407e-05,
+      "loss": 1.4608,
+      "step": 62340
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3068444444444444e-05,
+      "loss": 1.489,
+      "step": 62345
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.3065481481481481e-05,
+      "loss": 1.4414,
+      "step": 62350
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3062518518518518e-05,
+      "loss": 1.5302,
+      "step": 62355
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3059555555555555e-05,
+      "loss": 1.4479,
+      "step": 62360
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3056592592592593e-05,
+      "loss": 1.486,
+      "step": 62365
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.305362962962963e-05,
+      "loss": 1.4216,
+      "step": 62370
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3050666666666667e-05,
+      "loss": 1.4287,
+      "step": 62375
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3047703703703704e-05,
+      "loss": 1.3774,
+      "step": 62380
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3044740740740741e-05,
+      "loss": 1.4408,
+      "step": 62385
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3041777777777778e-05,
+      "loss": 1.4273,
+      "step": 62390
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3038814814814815e-05,
+      "loss": 1.4712,
+      "step": 62395
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.3035851851851852e-05,
+      "loss": 1.4498,
+      "step": 62400
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 1.303288888888889e-05,
+      "loss": 1.4075,
+      "step": 62405
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3029925925925927e-05,
+      "loss": 1.4966,
+      "step": 62410
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3026962962962964e-05,
+      "loss": 1.3855,
+      "step": 62415
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3024e-05,
+      "loss": 1.3437,
+      "step": 62420
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3021037037037036e-05,
+      "loss": 1.4148,
+      "step": 62425
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3018074074074074e-05,
+      "loss": 1.538,
+      "step": 62430
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.301511111111111e-05,
+      "loss": 1.369,
+      "step": 62435
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3012148148148148e-05,
+      "loss": 1.5188,
+      "step": 62440
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3009185185185185e-05,
+      "loss": 1.36,
+      "step": 62445
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3006222222222222e-05,
+      "loss": 1.449,
+      "step": 62450
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.300325925925926e-05,
+      "loss": 1.4689,
+      "step": 62455
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.3000296296296296e-05,
+      "loss": 1.3568,
+      "step": 62460
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.2997333333333333e-05,
+      "loss": 1.4439,
+      "step": 62465
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.299437037037037e-05,
+      "loss": 1.4756,
+      "step": 62470
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2991407407407408e-05,
+      "loss": 1.5335,
+      "step": 62475
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2988444444444445e-05,
+      "loss": 1.4442,
+      "step": 62480
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2985481481481482e-05,
+      "loss": 1.3576,
+      "step": 62485
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2982518518518519e-05,
+      "loss": 1.3434,
+      "step": 62490
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2979555555555556e-05,
+      "loss": 1.3303,
+      "step": 62495
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2976592592592593e-05,
+      "loss": 1.3664,
+      "step": 62500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.297362962962963e-05,
+      "loss": 1.352,
+      "step": 62505
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2970666666666668e-05,
+      "loss": 1.454,
+      "step": 62510
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2967703703703705e-05,
+      "loss": 1.3895,
+      "step": 62515
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 1.2964740740740742e-05,
+      "loss": 1.4973,
+      "step": 62520
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2961777777777779e-05,
+      "loss": 1.4923,
+      "step": 62525
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2958814814814816e-05,
+      "loss": 1.4156,
+      "step": 62530
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2955851851851852e-05,
+      "loss": 1.3343,
+      "step": 62535
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2952888888888889e-05,
+      "loss": 1.4482,
+      "step": 62540
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2949925925925926e-05,
+      "loss": 1.4333,
+      "step": 62545
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2946962962962963e-05,
+      "loss": 1.4253,
+      "step": 62550
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2944e-05,
+      "loss": 1.4877,
+      "step": 62555
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2941037037037037e-05,
+      "loss": 1.4449,
+      "step": 62560
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2938074074074074e-05,
+      "loss": 1.4745,
+      "step": 62565
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2935111111111111e-05,
+      "loss": 1.4347,
+      "step": 62570
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 1.2932148148148149e-05,
+      "loss": 1.4227,
+      "step": 62575
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2929185185185186e-05,
+      "loss": 1.4772,
+      "step": 62580
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2926222222222223e-05,
+      "loss": 1.3895,
+      "step": 62585
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.292325925925926e-05,
+      "loss": 1.464,
+      "step": 62590
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2920296296296297e-05,
+      "loss": 1.3754,
+      "step": 62595
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2917333333333334e-05,
+      "loss": 1.4082,
+      "step": 62600
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2914370370370371e-05,
+      "loss": 1.5368,
+      "step": 62605
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2911407407407408e-05,
+      "loss": 1.6053,
+      "step": 62610
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2908444444444446e-05,
+      "loss": 1.445,
+      "step": 62615
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2905481481481483e-05,
+      "loss": 1.3734,
+      "step": 62620
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.290251851851852e-05,
+      "loss": 1.4722,
+      "step": 62625
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.2899555555555557e-05,
+      "loss": 1.423,
+      "step": 62630
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2896592592592594e-05,
+      "loss": 1.3289,
+      "step": 62635
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2893629629629631e-05,
+      "loss": 1.4246,
+      "step": 62640
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2890666666666668e-05,
+      "loss": 1.4601,
+      "step": 62645
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2887703703703704e-05,
+      "loss": 1.4235,
+      "step": 62650
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2884740740740741e-05,
+      "loss": 1.4778,
+      "step": 62655
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2881777777777778e-05,
+      "loss": 1.5063,
+      "step": 62660
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2878814814814815e-05,
+      "loss": 1.4612,
+      "step": 62665
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2875851851851852e-05,
+      "loss": 1.4322,
+      "step": 62670
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.287288888888889e-05,
+      "loss": 1.3647,
+      "step": 62675
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2869925925925927e-05,
+      "loss": 1.4895,
+      "step": 62680
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2866962962962964e-05,
+      "loss": 1.4091,
+      "step": 62685
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 1.2864e-05,
+      "loss": 1.5034,
+      "step": 62690
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2861037037037038e-05,
+      "loss": 1.5016,
+      "step": 62695
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2858074074074075e-05,
+      "loss": 1.406,
+      "step": 62700
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2855111111111112e-05,
+      "loss": 1.4634,
+      "step": 62705
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.285214814814815e-05,
+      "loss": 1.4158,
+      "step": 62710
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2849185185185186e-05,
+      "loss": 1.5406,
+      "step": 62715
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2846222222222224e-05,
+      "loss": 1.5597,
+      "step": 62720
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.284325925925926e-05,
+      "loss": 1.418,
+      "step": 62725
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2840888888888889e-05,
+      "loss": 1.3906,
+      "step": 62730
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2837925925925926e-05,
+      "loss": 1.4865,
+      "step": 62735
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2834962962962963e-05,
+      "loss": 1.5078,
+      "step": 62740
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.2832e-05,
+      "loss": 1.4061,
+      "step": 62745
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2829037037037037e-05,
+      "loss": 1.441,
+      "step": 62750
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2826074074074074e-05,
+      "loss": 1.4335,
+      "step": 62755
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2823111111111112e-05,
+      "loss": 1.4624,
+      "step": 62760
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2820148148148149e-05,
+      "loss": 1.4073,
+      "step": 62765
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2817185185185184e-05,
+      "loss": 1.508,
+      "step": 62770
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2814222222222221e-05,
+      "loss": 1.3509,
+      "step": 62775
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2811259259259258e-05,
+      "loss": 1.4573,
+      "step": 62780
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2808296296296295e-05,
+      "loss": 1.3846,
+      "step": 62785
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2805333333333333e-05,
+      "loss": 1.4614,
+      "step": 62790
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.280237037037037e-05,
+      "loss": 1.4623,
+      "step": 62795
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 1.2799407407407407e-05,
+      "loss": 1.508,
+      "step": 62800
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2796444444444444e-05,
+      "loss": 1.4451,
+      "step": 62805
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2793481481481481e-05,
+      "loss": 1.4005,
+      "step": 62810
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2790518518518518e-05,
+      "loss": 1.3819,
+      "step": 62815
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2787555555555555e-05,
+      "loss": 1.4465,
+      "step": 62820
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2784592592592592e-05,
+      "loss": 1.4123,
+      "step": 62825
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.278162962962963e-05,
+      "loss": 1.3381,
+      "step": 62830
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2778666666666667e-05,
+      "loss": 1.3782,
+      "step": 62835
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2775703703703704e-05,
+      "loss": 1.4416,
+      "step": 62840
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2772740740740741e-05,
+      "loss": 1.4405,
+      "step": 62845
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2769777777777778e-05,
+      "loss": 1.3594,
+      "step": 62850
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.2766814814814815e-05,
+      "loss": 1.4442,
+      "step": 62855
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2763851851851852e-05,
+      "loss": 1.4712,
+      "step": 62860
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.276088888888889e-05,
+      "loss": 1.4451,
+      "step": 62865
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2757925925925927e-05,
+      "loss": 1.4091,
+      "step": 62870
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2754962962962964e-05,
+      "loss": 1.3955,
+      "step": 62875
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2752000000000001e-05,
+      "loss": 1.4398,
+      "step": 62880
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2749037037037036e-05,
+      "loss": 1.4119,
+      "step": 62885
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2746074074074073e-05,
+      "loss": 1.2501,
+      "step": 62890
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.274311111111111e-05,
+      "loss": 1.3734,
+      "step": 62895
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2740148148148148e-05,
+      "loss": 1.5163,
+      "step": 62900
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2737185185185185e-05,
+      "loss": 1.4326,
+      "step": 62905
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2734222222222222e-05,
+      "loss": 1.4178,
+      "step": 62910
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 1.2731259259259259e-05,
+      "loss": 1.3932,
+      "step": 62915
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2728296296296296e-05,
+      "loss": 1.4501,
+      "step": 62920
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2725333333333333e-05,
+      "loss": 1.4281,
+      "step": 62925
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.272237037037037e-05,
+      "loss": 1.4741,
+      "step": 62930
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2719407407407408e-05,
+      "loss": 1.3972,
+      "step": 62935
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2716444444444445e-05,
+      "loss": 1.424,
+      "step": 62940
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2713481481481482e-05,
+      "loss": 1.3897,
+      "step": 62945
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2710518518518519e-05,
+      "loss": 1.3829,
+      "step": 62950
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2707555555555556e-05,
+      "loss": 1.4174,
+      "step": 62955
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2704592592592593e-05,
+      "loss": 1.4653,
+      "step": 62960
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.270162962962963e-05,
+      "loss": 1.3815,
+      "step": 62965
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.2698666666666668e-05,
+      "loss": 1.3248,
+      "step": 62970
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2695703703703705e-05,
+      "loss": 1.4613,
+      "step": 62975
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2692740740740742e-05,
+      "loss": 1.459,
+      "step": 62980
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2689777777777779e-05,
+      "loss": 1.462,
+      "step": 62985
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2686814814814816e-05,
+      "loss": 1.378,
+      "step": 62990
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2683851851851853e-05,
+      "loss": 1.5159,
+      "step": 62995
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2680888888888889e-05,
+      "loss": 1.3656,
+      "step": 63000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2677925925925926e-05,
+      "loss": 1.3683,
+      "step": 63005
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2674962962962963e-05,
+      "loss": 1.4099,
+      "step": 63010
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2672e-05,
+      "loss": 1.4831,
+      "step": 63015
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2669037037037037e-05,
+      "loss": 1.5326,
+      "step": 63020
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 1.2666074074074074e-05,
+      "loss": 1.45,
+      "step": 63025
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2663111111111111e-05,
+      "loss": 1.438,
+      "step": 63030
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2660148148148148e-05,
+      "loss": 1.4529,
+      "step": 63035
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2657185185185186e-05,
+      "loss": 1.3854,
+      "step": 63040
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2654222222222223e-05,
+      "loss": 1.4134,
+      "step": 63045
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.265125925925926e-05,
+      "loss": 1.444,
+      "step": 63050
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2648296296296297e-05,
+      "loss": 1.4758,
+      "step": 63055
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2645333333333334e-05,
+      "loss": 1.4061,
+      "step": 63060
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2642370370370371e-05,
+      "loss": 1.5346,
+      "step": 63065
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2639407407407408e-05,
+      "loss": 1.448,
+      "step": 63070
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2636444444444446e-05,
+      "loss": 1.3911,
+      "step": 63075
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 1.2633481481481483e-05,
+      "loss": 1.5641,
+      "step": 63080
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.263051851851852e-05,
+      "loss": 1.4784,
+      "step": 63085
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2627555555555557e-05,
+      "loss": 1.477,
+      "step": 63090
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2624592592592594e-05,
+      "loss": 1.302,
+      "step": 63095
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2621629629629631e-05,
+      "loss": 1.4063,
+      "step": 63100
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2618666666666668e-05,
+      "loss": 1.4619,
+      "step": 63105
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2615703703703705e-05,
+      "loss": 1.3234,
+      "step": 63110
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.261274074074074e-05,
+      "loss": 1.3497,
+      "step": 63115
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2609777777777778e-05,
+      "loss": 1.5233,
+      "step": 63120
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2606814814814815e-05,
+      "loss": 1.3763,
+      "step": 63125
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2603851851851852e-05,
+      "loss": 1.5741,
+      "step": 63130
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.260088888888889e-05,
+      "loss": 1.4273,
+      "step": 63135
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.2597925925925926e-05,
+      "loss": 1.4434,
+      "step": 63140
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2594962962962964e-05,
+      "loss": 1.4281,
+      "step": 63145
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2592e-05,
+      "loss": 1.6038,
+      "step": 63150
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2589037037037038e-05,
+      "loss": 1.5052,
+      "step": 63155
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2586074074074075e-05,
+      "loss": 1.3966,
+      "step": 63160
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2583111111111112e-05,
+      "loss": 1.2816,
+      "step": 63165
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.258014814814815e-05,
+      "loss": 1.4384,
+      "step": 63170
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2577185185185186e-05,
+      "loss": 1.4818,
+      "step": 63175
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2574222222222224e-05,
+      "loss": 1.391,
+      "step": 63180
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.257125925925926e-05,
+      "loss": 1.3863,
+      "step": 63185
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2568296296296298e-05,
+      "loss": 1.5218,
+      "step": 63190
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 1.2565333333333335e-05,
+      "loss": 1.5048,
+      "step": 63195
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2562370370370372e-05,
+      "loss": 1.4771,
+      "step": 63200
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2559407407407409e-05,
+      "loss": 1.3952,
+      "step": 63205
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2556444444444446e-05,
+      "loss": 1.2982,
+      "step": 63210
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2553481481481483e-05,
+      "loss": 1.248,
+      "step": 63215
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.255051851851852e-05,
+      "loss": 1.4309,
+      "step": 63220
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2547555555555558e-05,
+      "loss": 1.4947,
+      "step": 63225
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2544592592592593e-05,
+      "loss": 1.4467,
+      "step": 63230
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.254162962962963e-05,
+      "loss": 1.4155,
+      "step": 63235
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2538666666666667e-05,
+      "loss": 1.5096,
+      "step": 63240
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2535703703703704e-05,
+      "loss": 1.4144,
+      "step": 63245
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.2532740740740742e-05,
+      "loss": 1.3855,
+      "step": 63250
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2529777777777779e-05,
+      "loss": 1.5263,
+      "step": 63255
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2526814814814816e-05,
+      "loss": 1.3757,
+      "step": 63260
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2523851851851853e-05,
+      "loss": 1.4024,
+      "step": 63265
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.252088888888889e-05,
+      "loss": 1.4251,
+      "step": 63270
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2517925925925927e-05,
+      "loss": 1.4224,
+      "step": 63275
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2514962962962964e-05,
+      "loss": 1.4568,
+      "step": 63280
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2512000000000002e-05,
+      "loss": 1.4631,
+      "step": 63285
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2509037037037039e-05,
+      "loss": 1.37,
+      "step": 63290
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2506074074074076e-05,
+      "loss": 1.5101,
+      "step": 63295
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.2503111111111113e-05,
+      "loss": 1.494,
+      "step": 63300
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 1.250014814814815e-05,
+      "loss": 1.4636,
+      "step": 63305
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2497185185185185e-05,
+      "loss": 1.4333,
+      "step": 63310
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2494222222222223e-05,
+      "loss": 1.2915,
+      "step": 63315
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.249125925925926e-05,
+      "loss": 1.5589,
+      "step": 63320
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2488296296296297e-05,
+      "loss": 1.4958,
+      "step": 63325
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2485333333333334e-05,
+      "loss": 1.3333,
+      "step": 63330
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2482370370370371e-05,
+      "loss": 1.3119,
+      "step": 63335
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2479407407407408e-05,
+      "loss": 1.4723,
+      "step": 63340
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2476444444444445e-05,
+      "loss": 1.5131,
+      "step": 63345
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2473481481481482e-05,
+      "loss": 1.5058,
+      "step": 63350
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.247051851851852e-05,
+      "loss": 1.4424,
+      "step": 63355
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2467555555555557e-05,
+      "loss": 1.4529,
+      "step": 63360
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.2464592592592592e-05,
+      "loss": 1.4008,
+      "step": 63365
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.246162962962963e-05,
+      "loss": 1.4442,
+      "step": 63370
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2458666666666666e-05,
+      "loss": 1.4131,
+      "step": 63375
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2455703703703704e-05,
+      "loss": 1.4425,
+      "step": 63380
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.245274074074074e-05,
+      "loss": 1.4593,
+      "step": 63385
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2449777777777778e-05,
+      "loss": 1.3284,
+      "step": 63390
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2446814814814815e-05,
+      "loss": 1.3679,
+      "step": 63395
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2443851851851852e-05,
+      "loss": 1.449,
+      "step": 63400
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.244088888888889e-05,
+      "loss": 1.3136,
+      "step": 63405
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2437925925925926e-05,
+      "loss": 1.5921,
+      "step": 63410
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2434962962962963e-05,
+      "loss": 1.3995,
+      "step": 63415
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 1.2432e-05,
+      "loss": 1.5115,
+      "step": 63420
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2429037037037038e-05,
+      "loss": 1.3916,
+      "step": 63425
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2426074074074075e-05,
+      "loss": 1.4975,
+      "step": 63430
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2423111111111112e-05,
+      "loss": 1.4304,
+      "step": 63435
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2420148148148149e-05,
+      "loss": 1.3127,
+      "step": 63440
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2417185185185186e-05,
+      "loss": 1.3067,
+      "step": 63445
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2414222222222223e-05,
+      "loss": 1.4566,
+      "step": 63450
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.241125925925926e-05,
+      "loss": 1.3178,
+      "step": 63455
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2408296296296298e-05,
+      "loss": 1.4906,
+      "step": 63460
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2405333333333335e-05,
+      "loss": 1.3916,
+      "step": 63465
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2402370370370372e-05,
+      "loss": 1.4068,
+      "step": 63470
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.2399407407407409e-05,
+      "loss": 1.4206,
+      "step": 63475
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2396444444444444e-05,
+      "loss": 1.4474,
+      "step": 63480
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2393481481481482e-05,
+      "loss": 1.4051,
+      "step": 63485
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2390518518518519e-05,
+      "loss": 1.3416,
+      "step": 63490
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2387555555555556e-05,
+      "loss": 1.3681,
+      "step": 63495
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2384592592592593e-05,
+      "loss": 1.4256,
+      "step": 63500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.238162962962963e-05,
+      "loss": 1.4887,
+      "step": 63505
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2378666666666667e-05,
+      "loss": 1.4836,
+      "step": 63510
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2375703703703704e-05,
+      "loss": 1.4317,
+      "step": 63515
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2372740740740741e-05,
+      "loss": 1.5093,
+      "step": 63520
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2369777777777779e-05,
+      "loss": 1.4812,
+      "step": 63525
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 1.2366814814814816e-05,
+      "loss": 1.4008,
+      "step": 63530
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2363851851851853e-05,
+      "loss": 1.4007,
+      "step": 63535
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.236088888888889e-05,
+      "loss": 1.4426,
+      "step": 63540
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2357925925925927e-05,
+      "loss": 1.4207,
+      "step": 63545
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2354962962962964e-05,
+      "loss": 1.388,
+      "step": 63550
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2352000000000001e-05,
+      "loss": 1.4497,
+      "step": 63555
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2349037037037038e-05,
+      "loss": 1.3574,
+      "step": 63560
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2346074074074076e-05,
+      "loss": 1.5136,
+      "step": 63565
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2343111111111113e-05,
+      "loss": 1.396,
+      "step": 63570
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.234014814814815e-05,
+      "loss": 1.4039,
+      "step": 63575
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2337185185185187e-05,
+      "loss": 1.4086,
+      "step": 63580
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2334222222222224e-05,
+      "loss": 1.3543,
+      "step": 63585
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 1.2331259259259261e-05,
+      "loss": 1.3999,
+      "step": 63590
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2328296296296297e-05,
+      "loss": 1.4088,
+      "step": 63595
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2325333333333334e-05,
+      "loss": 1.3915,
+      "step": 63600
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2322370370370371e-05,
+      "loss": 1.4533,
+      "step": 63605
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2319407407407408e-05,
+      "loss": 1.3593,
+      "step": 63610
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2316444444444445e-05,
+      "loss": 1.4548,
+      "step": 63615
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.231348148148148e-05,
+      "loss": 1.4336,
+      "step": 63620
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2310518518518518e-05,
+      "loss": 1.5669,
+      "step": 63625
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2307555555555555e-05,
+      "loss": 1.4884,
+      "step": 63630
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2304592592592592e-05,
+      "loss": 1.51,
+      "step": 63635
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2301629629629629e-05,
+      "loss": 1.4248,
+      "step": 63640
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.2298666666666666e-05,
+      "loss": 1.3863,
+      "step": 63645
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2295703703703703e-05,
+      "loss": 1.4853,
+      "step": 63650
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.229274074074074e-05,
+      "loss": 1.6204,
+      "step": 63655
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2289777777777778e-05,
+      "loss": 1.3527,
+      "step": 63660
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2286814814814815e-05,
+      "loss": 1.4132,
+      "step": 63665
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2283851851851852e-05,
+      "loss": 1.4646,
+      "step": 63670
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2280888888888889e-05,
+      "loss": 1.5045,
+      "step": 63675
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2277925925925926e-05,
+      "loss": 1.5071,
+      "step": 63680
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2274962962962963e-05,
+      "loss": 1.3902,
+      "step": 63685
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2272e-05,
+      "loss": 1.4731,
+      "step": 63690
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2269037037037038e-05,
+      "loss": 1.4404,
+      "step": 63695
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 1.2266074074074075e-05,
+      "loss": 1.5084,
+      "step": 63700
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2263111111111112e-05,
+      "loss": 1.4721,
+      "step": 63705
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2260148148148149e-05,
+      "loss": 1.2777,
+      "step": 63710
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2257185185185186e-05,
+      "loss": 1.4879,
+      "step": 63715
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2254222222222223e-05,
+      "loss": 1.4895,
+      "step": 63720
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.225125925925926e-05,
+      "loss": 1.4538,
+      "step": 63725
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2248296296296297e-05,
+      "loss": 1.4,
+      "step": 63730
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2245333333333333e-05,
+      "loss": 1.4021,
+      "step": 63735
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.224237037037037e-05,
+      "loss": 1.4455,
+      "step": 63740
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2239407407407407e-05,
+      "loss": 1.4138,
+      "step": 63745
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2236444444444444e-05,
+      "loss": 1.4656,
+      "step": 63750
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.2233481481481481e-05,
+      "loss": 1.3558,
+      "step": 63755
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2230518518518519e-05,
+      "loss": 1.409,
+      "step": 63760
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2227555555555556e-05,
+      "loss": 1.395,
+      "step": 63765
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2224592592592593e-05,
+      "loss": 1.3842,
+      "step": 63770
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.222162962962963e-05,
+      "loss": 1.4286,
+      "step": 63775
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2218666666666667e-05,
+      "loss": 1.4097,
+      "step": 63780
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2215703703703704e-05,
+      "loss": 1.3628,
+      "step": 63785
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2212740740740741e-05,
+      "loss": 1.4684,
+      "step": 63790
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2209777777777778e-05,
+      "loss": 1.4058,
+      "step": 63795
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2206814814814816e-05,
+      "loss": 1.4203,
+      "step": 63800
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2203851851851853e-05,
+      "loss": 1.5195,
+      "step": 63805
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.220088888888889e-05,
+      "loss": 1.4547,
+      "step": 63810
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 1.2197925925925927e-05,
+      "loss": 1.4276,
+      "step": 63815
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2194962962962964e-05,
+      "loss": 1.3875,
+      "step": 63820
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2192000000000001e-05,
+      "loss": 1.476,
+      "step": 63825
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2189037037037038e-05,
+      "loss": 1.4638,
+      "step": 63830
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2186074074074075e-05,
+      "loss": 1.3933,
+      "step": 63835
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2183111111111113e-05,
+      "loss": 1.3857,
+      "step": 63840
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.218014814814815e-05,
+      "loss": 1.5262,
+      "step": 63845
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2177185185185185e-05,
+      "loss": 1.4174,
+      "step": 63850
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2174222222222222e-05,
+      "loss": 1.4584,
+      "step": 63855
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.217125925925926e-05,
+      "loss": 1.5363,
+      "step": 63860
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2168296296296297e-05,
+      "loss": 1.537,
+      "step": 63865
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.2165333333333334e-05,
+      "loss": 1.4268,
+      "step": 63870
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.216237037037037e-05,
+      "loss": 1.4448,
+      "step": 63875
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2159407407407408e-05,
+      "loss": 1.4206,
+      "step": 63880
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2156444444444445e-05,
+      "loss": 1.3317,
+      "step": 63885
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2153481481481482e-05,
+      "loss": 1.5224,
+      "step": 63890
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.215051851851852e-05,
+      "loss": 1.4148,
+      "step": 63895
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2147555555555556e-05,
+      "loss": 1.4214,
+      "step": 63900
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2144592592592594e-05,
+      "loss": 1.3718,
+      "step": 63905
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.214162962962963e-05,
+      "loss": 1.3296,
+      "step": 63910
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2138666666666668e-05,
+      "loss": 1.447,
+      "step": 63915
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2135703703703705e-05,
+      "loss": 1.399,
+      "step": 63920
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 1.2132740740740742e-05,
+      "loss": 1.3159,
+      "step": 63925
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.212977777777778e-05,
+      "loss": 1.4368,
+      "step": 63930
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2126814814814816e-05,
+      "loss": 1.5324,
+      "step": 63935
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2123851851851853e-05,
+      "loss": 1.5732,
+      "step": 63940
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.212088888888889e-05,
+      "loss": 1.4191,
+      "step": 63945
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2117925925925928e-05,
+      "loss": 1.4841,
+      "step": 63950
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2114962962962965e-05,
+      "loss": 1.4358,
+      "step": 63955
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2112000000000002e-05,
+      "loss": 1.3871,
+      "step": 63960
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2109037037037037e-05,
+      "loss": 1.4708,
+      "step": 63965
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2106074074074075e-05,
+      "loss": 1.4338,
+      "step": 63970
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2103111111111112e-05,
+      "loss": 1.6058,
+      "step": 63975
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.2100148148148149e-05,
+      "loss": 1.4204,
+      "step": 63980
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2097185185185186e-05,
+      "loss": 1.3751,
+      "step": 63985
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2094222222222223e-05,
+      "loss": 1.4953,
+      "step": 63990
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.209125925925926e-05,
+      "loss": 1.5018,
+      "step": 63995
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2088296296296297e-05,
+      "loss": 1.4578,
+      "step": 64000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2085333333333333e-05,
+      "loss": 1.3368,
+      "step": 64005
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.208237037037037e-05,
+      "loss": 1.4855,
+      "step": 64010
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2079407407407407e-05,
+      "loss": 1.5567,
+      "step": 64015
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2076444444444444e-05,
+      "loss": 1.4216,
+      "step": 64020
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2073481481481481e-05,
+      "loss": 1.4041,
+      "step": 64025
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2070518518518518e-05,
+      "loss": 1.501,
+      "step": 64030
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2067555555555555e-05,
+      "loss": 1.3653,
+      "step": 64035
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 1.2064592592592593e-05,
+      "loss": 1.441,
+      "step": 64040
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.206162962962963e-05,
+      "loss": 1.3726,
+      "step": 64045
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2058666666666667e-05,
+      "loss": 1.4443,
+      "step": 64050
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2055703703703704e-05,
+      "loss": 1.2689,
+      "step": 64055
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2052740740740741e-05,
+      "loss": 1.3833,
+      "step": 64060
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2049777777777778e-05,
+      "loss": 1.4577,
+      "step": 64065
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2046814814814815e-05,
+      "loss": 1.4814,
+      "step": 64070
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2043851851851853e-05,
+      "loss": 1.4675,
+      "step": 64075
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.204088888888889e-05,
+      "loss": 1.5369,
+      "step": 64080
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2037925925925927e-05,
+      "loss": 1.4042,
+      "step": 64085
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2034962962962964e-05,
+      "loss": 1.4081,
+      "step": 64090
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 1.2032000000000001e-05,
+      "loss": 1.4858,
+      "step": 64095
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2029037037037036e-05,
+      "loss": 1.3878,
+      "step": 64100
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2026074074074074e-05,
+      "loss": 1.4497,
+      "step": 64105
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.202311111111111e-05,
+      "loss": 1.2881,
+      "step": 64110
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2020148148148148e-05,
+      "loss": 1.472,
+      "step": 64115
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2017185185185185e-05,
+      "loss": 1.409,
+      "step": 64120
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2014222222222222e-05,
+      "loss": 1.4294,
+      "step": 64125
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.201125925925926e-05,
+      "loss": 1.4415,
+      "step": 64130
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2008296296296296e-05,
+      "loss": 1.3761,
+      "step": 64135
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.2005333333333333e-05,
+      "loss": 1.4763,
+      "step": 64140
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.200237037037037e-05,
+      "loss": 1.3813,
+      "step": 64145
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.1999407407407408e-05,
+      "loss": 1.4211,
+      "step": 64150
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1996444444444445e-05,
+      "loss": 1.4975,
+      "step": 64155
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1993481481481482e-05,
+      "loss": 1.3922,
+      "step": 64160
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1990518518518519e-05,
+      "loss": 1.4213,
+      "step": 64165
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1987555555555556e-05,
+      "loss": 1.4342,
+      "step": 64170
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1984592592592593e-05,
+      "loss": 1.5983,
+      "step": 64175
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.198162962962963e-05,
+      "loss": 1.383,
+      "step": 64180
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1978666666666668e-05,
+      "loss": 1.4656,
+      "step": 64185
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1975703703703705e-05,
+      "loss": 1.4681,
+      "step": 64190
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1972740740740742e-05,
+      "loss": 1.5319,
+      "step": 64195
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1969777777777779e-05,
+      "loss": 1.3237,
+      "step": 64200
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 1.1966814814814816e-05,
+      "loss": 1.478,
+      "step": 64205
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1963851851851853e-05,
+      "loss": 1.4051,
+      "step": 64210
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1960888888888889e-05,
+      "loss": 1.5332,
+      "step": 64215
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1957925925925926e-05,
+      "loss": 1.4351,
+      "step": 64220
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1954962962962963e-05,
+      "loss": 1.4731,
+      "step": 64225
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1952e-05,
+      "loss": 1.5974,
+      "step": 64230
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1949037037037037e-05,
+      "loss": 1.3687,
+      "step": 64235
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1946074074074074e-05,
+      "loss": 1.382,
+      "step": 64240
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1943111111111111e-05,
+      "loss": 1.3731,
+      "step": 64245
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1940148148148149e-05,
+      "loss": 1.4651,
+      "step": 64250
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1937185185185186e-05,
+      "loss": 1.3834,
+      "step": 64255
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.1934222222222223e-05,
+      "loss": 1.4097,
+      "step": 64260
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.193125925925926e-05,
+      "loss": 1.5593,
+      "step": 64265
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1928296296296297e-05,
+      "loss": 1.4307,
+      "step": 64270
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1925333333333334e-05,
+      "loss": 1.4859,
+      "step": 64275
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1922370370370371e-05,
+      "loss": 1.4241,
+      "step": 64280
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1919407407407408e-05,
+      "loss": 1.4335,
+      "step": 64285
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1916444444444446e-05,
+      "loss": 1.5566,
+      "step": 64290
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1913481481481483e-05,
+      "loss": 1.5053,
+      "step": 64295
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.191051851851852e-05,
+      "loss": 1.5723,
+      "step": 64300
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1907555555555557e-05,
+      "loss": 1.4238,
+      "step": 64305
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1904592592592594e-05,
+      "loss": 1.3484,
+      "step": 64310
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1901629629629631e-05,
+      "loss": 1.5124,
+      "step": 64315
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 1.1898666666666668e-05,
+      "loss": 1.4343,
+      "step": 64320
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1895703703703706e-05,
+      "loss": 1.335,
+      "step": 64325
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1892740740740741e-05,
+      "loss": 1.4694,
+      "step": 64330
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1889777777777778e-05,
+      "loss": 1.5566,
+      "step": 64335
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1886814814814815e-05,
+      "loss": 1.5253,
+      "step": 64340
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1883851851851852e-05,
+      "loss": 1.4767,
+      "step": 64345
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.188088888888889e-05,
+      "loss": 1.4058,
+      "step": 64350
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1877925925925927e-05,
+      "loss": 1.5159,
+      "step": 64355
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1874962962962964e-05,
+      "loss": 1.5364,
+      "step": 64360
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1872000000000001e-05,
+      "loss": 1.6023,
+      "step": 64365
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1869037037037038e-05,
+      "loss": 1.3556,
+      "step": 64370
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.1866074074074075e-05,
+      "loss": 1.4972,
+      "step": 64375
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1863111111111112e-05,
+      "loss": 1.5108,
+      "step": 64380
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.186014814814815e-05,
+      "loss": 1.363,
+      "step": 64385
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1857185185185185e-05,
+      "loss": 1.4665,
+      "step": 64390
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1854222222222222e-05,
+      "loss": 1.4564,
+      "step": 64395
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1851259259259259e-05,
+      "loss": 1.4022,
+      "step": 64400
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1848296296296296e-05,
+      "loss": 1.5894,
+      "step": 64405
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1845333333333333e-05,
+      "loss": 1.4424,
+      "step": 64410
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.184237037037037e-05,
+      "loss": 1.4059,
+      "step": 64415
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1839407407407408e-05,
+      "loss": 1.559,
+      "step": 64420
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1836444444444445e-05,
+      "loss": 1.5001,
+      "step": 64425
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 1.1833481481481482e-05,
+      "loss": 1.4889,
+      "step": 64430
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1830518518518519e-05,
+      "loss": 1.4373,
+      "step": 64435
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1827555555555556e-05,
+      "loss": 1.4314,
+      "step": 64440
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1824592592592593e-05,
+      "loss": 1.4094,
+      "step": 64445
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.182162962962963e-05,
+      "loss": 1.4187,
+      "step": 64450
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1818666666666667e-05,
+      "loss": 1.4395,
+      "step": 64455
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1815703703703705e-05,
+      "loss": 1.4671,
+      "step": 64460
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1812740740740742e-05,
+      "loss": 1.4288,
+      "step": 64465
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1809777777777777e-05,
+      "loss": 1.4545,
+      "step": 64470
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1806814814814814e-05,
+      "loss": 1.3199,
+      "step": 64475
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1803851851851851e-05,
+      "loss": 1.4931,
+      "step": 64480
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1800888888888889e-05,
+      "loss": 1.4519,
+      "step": 64485
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.1797925925925926e-05,
+      "loss": 1.4743,
+      "step": 64490
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1794962962962963e-05,
+      "loss": 1.4663,
+      "step": 64495
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1792e-05,
+      "loss": 1.3287,
+      "step": 64500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1789037037037037e-05,
+      "loss": 1.3607,
+      "step": 64505
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1786074074074074e-05,
+      "loss": 1.5085,
+      "step": 64510
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1783111111111111e-05,
+      "loss": 1.4652,
+      "step": 64515
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1780148148148148e-05,
+      "loss": 1.4134,
+      "step": 64520
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1777185185185186e-05,
+      "loss": 1.4206,
+      "step": 64525
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1774222222222223e-05,
+      "loss": 1.3942,
+      "step": 64530
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.177125925925926e-05,
+      "loss": 1.4854,
+      "step": 64535
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1768296296296297e-05,
+      "loss": 1.4309,
+      "step": 64540
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 1.1765333333333334e-05,
+      "loss": 1.5314,
+      "step": 64545
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1762370370370371e-05,
+      "loss": 1.3848,
+      "step": 64550
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1759407407407408e-05,
+      "loss": 1.3669,
+      "step": 64555
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1756444444444445e-05,
+      "loss": 1.517,
+      "step": 64560
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1753481481481483e-05,
+      "loss": 1.5267,
+      "step": 64565
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.175051851851852e-05,
+      "loss": 1.4814,
+      "step": 64570
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1747555555555557e-05,
+      "loss": 1.3463,
+      "step": 64575
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1744592592592594e-05,
+      "loss": 1.4455,
+      "step": 64580
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.174162962962963e-05,
+      "loss": 1.5548,
+      "step": 64585
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1738666666666667e-05,
+      "loss": 1.5267,
+      "step": 64590
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.1735703703703704e-05,
+      "loss": 1.3188,
+      "step": 64595
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 1.173274074074074e-05,
+      "loss": 1.5194,
+      "step": 64600
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1729777777777778e-05,
+      "loss": 1.374,
+      "step": 64605
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1726814814814815e-05,
+      "loss": 1.3236,
+      "step": 64610
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1723851851851852e-05,
+      "loss": 1.4847,
+      "step": 64615
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.172088888888889e-05,
+      "loss": 1.4281,
+      "step": 64620
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1717925925925926e-05,
+      "loss": 1.5842,
+      "step": 64625
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1714962962962964e-05,
+      "loss": 1.4239,
+      "step": 64630
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1712e-05,
+      "loss": 1.499,
+      "step": 64635
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1709037037037038e-05,
+      "loss": 1.3439,
+      "step": 64640
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1706074074074075e-05,
+      "loss": 1.3417,
+      "step": 64645
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.1703111111111112e-05,
+      "loss": 1.4448,
+      "step": 64650
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.170014814814815e-05,
+      "loss": 1.4739,
+      "step": 64655
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1697185185185186e-05,
+      "loss": 1.4692,
+      "step": 64660
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1694222222222223e-05,
+      "loss": 1.4785,
+      "step": 64665
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.169125925925926e-05,
+      "loss": 1.507,
+      "step": 64670
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1688296296296298e-05,
+      "loss": 1.4331,
+      "step": 64675
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1685333333333335e-05,
+      "loss": 1.3896,
+      "step": 64680
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1682370370370372e-05,
+      "loss": 1.4109,
+      "step": 64685
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1679407407407409e-05,
+      "loss": 1.4181,
+      "step": 64690
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1676444444444446e-05,
+      "loss": 1.4497,
+      "step": 64695
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1673481481481482e-05,
+      "loss": 1.434,
+      "step": 64700
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1670518518518519e-05,
+      "loss": 1.4931,
+      "step": 64705
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1667555555555556e-05,
+      "loss": 1.5909,
+      "step": 64710
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 1.1664592592592593e-05,
+      "loss": 1.3753,
+      "step": 64715
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.166162962962963e-05,
+      "loss": 1.343,
+      "step": 64720
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1658666666666667e-05,
+      "loss": 1.4428,
+      "step": 64725
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1655703703703704e-05,
+      "loss": 1.447,
+      "step": 64730
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1652740740740742e-05,
+      "loss": 1.3478,
+      "step": 64735
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1649777777777779e-05,
+      "loss": 1.4299,
+      "step": 64740
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1646814814814816e-05,
+      "loss": 1.5469,
+      "step": 64745
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1643851851851853e-05,
+      "loss": 1.4951,
+      "step": 64750
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.164088888888889e-05,
+      "loss": 1.4089,
+      "step": 64755
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1637925925925927e-05,
+      "loss": 1.4509,
+      "step": 64760
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1634962962962964e-05,
+      "loss": 1.4477,
+      "step": 64765
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.1632000000000001e-05,
+      "loss": 1.3557,
+      "step": 64770
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1629037037037039e-05,
+      "loss": 1.3025,
+      "step": 64775
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1626074074074074e-05,
+      "loss": 1.4078,
+      "step": 64780
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1623111111111111e-05,
+      "loss": 1.4943,
+      "step": 64785
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1620148148148148e-05,
+      "loss": 1.2919,
+      "step": 64790
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1617185185185185e-05,
+      "loss": 1.5148,
+      "step": 64795
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1614222222222223e-05,
+      "loss": 1.444,
+      "step": 64800
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.161125925925926e-05,
+      "loss": 1.384,
+      "step": 64805
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1608296296296297e-05,
+      "loss": 1.5051,
+      "step": 64810
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1605333333333334e-05,
+      "loss": 1.4891,
+      "step": 64815
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1602370370370371e-05,
+      "loss": 1.4748,
+      "step": 64820
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 1.1599407407407408e-05,
+      "loss": 1.4126,
+      "step": 64825
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.1596444444444445e-05,
+      "loss": 1.3885,
+      "step": 64830
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.159348148148148e-05,
+      "loss": 1.4666,
+      "step": 64835
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.1590518518518518e-05,
+      "loss": 1.5625,
+      "step": 64840
+    }
+  ],
+  "max_steps": 84375,
+  "num_train_epochs": 15,
+  "total_flos": 2.801300503831511e+18,
+  "trial_name": null,
+  "trial_params": null
+}