diff --git "a/checkpoint-21431/trainer_state.json" "b/checkpoint-21431/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-21431/trainer_state.json"
@@ -0,0 +1,25742 @@
+{
+  "best_metric": 0.05617095157504082,
+  "best_model_checkpoint": "wav2vec2-base-pem123-32-960h-la/checkpoint-21431",
+  "epoch": 1.0,
+  "global_step": 21431,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.5681,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.2419,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.4492,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.6105,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.2818,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.3872,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.4249,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.4935,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.5878,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.3044,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1e-06,
+      "loss": 2.1617,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.8407,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3e-06,
+      "loss": 2.6913,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.1341,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 2.6384,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.3875,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.6405,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.6826,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.6435,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.9612,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 1.6984,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 1.6943,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 1.9857,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.0135,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 1.9529,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 2.1156,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 1.8468,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 1.9046,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-06,
+      "loss": 2.0311,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 2.0449,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 1.3589,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 1.2662,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 1.1506,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.3545,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 1.4116,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 1.402,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-06,
+      "loss": 1.5328,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 1.475,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 1.6633,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.138,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-06,
+      "loss": 1.2597,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 1.0349,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3e-06,
+      "loss": 1.0808,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4e-06,
+      "loss": 1.044,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 1.2246,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 1.112,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7e-06,
+      "loss": 1.3231,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 1.5087,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.5727,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.7501,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1e-06,
+      "loss": 1.1905,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-06,
+      "loss": 0.9608,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 1.1615,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 1.0021,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.9685,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 1.2904,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7e-06,
+      "loss": 1.2705,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8e-06,
+      "loss": 1.4223,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9e-06,
+      "loss": 1.5338,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 1.7456,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1e-06,
+      "loss": 1.1443,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 0.8596,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 0.9641,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.9228,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.0683,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 0.9611,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 1.2087,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 1.1336,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.9e-06,
+      "loss": 1.422,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-06,
+      "loss": 1.8331,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 1.0522,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.8421,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3e-06,
+      "loss": 0.7651,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.4e-06,
+      "loss": 0.9904,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8832,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 0.9663,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.7e-06,
+      "loss": 1.115,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 1.1533,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.9e-06,
+      "loss": 1.2886,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.0128,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.1e-06,
+      "loss": 0.9795,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.2e-06,
+      "loss": 0.7312,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.3e-06,
+      "loss": 0.7858,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.8727,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.5e-06,
+      "loss": 0.8088,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.6e-06,
+      "loss": 0.9188,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 1.1715,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.8e-06,
+      "loss": 1.3011,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 1.3686,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9e-06,
+      "loss": 1.5265,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 0.8523,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 0.6656,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.3e-06,
+      "loss": 0.8027,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.4e-06,
+      "loss": 0.6912,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.5e-06,
+      "loss": 0.8886,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.8875,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.7e-06,
+      "loss": 0.9732,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.051,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9e-06,
+      "loss": 1.1972,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.5824,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 0.9922,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02e-05,
+      "loss": 0.6785,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0300000000000001e-05,
+      "loss": 0.6366,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.04e-05,
+      "loss": 0.7643,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.8867,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0600000000000002e-05,
+      "loss": 0.7378,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 0.8769,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0800000000000002e-05,
+      "loss": 1.0011,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0900000000000002e-05,
+      "loss": 1.1308,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 1.6817,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1100000000000002e-05,
+      "loss": 0.8498,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.6052,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.13e-05,
+      "loss": 0.6782,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.14e-05,
+      "loss": 0.6536,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.15e-05,
+      "loss": 0.7006,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.16e-05,
+      "loss": 0.9329,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.17e-05,
+      "loss": 0.7805,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.18e-05,
+      "loss": 0.872,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1900000000000001e-05,
+      "loss": 0.8867,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3013,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 0.9019,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.22e-05,
+      "loss": 0.5602,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.23e-05,
+      "loss": 0.632,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2400000000000002e-05,
+      "loss": 0.6756,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7106,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.7885,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.27e-05,
+      "loss": 0.8643,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.953,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2900000000000002e-05,
+      "loss": 1.2641,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 1.4198,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.82,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3200000000000002e-05,
+      "loss": 0.6065,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 0.5797,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.7074,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.5688,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.6267,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3700000000000003e-05,
+      "loss": 0.9009,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.38e-05,
+      "loss": 0.86,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.39e-05,
+      "loss": 0.9638,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4e-05,
+      "loss": 1.6872,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.41e-05,
+      "loss": 0.7373,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4200000000000001e-05,
+      "loss": 0.5019,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.43e-05,
+      "loss": 0.6072,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.6512,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.45e-05,
+      "loss": 0.6652,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.46e-05,
+      "loss": 0.7577,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4700000000000002e-05,
+      "loss": 1.0377,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.48e-05,
+      "loss": 0.8039,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4900000000000001e-05,
+      "loss": 0.9909,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.2967,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5100000000000001e-05,
+      "loss": 0.7075,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.535,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5300000000000003e-05,
+      "loss": 0.5943,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.54e-05,
+      "loss": 0.7975,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.55e-05,
+      "loss": 0.606,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5600000000000003e-05,
+      "loss": 0.748,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 0.714,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.58e-05,
+      "loss": 0.9922,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5900000000000004e-05,
+      "loss": 0.9972,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.2826,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.5911,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.62e-05,
+      "loss": 0.5172,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.63e-05,
+      "loss": 0.455,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.64e-05,
+      "loss": 0.6032,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.65e-05,
+      "loss": 0.6292,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.66e-05,
+      "loss": 0.658,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.67e-05,
+      "loss": 0.7158,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.8783,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.69e-05,
+      "loss": 0.9988,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7e-05,
+      "loss": 1.1596,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.7145,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.72e-05,
+      "loss": 0.4501,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.73e-05,
+      "loss": 0.5174,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7400000000000003e-05,
+      "loss": 0.5951,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.7152,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.76e-05,
+      "loss": 0.6649,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.77e-05,
+      "loss": 0.6626,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7800000000000002e-05,
+      "loss": 0.7571,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.79e-05,
+      "loss": 1.1449,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8e-05,
+      "loss": 1.3764,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8100000000000003e-05,
+      "loss": 0.7015,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.4139,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.83e-05,
+      "loss": 0.5709,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.4689,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.5459,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.86e-05,
+      "loss": 0.6105,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8700000000000004e-05,
+      "loss": 0.7417,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.88e-05,
+      "loss": 0.7044,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 0.8603,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9e-05,
+      "loss": 1.2372,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.91e-05,
+      "loss": 0.6774,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.3458,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.93e-05,
+      "loss": 0.4924,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.94e-05,
+      "loss": 0.4699,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.95e-05,
+      "loss": 0.5591,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.7706,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.97e-05,
+      "loss": 0.6164,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.98e-05,
+      "loss": 0.6035,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 0.8199,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2e-05,
+      "loss": 1.3823,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999953119872486e-05,
+      "loss": 0.6102,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999906239744972e-05,
+      "loss": 0.3826,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999859359617458e-05,
+      "loss": 0.4675,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999812479489944e-05,
+      "loss": 0.4567,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9997655993624305e-05,
+      "loss": 0.6292,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9997187192349165e-05,
+      "loss": 0.6357,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9996718391074025e-05,
+      "loss": 0.666,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9996249589798888e-05,
+      "loss": 0.6685,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9995780788523748e-05,
+      "loss": 0.8939,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9995311987248608e-05,
+      "loss": 1.0384,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994843185973468e-05,
+      "loss": 0.4313,
+      "step": 1055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9994374384698328e-05,
+      "loss": 0.3909,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9993905583423188e-05,
+      "loss": 0.4668,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999343678214805e-05,
+      "loss": 0.4648,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999296798087291e-05,
+      "loss": 0.5502,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999249917959777e-05,
+      "loss": 0.5675,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999203037832263e-05,
+      "loss": 0.7048,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999156157704749e-05,
+      "loss": 0.7512,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999109277577235e-05,
+      "loss": 0.91,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999062397449721e-05,
+      "loss": 1.4923,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9990155173222074e-05,
+      "loss": 0.593,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989686371946934e-05,
+      "loss": 0.412,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989217570671794e-05,
+      "loss": 0.4295,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988748769396657e-05,
+      "loss": 0.5324,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988279968121517e-05,
+      "loss": 0.5146,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987811166846377e-05,
+      "loss": 0.6231,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987342365571237e-05,
+      "loss": 0.6736,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986873564296097e-05,
+      "loss": 0.7737,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986404763020957e-05,
+      "loss": 0.796,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985935961745817e-05,
+      "loss": 1.3116,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985467160470677e-05,
+      "loss": 0.5599,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984998359195536e-05,
+      "loss": 0.3027,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99845295579204e-05,
+      "loss": 0.4646,
+      "step": 1165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998406075664526e-05,
+      "loss": 0.5406,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998359195537012e-05,
+      "loss": 0.5461,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998312315409498e-05,
+      "loss": 0.6237,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9982654352819843e-05,
+      "loss": 0.6277,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9982185551544703e-05,
+      "loss": 0.6304,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981716750269563e-05,
+      "loss": 0.8886,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981247948994423e-05,
+      "loss": 1.1062,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980779147719283e-05,
+      "loss": 0.597,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980310346444146e-05,
+      "loss": 0.3313,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979841545169006e-05,
+      "loss": 0.3761,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9979372743893866e-05,
+      "loss": 0.4524,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978903942618726e-05,
+      "loss": 0.5068,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9978435141343586e-05,
+      "loss": 0.5899,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9977966340068446e-05,
+      "loss": 0.5973,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9977497538793306e-05,
+      "loss": 0.7986,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9977028737518165e-05,
+      "loss": 0.749,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997655993624303e-05,
+      "loss": 1.074,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997609113496789e-05,
+      "loss": 0.511,
+      "step": 1255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997562233369275e-05,
+      "loss": 0.3073,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9975153532417612e-05,
+      "loss": 0.3776,
+      "step": 1265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974684731142472e-05,
+      "loss": 0.458,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974215929867332e-05,
+      "loss": 0.4236,
+      "step": 1275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973747128592192e-05,
+      "loss": 0.53,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973278327317052e-05,
+      "loss": 0.5216,
+      "step": 1285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997280952604191e-05,
+      "loss": 0.8459,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997234072476677e-05,
+      "loss": 1.0025,
+      "step": 1295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997187192349163e-05,
+      "loss": 1.0659,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971403122216495e-05,
+      "loss": 0.5928,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970934320941355e-05,
+      "loss": 0.2835,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9970465519666215e-05,
+      "loss": 0.4103,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969996718391075e-05,
+      "loss": 0.424,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969527917115935e-05,
+      "loss": 0.505,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969059115840798e-05,
+      "loss": 0.5272,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968590314565658e-05,
+      "loss": 0.5859,
+      "step": 1335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968121513290518e-05,
+      "loss": 0.7147,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967652712015378e-05,
+      "loss": 0.6935,
+      "step": 1345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996718391074024e-05,
+      "loss": 1.0448,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99667151094651e-05,
+      "loss": 0.5598,
+      "step": 1355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996624630818996e-05,
+      "loss": 0.2887,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996577750691482e-05,
+      "loss": 0.3524,
+      "step": 1365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996530870563968e-05,
+      "loss": 0.3648,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996483990436454e-05,
+      "loss": 0.5021,
+      "step": 1375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99643711030894e-05,
+      "loss": 0.5777,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996390230181426e-05,
+      "loss": 0.5879,
+      "step": 1385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963433500539124e-05,
+      "loss": 0.6235,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962964699263984e-05,
+      "loss": 0.5928,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962495897988844e-05,
+      "loss": 1.172,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9962027096713704e-05,
+      "loss": 0.5087,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961558295438567e-05,
+      "loss": 0.3315,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9961089494163427e-05,
+      "loss": 0.405,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9960620692888287e-05,
+      "loss": 0.407,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9960151891613147e-05,
+      "loss": 0.5505,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9959683090338007e-05,
+      "loss": 0.5691,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9959214289062867e-05,
+      "loss": 0.502,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9958745487787727e-05,
+      "loss": 0.8324,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995827668651259e-05,
+      "loss": 0.6685,
+      "step": 1445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995780788523745e-05,
+      "loss": 1.0913,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995733908396231e-05,
+      "loss": 0.4889,
+      "step": 1455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995687028268717e-05,
+      "loss": 0.3823,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.995640148141203e-05,
+      "loss": 0.4376,
+      "step": 1465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9955932680136893e-05,
+      "loss": 0.4689,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9955463878861753e-05,
+      "loss": 0.4048,
+      "step": 1475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9954995077586613e-05,
+      "loss": 0.5842,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9954526276311473e-05,
+      "loss": 0.5643,
+      "step": 1485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9954057475036336e-05,
+      "loss": 0.8457,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9953588673761196e-05,
+      "loss": 0.5716,
+      "step": 1495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9953119872486056e-05,
+      "loss": 1.1056,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9952651071210916e-05,
+      "loss": 0.4777,
+      "step": 1505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9952182269935776e-05,
+      "loss": 0.277,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9951713468660636e-05,
+      "loss": 0.3154,
+      "step": 1515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9951244667385496e-05,
+      "loss": 0.4537,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9950775866110356e-05,
+      "loss": 0.517,
+      "step": 1525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9950307064835216e-05,
+      "loss": 0.5247,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994983826356008e-05,
+      "loss": 0.4861,
+      "step": 1535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994936946228494e-05,
+      "loss": 0.5592,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99489006610098e-05,
+      "loss": 0.9036,
+      "step": 1545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9948431859734662e-05,
+      "loss": 1.0261,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9947963058459522e-05,
+      "loss": 0.5694,
+      "step": 1555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9947494257184382e-05,
+      "loss": 0.2699,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9947025455909242e-05,
+      "loss": 0.3443,
+      "step": 1565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9946556654634102e-05,
+      "loss": 0.563,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9946087853358962e-05,
+      "loss": 0.3772,
+      "step": 1575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9945619052083822e-05,
+      "loss": 0.5209,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9945150250808685e-05,
+      "loss": 0.4973,
+      "step": 1585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9944681449533545e-05,
+      "loss": 0.4882,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9944212648258405e-05,
+      "loss": 0.7922,
+      "step": 1595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943743846983265e-05,
+      "loss": 0.7772,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943275045708125e-05,
+      "loss": 0.555,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9942806244432985e-05,
+      "loss": 0.4321,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9942337443157848e-05,
+      "loss": 0.4047,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9941868641882708e-05,
+      "loss": 0.352,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9941399840607568e-05,
+      "loss": 0.3804,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994093103933243e-05,
+      "loss": 0.4649,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.994046223805729e-05,
+      "loss": 0.5527,
+      "step": 1635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993999343678215e-05,
+      "loss": 0.7,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993952463550701e-05,
+      "loss": 0.7865,
+      "step": 1645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993905583423187e-05,
+      "loss": 1.0613,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993858703295673e-05,
+      "loss": 0.533,
+      "step": 1655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993811823168159e-05,
+      "loss": 0.2746,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993764943040645e-05,
+      "loss": 0.3082,
+      "step": 1665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993718062913131e-05,
+      "loss": 0.4,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9936711827856174e-05,
+      "loss": 0.4761,
+      "step": 1675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9936243026581034e-05,
+      "loss": 0.4022,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9935774225305894e-05,
+      "loss": 0.6181,
+      "step": 1685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9935305424030754e-05,
+      "loss": 0.6173,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9934836622755617e-05,
+      "loss": 0.6797,
+      "step": 1695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9934367821480477e-05,
+      "loss": 1.1921,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9933899020205337e-05,
+      "loss": 0.5224,
+      "step": 1705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9933430218930197e-05,
+      "loss": 0.3617,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9932961417655057e-05,
+      "loss": 0.3178,
+      "step": 1715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993249261637992e-05,
+      "loss": 0.4298,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993202381510478e-05,
+      "loss": 0.4292,
+      "step": 1725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993155501382964e-05,
+      "loss": 0.4192,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99310862125545e-05,
+      "loss": 0.5256,
+      "step": 1735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993061741127936e-05,
+      "loss": 0.5433,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.993014861000422e-05,
+      "loss": 0.7419,
+      "step": 1745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992967980872908e-05,
+      "loss": 1.1098,
+      "step": 1750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992921100745394e-05,
+      "loss": 0.4264,
+      "step": 1755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9928742206178803e-05,
+      "loss": 0.2677,
+      "step": 1760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9928273404903663e-05,
+      "loss": 0.3569,
+      "step": 1765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9927804603628523e-05,
+      "loss": 0.3524,
+      "step": 1770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9927335802353386e-05,
+      "loss": 0.469,
+      "step": 1775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9926867001078246e-05,
+      "loss": 0.3399,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9926398199803106e-05,
+      "loss": 0.583,
+      "step": 1785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9925929398527966e-05,
+      "loss": 0.5959,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9925460597252826e-05,
+      "loss": 0.7883,
+      "step": 1795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9924991795977686e-05,
+      "loss": 0.8624,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9924522994702546e-05,
+      "loss": 0.4495,
+      "step": 1805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9924054193427406e-05,
+      "loss": 0.2398,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992358539215227e-05,
+      "loss": 0.285,
+      "step": 1815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.992311659087713e-05,
+      "loss": 0.3657,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992264778960199e-05,
+      "loss": 0.4166,
+      "step": 1825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992217898832685e-05,
+      "loss": 0.4724,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.992171018705171e-05,
+      "loss": 0.5657,
+      "step": 1835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9921241385776572e-05,
+      "loss": 0.4912,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9920772584501432e-05,
+      "loss": 0.8323,
+      "step": 1845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9920303783226292e-05,
+      "loss": 1.2386,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9919834981951152e-05,
+      "loss": 0.481,
+      "step": 1855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9919366180676015e-05,
+      "loss": 0.2647,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9918897379400875e-05,
+      "loss": 0.3251,
+      "step": 1865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9918428578125735e-05,
+      "loss": 0.4379,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9917959776850595e-05,
+      "loss": 0.3494,
+      "step": 1875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9917490975575455e-05,
+      "loss": 0.4675,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9917022174300315e-05,
+      "loss": 0.5453,
+      "step": 1885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9916553373025175e-05,
+      "loss": 0.7557,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9916084571750035e-05,
+      "loss": 0.8384,
+      "step": 1895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9915615770474898e-05,
+      "loss": 0.9389,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9915146969199758e-05,
+      "loss": 0.595,
+      "step": 1905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9914678167924618e-05,
+      "loss": 0.2589,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9914209366649478e-05,
+      "loss": 0.2753,
+      "step": 1915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991374056537434e-05,
+      "loss": 0.327,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99132717640992e-05,
+      "loss": 0.4068,
+      "step": 1925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991280296282406e-05,
+      "loss": 0.3982,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991233416154892e-05,
+      "loss": 0.4792,
+      "step": 1935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991186536027378e-05,
+      "loss": 0.5322,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.991139655899864e-05,
+      "loss": 0.5524,
+      "step": 1945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99109277577235e-05,
+      "loss": 0.7718,
+      "step": 1950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9910458956448364e-05,
+      "loss": 0.4615,
+      "step": 1955
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9909990155173224e-05,
+      "loss": 0.23,
+      "step": 1960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9909521353898084e-05,
+      "loss": 0.3236,
+      "step": 1965
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9909052552622944e-05,
+      "loss": 0.3613,
+      "step": 1970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9908583751347804e-05,
+      "loss": 0.4707,
+      "step": 1975
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9908114950072667e-05,
+      "loss": 0.4895,
+      "step": 1980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9907646148797527e-05,
+      "loss": 0.4998,
+      "step": 1985
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9907177347522387e-05,
+      "loss": 0.4744,
+      "step": 1990
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9906708546247247e-05,
+      "loss": 0.6763,
+      "step": 1995
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990623974497211e-05,
+      "loss": 0.8183,
+      "step": 2000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990577094369697e-05,
+      "loss": 0.4009,
+      "step": 2005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990530214242183e-05,
+      "loss": 0.2849,
+      "step": 2010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990483334114669e-05,
+      "loss": 0.3281,
+      "step": 2015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990436453987155e-05,
+      "loss": 0.2439,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990389573859641e-05,
+      "loss": 0.4259,
+      "step": 2025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990342693732127e-05,
+      "loss": 0.4708,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.990295813604613e-05,
+      "loss": 0.4431,
+      "step": 2035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.990248933477099e-05,
+      "loss": 0.6112,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9902020533495853e-05,
+      "loss": 0.7258,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9901551732220713e-05,
+      "loss": 1.1696,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9901082930945573e-05,
+      "loss": 0.4031,
+      "step": 2055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9900614129670436e-05,
+      "loss": 0.2141,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9900145328395296e-05,
+      "loss": 0.2085,
+      "step": 2065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9899676527120156e-05,
+      "loss": 0.3854,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9899207725845016e-05,
+      "loss": 0.447,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9898738924569876e-05,
+      "loss": 0.4033,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9898270123294736e-05,
+      "loss": 0.4619,
+      "step": 2085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9897801322019596e-05,
+      "loss": 0.4808,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989733252074446e-05,
+      "loss": 0.7021,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989686371946932e-05,
+      "loss": 1.0386,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989639491819418e-05,
+      "loss": 0.4082,
+      "step": 2105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989592611691904e-05,
+      "loss": 0.2731,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.98954573156439e-05,
+      "loss": 0.3498,
+      "step": 2115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989498851436876e-05,
+      "loss": 0.3212,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9894519713093622e-05,
+      "loss": 0.3869,
+      "step": 2125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9894050911818482e-05,
+      "loss": 0.3923,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9893582110543342e-05,
+      "loss": 0.4625,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9893113309268205e-05,
+      "loss": 0.5782,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9892644507993065e-05,
+      "loss": 0.845,
+      "step": 2145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9892175706717925e-05,
+      "loss": 1.0402,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9891706905442785e-05,
+      "loss": 0.4479,
+      "step": 2155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9891238104167645e-05,
+      "loss": 0.2284,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9890769302892505e-05,
+      "loss": 0.2668,
+      "step": 2165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9890300501617365e-05,
+      "loss": 0.3342,
+      "step": 2170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9889831700342225e-05,
+      "loss": 0.4317,
+      "step": 2175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9889362899067085e-05,
+      "loss": 0.4107,
+      "step": 2180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9888894097791948e-05,
+      "loss": 0.3555,
+      "step": 2185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9888425296516808e-05,
+      "loss": 0.5225,
+      "step": 2190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9887956495241668e-05,
+      "loss": 0.9319,
+      "step": 2195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9887487693966528e-05,
+      "loss": 0.9103,
+      "step": 2200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988701889269139e-05,
+      "loss": 0.5407,
+      "step": 2205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988655009141625e-05,
+      "loss": 0.2933,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988608129014111e-05,
+      "loss": 0.3158,
+      "step": 2215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988561248886597e-05,
+      "loss": 0.3756,
+      "step": 2220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988514368759083e-05,
+      "loss": 0.3151,
+      "step": 2225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.988467488631569e-05,
+      "loss": 0.3484,
+      "step": 2230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9884206085040554e-05,
+      "loss": 0.514,
+      "step": 2235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9883737283765414e-05,
+      "loss": 0.4266,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9883268482490274e-05,
+      "loss": 0.7312,
+      "step": 2245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9882799681215134e-05,
+      "loss": 0.7081,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9882330879939994e-05,
+      "loss": 0.4765,
+      "step": 2255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9881862078664854e-05,
+      "loss": 0.2093,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9881393277389714e-05,
+      "loss": 0.3321,
+      "step": 2265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9880924476114577e-05,
+      "loss": 0.3887,
+      "step": 2270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9880455674839437e-05,
+      "loss": 0.3139,
+      "step": 2275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9879986873564297e-05,
+      "loss": 0.3986,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987951807228916e-05,
+      "loss": 0.502,
+      "step": 2285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987904927101402e-05,
+      "loss": 0.4883,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987858046973888e-05,
+      "loss": 0.5289,
+      "step": 2295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987811166846374e-05,
+      "loss": 1.2296,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.98776428671886e-05,
+      "loss": 0.4979,
+      "step": 2305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987717406591346e-05,
+      "loss": 0.1878,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987670526463832e-05,
+      "loss": 0.3966,
+      "step": 2315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987623646336318e-05,
+      "loss": 0.331,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9875767662088043e-05,
+      "loss": 0.3299,
+      "step": 2325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9875298860812903e-05,
+      "loss": 0.3091,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9874830059537763e-05,
+      "loss": 0.4212,
+      "step": 2335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9874361258262623e-05,
+      "loss": 0.3994,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9873892456987483e-05,
+      "loss": 0.6617,
+      "step": 2345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9873423655712346e-05,
+      "loss": 0.7071,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9872954854437206e-05,
+      "loss": 0.3459,
+      "step": 2355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9872486053162066e-05,
+      "loss": 0.1769,
+      "step": 2360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9872017251886926e-05,
+      "loss": 0.3423,
+      "step": 2365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987154845061179e-05,
+      "loss": 0.4013,
+      "step": 2370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987107964933665e-05,
+      "loss": 0.368,
+      "step": 2375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987061084806151e-05,
+      "loss": 0.3948,
+      "step": 2380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987014204678637e-05,
+      "loss": 0.3829,
+      "step": 2385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986967324551123e-05,
+      "loss": 0.3998,
+      "step": 2390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986920444423609e-05,
+      "loss": 0.5129,
+      "step": 2395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986873564296095e-05,
+      "loss": 1.1603,
+      "step": 2400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986826684168581e-05,
+      "loss": 0.4495,
+      "step": 2405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9867798040410672e-05,
+      "loss": 0.2745,
+      "step": 2410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9867329239135532e-05,
+      "loss": 0.3008,
+      "step": 2415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9866860437860392e-05,
+      "loss": 0.3176,
+      "step": 2420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9866391636585256e-05,
+      "loss": 0.3361,
+      "step": 2425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9865922835310115e-05,
+      "loss": 0.4109,
+      "step": 2430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9865454034034975e-05,
+      "loss": 0.4295,
+      "step": 2435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9864985232759835e-05,
+      "loss": 0.6172,
+      "step": 2440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9864516431484695e-05,
+      "loss": 0.5372,
+      "step": 2445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9864047630209555e-05,
+      "loss": 1.0839,
+      "step": 2450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9863578828934415e-05,
+      "loss": 0.5672,
+      "step": 2455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9863110027659275e-05,
+      "loss": 0.2294,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.986264122638414e-05,
+      "loss": 0.4235,
+      "step": 2465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9862172425109e-05,
+      "loss": 0.3189,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9861703623833858e-05,
+      "loss": 0.2703,
+      "step": 2475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9861234822558718e-05,
+      "loss": 0.4522,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9860766021283578e-05,
+      "loss": 0.648,
+      "step": 2485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.986029722000844e-05,
+      "loss": 0.4643,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98598284187333e-05,
+      "loss": 0.5613,
+      "step": 2495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985935961745816e-05,
+      "loss": 0.9657,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985889081618302e-05,
+      "loss": 0.3885,
+      "step": 2505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9858422014907885e-05,
+      "loss": 0.252,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9857953213632744e-05,
+      "loss": 0.3157,
+      "step": 2515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9857484412357604e-05,
+      "loss": 0.331,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9857015611082464e-05,
+      "loss": 0.3348,
+      "step": 2525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9856546809807324e-05,
+      "loss": 0.4366,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9856078008532184e-05,
+      "loss": 0.5167,
+      "step": 2535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9855609207257044e-05,
+      "loss": 0.5484,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9855140405981904e-05,
+      "loss": 0.682,
+      "step": 2545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9854671604706764e-05,
+      "loss": 0.9292,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9854202803431627e-05,
+      "loss": 0.2851,
+      "step": 2555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9853734002156487e-05,
+      "loss": 0.319,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9853265200881347e-05,
+      "loss": 0.3735,
+      "step": 2565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985279639960621e-05,
+      "loss": 0.3128,
+      "step": 2570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985232759833107e-05,
+      "loss": 0.3411,
+      "step": 2575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985185879705593e-05,
+      "loss": 0.4413,
+      "step": 2580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985138999578079e-05,
+      "loss": 0.5069,
+      "step": 2585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985092119450565e-05,
+      "loss": 0.6175,
+      "step": 2590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.985045239323051e-05,
+      "loss": 0.5128,
+      "step": 2595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984998359195537e-05,
+      "loss": 0.7707,
+      "step": 2600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9849514790680233e-05,
+      "loss": 0.388,
+      "step": 2605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9849045989405093e-05,
+      "loss": 0.2559,
+      "step": 2610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9848577188129953e-05,
+      "loss": 0.3089,
+      "step": 2615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9848108386854813e-05,
+      "loss": 0.2978,
+      "step": 2620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9847639585579673e-05,
+      "loss": 0.2533,
+      "step": 2625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9847170784304533e-05,
+      "loss": 0.3758,
+      "step": 2630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9846701983029396e-05,
+      "loss": 0.2824,
+      "step": 2635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9846233181754256e-05,
+      "loss": 0.4697,
+      "step": 2640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9845764380479116e-05,
+      "loss": 0.4773,
+      "step": 2645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984529557920398e-05,
+      "loss": 0.9044,
+      "step": 2650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984482677792884e-05,
+      "loss": 0.3903,
+      "step": 2655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98443579766537e-05,
+      "loss": 0.268,
+      "step": 2660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984388917537856e-05,
+      "loss": 0.205,
+      "step": 2665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984342037410342e-05,
+      "loss": 0.3019,
+      "step": 2670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.984295157282828e-05,
+      "loss": 0.4408,
+      "step": 2675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.984248277155314e-05,
+      "loss": 0.3562,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9842013970278e-05,
+      "loss": 0.4502,
+      "step": 2685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.984154516900286e-05,
+      "loss": 0.6325,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9841076367727722e-05,
+      "loss": 0.734,
+      "step": 2695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9840607566452582e-05,
+      "loss": 0.825,
+      "step": 2700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9840138765177442e-05,
+      "loss": 0.4505,
+      "step": 2705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9839669963902302e-05,
+      "loss": 0.2048,
+      "step": 2710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9839201162627166e-05,
+      "loss": 0.2263,
+      "step": 2715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9838732361352025e-05,
+      "loss": 0.2975,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9838263560076885e-05,
+      "loss": 0.3431,
+      "step": 2725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9837794758801745e-05,
+      "loss": 0.4268,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9837325957526605e-05,
+      "loss": 0.4063,
+      "step": 2735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9836857156251465e-05,
+      "loss": 0.4808,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983638835497633e-05,
+      "loss": 0.6248,
+      "step": 2745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983591955370119e-05,
+      "loss": 0.9324,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983545075242605e-05,
+      "loss": 0.4698,
+      "step": 2755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983498195115091e-05,
+      "loss": 0.2148,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9834513149875768e-05,
+      "loss": 0.2496,
+      "step": 2765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9834044348600628e-05,
+      "loss": 0.3243,
+      "step": 2770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983357554732549e-05,
+      "loss": 0.312,
+      "step": 2775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983310674605035e-05,
+      "loss": 0.3524,
+      "step": 2780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983263794477521e-05,
+      "loss": 0.4036,
+      "step": 2785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.983216914350007e-05,
+      "loss": 0.4246,
+      "step": 2790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9831700342224935e-05,
+      "loss": 0.6055,
+      "step": 2795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9831231540949795e-05,
+      "loss": 0.6449,
+      "step": 2800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9830762739674655e-05,
+      "loss": 0.3477,
+      "step": 2805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9830293938399514e-05,
+      "loss": 0.3349,
+      "step": 2810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9829825137124374e-05,
+      "loss": 0.2735,
+      "step": 2815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9829356335849234e-05,
+      "loss": 0.294,
+      "step": 2820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9828887534574094e-05,
+      "loss": 0.3747,
+      "step": 2825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9828418733298954e-05,
+      "loss": 0.3981,
+      "step": 2830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9827949932023818e-05,
+      "loss": 0.5341,
+      "step": 2835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9827481130748677e-05,
+      "loss": 0.4484,
+      "step": 2840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9827012329473537e-05,
+      "loss": 0.6313,
+      "step": 2845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9826543528198397e-05,
+      "loss": 0.8526,
+      "step": 2850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.982607472692326e-05,
+      "loss": 0.365,
+      "step": 2855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.982560592564812e-05,
+      "loss": 0.2652,
+      "step": 2860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.982513712437298e-05,
+      "loss": 0.239,
+      "step": 2865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.982466832309784e-05,
+      "loss": 0.3384,
+      "step": 2870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.98241995218227e-05,
+      "loss": 0.3417,
+      "step": 2875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.982373072054756e-05,
+      "loss": 0.348,
+      "step": 2880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9823261919272424e-05,
+      "loss": 0.3826,
+      "step": 2885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9822793117997284e-05,
+      "loss": 0.4936,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9822324316722143e-05,
+      "loss": 0.5807,
+      "step": 2895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9821855515447003e-05,
+      "loss": 1.0115,
+      "step": 2900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9821386714171863e-05,
+      "loss": 0.3353,
+      "step": 2905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9820917912896723e-05,
+      "loss": 0.2495,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9820449111621583e-05,
+      "loss": 0.2535,
+      "step": 2915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9819980310346447e-05,
+      "loss": 0.3094,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9819511509071306e-05,
+      "loss": 0.2988,
+      "step": 2925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9819042707796166e-05,
+      "loss": 0.3186,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981857390652103e-05,
+      "loss": 0.4027,
+      "step": 2935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981810510524589e-05,
+      "loss": 0.4593,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981763630397075e-05,
+      "loss": 0.6236,
+      "step": 2945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981716750269561e-05,
+      "loss": 0.9071,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981669870142047e-05,
+      "loss": 0.4223,
+      "step": 2955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981622990014533e-05,
+      "loss": 0.197,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981576109887019e-05,
+      "loss": 0.298,
+      "step": 2965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981529229759505e-05,
+      "loss": 0.3526,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9814823496319913e-05,
+      "loss": 0.4274,
+      "step": 2975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9814354695044773e-05,
+      "loss": 0.3132,
+      "step": 2980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9813885893769632e-05,
+      "loss": 0.4685,
+      "step": 2985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9813417092494492e-05,
+      "loss": 0.5327,
+      "step": 2990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9812948291219352e-05,
+      "loss": 0.5887,
+      "step": 2995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9812479489944216e-05,
+      "loss": 0.8516,
+      "step": 3000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9812010688669076e-05,
+      "loss": 0.3981,
+      "step": 3005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9811541887393936e-05,
+      "loss": 0.2334,
+      "step": 3010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9811073086118795e-05,
+      "loss": 0.2776,
+      "step": 3015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981060428484366e-05,
+      "loss": 0.2129,
+      "step": 3020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.981013548356852e-05,
+      "loss": 0.3171,
+      "step": 3025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980966668229338e-05,
+      "loss": 0.3339,
+      "step": 3030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980919788101824e-05,
+      "loss": 0.4108,
+      "step": 3035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98087290797431e-05,
+      "loss": 0.4631,
+      "step": 3040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980826027846796e-05,
+      "loss": 0.5093,
+      "step": 3045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980779147719282e-05,
+      "loss": 0.8253,
+      "step": 3050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980732267591768e-05,
+      "loss": 0.5317,
+      "step": 3055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9806853874642538e-05,
+      "loss": 0.2176,
+      "step": 3060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.98063850733674e-05,
+      "loss": 0.2664,
+      "step": 3065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980591627209226e-05,
+      "loss": 0.2988,
+      "step": 3070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.980544747081712e-05,
+      "loss": 0.3123,
+      "step": 3075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9804978669541985e-05,
+      "loss": 0.3984,
+      "step": 3080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9804509868266845e-05,
+      "loss": 0.4429,
+      "step": 3085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9804041066991705e-05,
+      "loss": 0.4721,
+      "step": 3090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9803572265716565e-05,
+      "loss": 0.608,
+      "step": 3095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9803103464441424e-05,
+      "loss": 0.7198,
+      "step": 3100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9802634663166284e-05,
+      "loss": 0.4533,
+      "step": 3105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9802165861891144e-05,
+      "loss": 0.2521,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9801697060616008e-05,
+      "loss": 0.3518,
+      "step": 3115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9801228259340868e-05,
+      "loss": 0.3765,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9800759458065728e-05,
+      "loss": 0.4222,
+      "step": 3125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9800290656790587e-05,
+      "loss": 0.3578,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9799821855515447e-05,
+      "loss": 0.3973,
+      "step": 3135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9799353054240307e-05,
+      "loss": 0.4282,
+      "step": 3140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979888425296517e-05,
+      "loss": 0.5091,
+      "step": 3145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979841545169003e-05,
+      "loss": 1.0688,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979794665041489e-05,
+      "loss": 0.4779,
+      "step": 3155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9797477849139754e-05,
+      "loss": 0.2231,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9797009047864614e-05,
+      "loss": 0.2028,
+      "step": 3165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9796540246589474e-05,
+      "loss": 0.2761,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9796071445314334e-05,
+      "loss": 0.4209,
+      "step": 3175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9795602644039194e-05,
+      "loss": 0.3276,
+      "step": 3180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9795133842764054e-05,
+      "loss": 0.4656,
+      "step": 3185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9794665041488913e-05,
+      "loss": 0.4872,
+      "step": 3190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9794196240213773e-05,
+      "loss": 0.573,
+      "step": 3195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9793727438938633e-05,
+      "loss": 0.9714,
+      "step": 3200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9793258637663497e-05,
+      "loss": 0.449,
+      "step": 3205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9792789836388357e-05,
+      "loss": 0.2523,
+      "step": 3210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9792321035113217e-05,
+      "loss": 0.2467,
+      "step": 3215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9791852233838076e-05,
+      "loss": 0.2898,
+      "step": 3220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979138343256294e-05,
+      "loss": 0.3189,
+      "step": 3225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.97909146312878e-05,
+      "loss": 0.4171,
+      "step": 3230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.979044583001266e-05,
+      "loss": 0.4367,
+      "step": 3235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978997702873752e-05,
+      "loss": 0.4654,
+      "step": 3240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978950822746238e-05,
+      "loss": 0.5939,
+      "step": 3245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978903942618724e-05,
+      "loss": 1.0234,
+      "step": 3250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9788570624912103e-05,
+      "loss": 0.4174,
+      "step": 3255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9788101823636963e-05,
+      "loss": 0.1468,
+      "step": 3260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9787633022361823e-05,
+      "loss": 0.2742,
+      "step": 3265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9787164221086683e-05,
+      "loss": 0.3599,
+      "step": 3270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9786695419811542e-05,
+      "loss": 0.2513,
+      "step": 3275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9786226618536402e-05,
+      "loss": 0.2906,
+      "step": 3280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9785757817261266e-05,
+      "loss": 0.4572,
+      "step": 3285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9785289015986126e-05,
+      "loss": 0.3669,
+      "step": 3290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9784820214710986e-05,
+      "loss": 0.5854,
+      "step": 3295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9784351413435846e-05,
+      "loss": 0.8587,
+      "step": 3300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978388261216071e-05,
+      "loss": 0.3697,
+      "step": 3305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978341381088557e-05,
+      "loss": 0.1952,
+      "step": 3310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978294500961043e-05,
+      "loss": 0.252,
+      "step": 3315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.978247620833529e-05,
+      "loss": 0.3353,
+      "step": 3320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978200740706015e-05,
+      "loss": 0.3767,
+      "step": 3325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978153860578501e-05,
+      "loss": 0.3057,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978106980450987e-05,
+      "loss": 0.4199,
+      "step": 3335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.978060100323473e-05,
+      "loss": 0.4672,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9780132201959592e-05,
+      "loss": 0.5891,
+      "step": 3345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977966340068445e-05,
+      "loss": 1.0011,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977919459940931e-05,
+      "loss": 0.3955,
+      "step": 3355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977872579813417e-05,
+      "loss": 0.2046,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9778256996859035e-05,
+      "loss": 0.1842,
+      "step": 3365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9777788195583895e-05,
+      "loss": 0.3045,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9777319394308755e-05,
+      "loss": 0.3251,
+      "step": 3375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9776850593033615e-05,
+      "loss": 0.4542,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9776381791758475e-05,
+      "loss": 0.428,
+      "step": 3385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9775912990483335e-05,
+      "loss": 0.602,
+      "step": 3390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9775444189208198e-05,
+      "loss": 0.472,
+      "step": 3395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9774975387933058e-05,
+      "loss": 0.8672,
+      "step": 3400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9774506586657918e-05,
+      "loss": 0.3989,
+      "step": 3405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9774037785382778e-05,
+      "loss": 0.1915,
+      "step": 3410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9773568984107638e-05,
+      "loss": 0.222,
+      "step": 3415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9773100182832498e-05,
+      "loss": 0.2807,
+      "step": 3420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9772631381557357e-05,
+      "loss": 0.369,
+      "step": 3425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977216258028222e-05,
+      "loss": 0.4569,
+      "step": 3430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977169377900708e-05,
+      "loss": 0.3986,
+      "step": 3435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.977122497773194e-05,
+      "loss": 0.5065,
+      "step": 3440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9770756176456804e-05,
+      "loss": 0.5453,
+      "step": 3445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9770287375181664e-05,
+      "loss": 0.879,
+      "step": 3450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9769818573906524e-05,
+      "loss": 0.4329,
+      "step": 3455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9769349772631384e-05,
+      "loss": 0.2126,
+      "step": 3460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9768880971356244e-05,
+      "loss": 0.3154,
+      "step": 3465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9768412170081104e-05,
+      "loss": 0.314,
+      "step": 3470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9767943368805964e-05,
+      "loss": 0.3373,
+      "step": 3475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9767474567530823e-05,
+      "loss": 0.2566,
+      "step": 3480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9767005766255687e-05,
+      "loss": 0.4568,
+      "step": 3485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9766536964980547e-05,
+      "loss": 0.3373,
+      "step": 3490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9766068163705407e-05,
+      "loss": 0.5181,
+      "step": 3495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9765599362430267e-05,
+      "loss": 0.8484,
+      "step": 3500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9765130561155127e-05,
+      "loss": 0.3833,
+      "step": 3505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.976466175987999e-05,
+      "loss": 0.1497,
+      "step": 3510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.976419295860485e-05,
+      "loss": 0.235,
+      "step": 3515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.976372415732971e-05,
+      "loss": 0.2564,
+      "step": 3520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.976325535605457e-05,
+      "loss": 0.2755,
+      "step": 3525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.976278655477943e-05,
+      "loss": 0.416,
+      "step": 3530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9762317753504293e-05,
+      "loss": 0.4785,
+      "step": 3535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9761848952229153e-05,
+      "loss": 0.5485,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9761380150954013e-05,
+      "loss": 0.6251,
+      "step": 3545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9760911349678873e-05,
+      "loss": 0.8421,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9760442548403733e-05,
+      "loss": 0.3436,
+      "step": 3555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9759973747128593e-05,
+      "loss": 0.1759,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9759504945853453e-05,
+      "loss": 0.2988,
+      "step": 3565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9759036144578312e-05,
+      "loss": 0.2328,
+      "step": 3570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9758567343303176e-05,
+      "loss": 0.3053,
+      "step": 3575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9758098542028036e-05,
+      "loss": 0.3479,
+      "step": 3580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9757629740752896e-05,
+      "loss": 0.377,
+      "step": 3585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975716093947776e-05,
+      "loss": 0.4114,
+      "step": 3590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975669213820262e-05,
+      "loss": 0.488,
+      "step": 3595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975622333692748e-05,
+      "loss": 0.5475,
+      "step": 3600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975575453565234e-05,
+      "loss": 0.5718,
+      "step": 3605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.97552857343772e-05,
+      "loss": 0.1882,
+      "step": 3610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975481693310206e-05,
+      "loss": 0.2593,
+      "step": 3615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975434813182692e-05,
+      "loss": 0.2558,
+      "step": 3620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9753879330551782e-05,
+      "loss": 0.2394,
+      "step": 3625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9753410529276642e-05,
+      "loss": 0.4207,
+      "step": 3630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9752941728001502e-05,
+      "loss": 0.4391,
+      "step": 3635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9752472926726362e-05,
+      "loss": 0.7291,
+      "step": 3640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975200412545122e-05,
+      "loss": 0.4494,
+      "step": 3645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.975153532417608e-05,
+      "loss": 0.7069,
+      "step": 3650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9751066522900945e-05,
+      "loss": 0.3945,
+      "step": 3655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9750597721625805e-05,
+      "loss": 0.2034,
+      "step": 3660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9750128920350665e-05,
+      "loss": 0.2243,
+      "step": 3665
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9749660119075528e-05,
+      "loss": 0.2732,
+      "step": 3670
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9749191317800388e-05,
+      "loss": 0.3621,
+      "step": 3675
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9748722516525248e-05,
+      "loss": 0.3214,
+      "step": 3680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9748253715250108e-05,
+      "loss": 0.5254,
+      "step": 3685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9747784913974968e-05,
+      "loss": 0.5203,
+      "step": 3690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9747316112699828e-05,
+      "loss": 0.7242,
+      "step": 3695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9746847311424688e-05,
+      "loss": 0.83,
+      "step": 3700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9746378510149548e-05,
+      "loss": 0.3907,
+      "step": 3705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9745909708874408e-05,
+      "loss": 0.1778,
+      "step": 3710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.974544090759927e-05,
+      "loss": 0.2049,
+      "step": 3715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.974497210632413e-05,
+      "loss": 0.3739,
+      "step": 3720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.974450330504899e-05,
+      "loss": 0.3008,
+      "step": 3725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.974403450377385e-05,
+      "loss": 0.2455,
+      "step": 3730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9743565702498714e-05,
+      "loss": 0.4178,
+      "step": 3735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9743096901223574e-05,
+      "loss": 0.5493,
+      "step": 3740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9742628099948434e-05,
+      "loss": 0.4286,
+      "step": 3745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9742159298673294e-05,
+      "loss": 0.6876,
+      "step": 3750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9741690497398154e-05,
+      "loss": 0.3451,
+      "step": 3755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9741221696123014e-05,
+      "loss": 0.1612,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9740752894847877e-05,
+      "loss": 0.2758,
+      "step": 3765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9740284093572737e-05,
+      "loss": 0.2768,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9739815292297597e-05,
+      "loss": 0.3298,
+      "step": 3775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9739346491022457e-05,
+      "loss": 0.3231,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9738877689747317e-05,
+      "loss": 0.3359,
+      "step": 3785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9738408888472177e-05,
+      "loss": 0.3687,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973794008719704e-05,
+      "loss": 0.5007,
+      "step": 3795
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.97374712859219e-05,
+      "loss": 0.9462,
+      "step": 3800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973700248464676e-05,
+      "loss": 0.4636,
+      "step": 3805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973653368337162e-05,
+      "loss": 0.1924,
+      "step": 3810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9736064882096483e-05,
+      "loss": 0.1923,
+      "step": 3815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9735596080821343e-05,
+      "loss": 0.2264,
+      "step": 3820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9735127279546203e-05,
+      "loss": 0.3736,
+      "step": 3825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9734658478271063e-05,
+      "loss": 0.4152,
+      "step": 3830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9734189676995923e-05,
+      "loss": 0.4174,
+      "step": 3835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9733720875720783e-05,
+      "loss": 0.4885,
+      "step": 3840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9733252074445643e-05,
+      "loss": 0.4311,
+      "step": 3845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9732783273170503e-05,
+      "loss": 0.6807,
+      "step": 3850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9732314471895363e-05,
+      "loss": 0.4367,
+      "step": 3855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9731845670620226e-05,
+      "loss": 0.1888,
+      "step": 3860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9731376869345086e-05,
+      "loss": 0.2237,
+      "step": 3865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9730908068069946e-05,
+      "loss": 0.2441,
+      "step": 3870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.973043926679481e-05,
+      "loss": 0.3274,
+      "step": 3875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972997046551967e-05,
+      "loss": 0.3156,
+      "step": 3880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972950166424453e-05,
+      "loss": 0.3368,
+      "step": 3885
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972903286296939e-05,
+      "loss": 0.4301,
+      "step": 3890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972856406169425e-05,
+      "loss": 0.6439,
+      "step": 3895
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972809526041911e-05,
+      "loss": 0.839,
+      "step": 3900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9727626459143972e-05,
+      "loss": 0.4082,
+      "step": 3905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9727157657868832e-05,
+      "loss": 0.1958,
+      "step": 3910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9726688856593692e-05,
+      "loss": 0.1713,
+      "step": 3915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9726220055318552e-05,
+      "loss": 0.2028,
+      "step": 3920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9725751254043412e-05,
+      "loss": 0.2407,
+      "step": 3925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9725282452768272e-05,
+      "loss": 0.4262,
+      "step": 3930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.972481365149313e-05,
+      "loss": 0.3567,
+      "step": 3935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9724344850217995e-05,
+      "loss": 0.4333,
+      "step": 3940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9723876048942855e-05,
+      "loss": 0.3769,
+      "step": 3945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9723407247667715e-05,
+      "loss": 0.9714,
+      "step": 3950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9722938446392578e-05,
+      "loss": 0.3421,
+      "step": 3955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9722469645117438e-05,
+      "loss": 0.1904,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9722000843842298e-05,
+      "loss": 0.2153,
+      "step": 3965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9721532042567158e-05,
+      "loss": 0.2275,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9721063241292018e-05,
+      "loss": 0.3917,
+      "step": 3975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9720594440016878e-05,
+      "loss": 0.4462,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9720125638741738e-05,
+      "loss": 0.319,
+      "step": 3985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9719656837466598e-05,
+      "loss": 0.2997,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971918803619146e-05,
+      "loss": 0.601,
+      "step": 3995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971871923491632e-05,
+      "loss": 0.6658,
+      "step": 4000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971825043364118e-05,
+      "loss": 0.3592,
+      "step": 4005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971778163236604e-05,
+      "loss": 0.1744,
+      "step": 4010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.97173128310909e-05,
+      "loss": 0.2081,
+      "step": 4015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9716844029815764e-05,
+      "loss": 0.2472,
+      "step": 4020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9716375228540624e-05,
+      "loss": 0.2668,
+      "step": 4025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9715906427265484e-05,
+      "loss": 0.2763,
+      "step": 4030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9715437625990344e-05,
+      "loss": 0.3903,
+      "step": 4035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9714968824715204e-05,
+      "loss": 0.4871,
+      "step": 4040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9714500023440067e-05,
+      "loss": 0.6568,
+      "step": 4045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9714031222164927e-05,
+      "loss": 0.7185,
+      "step": 4050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9713562420889787e-05,
+      "loss": 0.416,
+      "step": 4055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9713093619614647e-05,
+      "loss": 0.1623,
+      "step": 4060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9712624818339507e-05,
+      "loss": 0.1905,
+      "step": 4065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9712156017064367e-05,
+      "loss": 0.2243,
+      "step": 4070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9711687215789227e-05,
+      "loss": 0.2595,
+      "step": 4075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9711218414514087e-05,
+      "loss": 0.2343,
+      "step": 4080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971074961323895e-05,
+      "loss": 0.3081,
+      "step": 4085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.971028081196381e-05,
+      "loss": 0.4051,
+      "step": 4090
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970981201068867e-05,
+      "loss": 0.4353,
+      "step": 4095
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9709343209413533e-05,
+      "loss": 0.8488,
+      "step": 4100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9708874408138393e-05,
+      "loss": 0.3971,
+      "step": 4105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9708405606863253e-05,
+      "loss": 0.1639,
+      "step": 4110
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9707936805588113e-05,
+      "loss": 0.1971,
+      "step": 4115
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9707468004312973e-05,
+      "loss": 0.1918,
+      "step": 4120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9706999203037833e-05,
+      "loss": 0.2442,
+      "step": 4125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9706530401762693e-05,
+      "loss": 0.3431,
+      "step": 4130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9706061600487556e-05,
+      "loss": 0.4143,
+      "step": 4135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9705592799212416e-05,
+      "loss": 0.563,
+      "step": 4140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9705123997937276e-05,
+      "loss": 0.5536,
+      "step": 4145
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9704655196662136e-05,
+      "loss": 0.8869,
+      "step": 4150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9704186395386996e-05,
+      "loss": 0.3938,
+      "step": 4155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9703717594111856e-05,
+      "loss": 0.1912,
+      "step": 4160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970324879283672e-05,
+      "loss": 0.2132,
+      "step": 4165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970277999156158e-05,
+      "loss": 0.1632,
+      "step": 4170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.970231119028644e-05,
+      "loss": 0.2362,
+      "step": 4175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.97018423890113e-05,
+      "loss": 0.2287,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9701373587736162e-05,
+      "loss": 0.3168,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9700904786461022e-05,
+      "loss": 0.4876,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9700435985185882e-05,
+      "loss": 0.5353,
+      "step": 4195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9699967183910742e-05,
+      "loss": 0.6848,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9699498382635602e-05,
+      "loss": 0.4614,
+      "step": 4205
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9699029581360462e-05,
+      "loss": 0.2455,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9698560780085322e-05,
+      "loss": 0.1861,
+      "step": 4215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9698091978810182e-05,
+      "loss": 0.3312,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9697623177535045e-05,
+      "loss": 0.2629,
+      "step": 4225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9697154376259905e-05,
+      "loss": 0.2331,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9696685574984765e-05,
+      "loss": 0.4012,
+      "step": 4235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9696216773709628e-05,
+      "loss": 0.3765,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9695747972434488e-05,
+      "loss": 0.5471,
+      "step": 4245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9695279171159348e-05,
+      "loss": 0.723,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9694810369884208e-05,
+      "loss": 0.4313,
+      "step": 4255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9694341568609068e-05,
+      "loss": 0.13,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9693872767333928e-05,
+      "loss": 0.2133,
+      "step": 4265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9693403966058788e-05,
+      "loss": 0.2618,
+      "step": 4270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969293516478365e-05,
+      "loss": 0.2428,
+      "step": 4275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969246636350851e-05,
+      "loss": 0.3027,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969199756223337e-05,
+      "loss": 0.3958,
+      "step": 4285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969152876095823e-05,
+      "loss": 0.5407,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969105995968309e-05,
+      "loss": 0.5288,
+      "step": 4295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.969059115840795e-05,
+      "loss": 0.9622,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9690122357132814e-05,
+      "loss": 0.3357,
+      "step": 4305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9689653555857674e-05,
+      "loss": 0.2422,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9689184754582534e-05,
+      "loss": 0.2317,
+      "step": 4315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9688715953307397e-05,
+      "loss": 0.3139,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9688247152032257e-05,
+      "loss": 0.2611,
+      "step": 4325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9687778350757117e-05,
+      "loss": 0.274,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9687309549481977e-05,
+      "loss": 0.4473,
+      "step": 4335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9686840748206837e-05,
+      "loss": 0.3167,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9686371946931697e-05,
+      "loss": 0.5854,
+      "step": 4345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9685903145656557e-05,
+      "loss": 0.8531,
+      "step": 4350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9685434344381417e-05,
+      "loss": 0.3545,
+      "step": 4355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9684965543106277e-05,
+      "loss": 0.1751,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9684496741831137e-05,
+      "loss": 0.2515,
+      "step": 4365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9684027940556e-05,
+      "loss": 0.2662,
+      "step": 4370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968355913928086e-05,
+      "loss": 0.2842,
+      "step": 4375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.968309033800572e-05,
+      "loss": 0.4317,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9682621536730583e-05,
+      "loss": 0.3946,
+      "step": 4385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9682152735455443e-05,
+      "loss": 0.3546,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9681683934180303e-05,
+      "loss": 0.4726,
+      "step": 4395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9681215132905163e-05,
+      "loss": 0.6882,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9680746331630023e-05,
+      "loss": 0.3417,
+      "step": 4405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9680277530354883e-05,
+      "loss": 0.1884,
+      "step": 4410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9679808729079746e-05,
+      "loss": 0.2584,
+      "step": 4415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9679339927804606e-05,
+      "loss": 0.1648,
+      "step": 4420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9678871126529466e-05,
+      "loss": 0.3163,
+      "step": 4425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9678402325254326e-05,
+      "loss": 0.2756,
+      "step": 4430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9677933523979186e-05,
+      "loss": 0.3836,
+      "step": 4435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9677464722704046e-05,
+      "loss": 0.3857,
+      "step": 4440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9676995921428906e-05,
+      "loss": 0.5504,
+      "step": 4445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.967652712015377e-05,
+      "loss": 0.796,
+      "step": 4450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.967605831887863e-05,
+      "loss": 0.4518,
+      "step": 4455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.967558951760349e-05,
+      "loss": 0.2493,
+      "step": 4460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9675120716328352e-05,
+      "loss": 0.1496,
+      "step": 4465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9674651915053212e-05,
+      "loss": 0.2788,
+      "step": 4470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9674183113778072e-05,
+      "loss": 0.2969,
+      "step": 4475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9673714312502932e-05,
+      "loss": 0.3378,
+      "step": 4480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9673245511227792e-05,
+      "loss": 0.5223,
+      "step": 4485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9672776709952652e-05,
+      "loss": 0.4544,
+      "step": 4490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9672307908677512e-05,
+      "loss": 0.6628,
+      "step": 4495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9671839107402372e-05,
+      "loss": 0.898,
+      "step": 4500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9671370306127232e-05,
+      "loss": 0.3878,
+      "step": 4505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9670901504852095e-05,
+      "loss": 0.2018,
+      "step": 4510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9670432703576955e-05,
+      "loss": 0.2069,
+      "step": 4515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9669963902301815e-05,
+      "loss": 0.242,
+      "step": 4520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9669495101026675e-05,
+      "loss": 0.2806,
+      "step": 4525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9669026299751538e-05,
+      "loss": 0.3503,
+      "step": 4530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9668557498476398e-05,
+      "loss": 0.2483,
+      "step": 4535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9668088697201258e-05,
+      "loss": 0.3632,
+      "step": 4540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9667619895926118e-05,
+      "loss": 0.5749,
+      "step": 4545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9667151094650978e-05,
+      "loss": 0.6014,
+      "step": 4550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966668229337584e-05,
+      "loss": 0.3388,
+      "step": 4555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.96662134921007e-05,
+      "loss": 0.1989,
+      "step": 4560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966574469082556e-05,
+      "loss": 0.2078,
+      "step": 4565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966527588955042e-05,
+      "loss": 0.2163,
+      "step": 4570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966480708827528e-05,
+      "loss": 0.2737,
+      "step": 4575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.966433828700014e-05,
+      "loss": 0.347,
+      "step": 4580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9663869485725e-05,
+      "loss": 0.3929,
+      "step": 4585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9663400684449864e-05,
+      "loss": 0.4723,
+      "step": 4590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9662931883174724e-05,
+      "loss": 0.5059,
+      "step": 4595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9662463081899584e-05,
+      "loss": 0.689,
+      "step": 4600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9661994280624444e-05,
+      "loss": 0.374,
+      "step": 4605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9661525479349307e-05,
+      "loss": 0.1667,
+      "step": 4610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9661056678074167e-05,
+      "loss": 0.1707,
+      "step": 4615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660587876799027e-05,
+      "loss": 0.196,
+      "step": 4620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660119075523887e-05,
+      "loss": 0.3145,
+      "step": 4625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659650274248747e-05,
+      "loss": 0.3389,
+      "step": 4630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9659181472973607e-05,
+      "loss": 0.4153,
+      "step": 4635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9658712671698467e-05,
+      "loss": 0.3724,
+      "step": 4640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965824387042333e-05,
+      "loss": 0.5458,
+      "step": 4645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965777506914819e-05,
+      "loss": 0.8993,
+      "step": 4650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965730626787305e-05,
+      "loss": 0.3525,
+      "step": 4655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965683746659791e-05,
+      "loss": 0.2098,
+      "step": 4660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965636866532277e-05,
+      "loss": 0.1881,
+      "step": 4665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9655899864047633e-05,
+      "loss": 0.2259,
+      "step": 4670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9655431062772493e-05,
+      "loss": 0.3217,
+      "step": 4675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654962261497353e-05,
+      "loss": 0.3272,
+      "step": 4680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654493460222213e-05,
+      "loss": 0.3857,
+      "step": 4685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9654024658947073e-05,
+      "loss": 0.4295,
+      "step": 4690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653555857671936e-05,
+      "loss": 0.4549,
+      "step": 4695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9653087056396796e-05,
+      "loss": 0.6043,
+      "step": 4700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652618255121656e-05,
+      "loss": 0.3951,
+      "step": 4705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9652149453846516e-05,
+      "loss": 0.1837,
+      "step": 4710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651680652571376e-05,
+      "loss": 0.2177,
+      "step": 4715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9651211851296236e-05,
+      "loss": 0.2921,
+      "step": 4720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9650743050021096e-05,
+      "loss": 0.3297,
+      "step": 4725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9650274248745956e-05,
+      "loss": 0.3355,
+      "step": 4730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964980544747082e-05,
+      "loss": 0.4129,
+      "step": 4735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964933664619568e-05,
+      "loss": 0.425,
+      "step": 4740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964886784492054e-05,
+      "loss": 0.4326,
+      "step": 4745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9648399043645402e-05,
+      "loss": 0.72,
+      "step": 4750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9647930242370262e-05,
+      "loss": 0.4429,
+      "step": 4755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9647461441095122e-05,
+      "loss": 0.2009,
+      "step": 4760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9646992639819982e-05,
+      "loss": 0.195,
+      "step": 4765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9646523838544842e-05,
+      "loss": 0.2269,
+      "step": 4770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9646055037269702e-05,
+      "loss": 0.2516,
+      "step": 4775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9645586235994562e-05,
+      "loss": 0.2693,
+      "step": 4780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9645117434719425e-05,
+      "loss": 0.4334,
+      "step": 4785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9644648633444285e-05,
+      "loss": 0.4747,
+      "step": 4790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9644179832169145e-05,
+      "loss": 0.5062,
+      "step": 4795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9643711030894005e-05,
+      "loss": 0.627,
+      "step": 4800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9643242229618865e-05,
+      "loss": 0.2878,
+      "step": 4805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9642773428343725e-05,
+      "loss": 0.1459,
+      "step": 4810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964230462706859e-05,
+      "loss": 0.1581,
+      "step": 4815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964183582579345e-05,
+      "loss": 0.2001,
+      "step": 4820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9641367024518308e-05,
+      "loss": 0.2857,
+      "step": 4825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9640898223243168e-05,
+      "loss": 0.2737,
+      "step": 4830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.964042942196803e-05,
+      "loss": 0.2742,
+      "step": 4835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963996062069289e-05,
+      "loss": 0.4487,
+      "step": 4840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963949181941775e-05,
+      "loss": 0.4665,
+      "step": 4845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963902301814261e-05,
+      "loss": 0.8322,
+      "step": 4850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963855421686747e-05,
+      "loss": 0.3451,
+      "step": 4855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963808541559233e-05,
+      "loss": 0.1254,
+      "step": 4860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963761661431719e-05,
+      "loss": 0.1569,
+      "step": 4865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963714781304205e-05,
+      "loss": 0.2125,
+      "step": 4870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963667901176691e-05,
+      "loss": 0.2344,
+      "step": 4875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9636210210491774e-05,
+      "loss": 0.2472,
+      "step": 4880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635741409216634e-05,
+      "loss": 0.4773,
+      "step": 4885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9635272607941494e-05,
+      "loss": 0.4222,
+      "step": 4890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634803806666358e-05,
+      "loss": 0.4088,
+      "step": 4895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9634335005391217e-05,
+      "loss": 0.4062,
+      "step": 4900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9633866204116077e-05,
+      "loss": 0.3291,
+      "step": 4905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9633397402840937e-05,
+      "loss": 0.133,
+      "step": 4910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9632928601565797e-05,
+      "loss": 0.2487,
+      "step": 4915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9632459800290657e-05,
+      "loss": 0.1519,
+      "step": 4920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963199099901552e-05,
+      "loss": 0.3122,
+      "step": 4925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963152219774038e-05,
+      "loss": 0.4063,
+      "step": 4930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963105339646524e-05,
+      "loss": 0.3803,
+      "step": 4935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.96305845951901e-05,
+      "loss": 0.4153,
+      "step": 4940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963011579391496e-05,
+      "loss": 0.4193,
+      "step": 4945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962964699263982e-05,
+      "loss": 0.88,
+      "step": 4950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962917819136468e-05,
+      "loss": 0.4152,
+      "step": 4955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628709390089543e-05,
+      "loss": 0.1861,
+      "step": 4960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9628240588814403e-05,
+      "loss": 0.1928,
+      "step": 4965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9627771787539263e-05,
+      "loss": 0.3147,
+      "step": 4970
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9627302986264127e-05,
+      "loss": 0.2551,
+      "step": 4975
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9626834184988987e-05,
+      "loss": 0.3313,
+      "step": 4980
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9626365383713846e-05,
+      "loss": 0.4429,
+      "step": 4985
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625896582438706e-05,
+      "loss": 0.4012,
+      "step": 4990
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9625427781163566e-05,
+      "loss": 0.5365,
+      "step": 4995
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624958979888426e-05,
+      "loss": 0.7345,
+      "step": 5000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624490178613286e-05,
+      "loss": 0.3755,
+      "step": 5005
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9624021377338146e-05,
+      "loss": 0.3096,
+      "step": 5010
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9623552576063006e-05,
+      "loss": 0.1869,
+      "step": 5015
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962308377478787e-05,
+      "loss": 0.3091,
+      "step": 5020
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962261497351273e-05,
+      "loss": 0.3136,
+      "step": 5025
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962214617223759e-05,
+      "loss": 0.2676,
+      "step": 5030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962167737096245e-05,
+      "loss": 0.3055,
+      "step": 5035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9621208569687313e-05,
+      "loss": 0.3587,
+      "step": 5040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9620739768412172e-05,
+      "loss": 0.4906,
+      "step": 5045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9620270967137032e-05,
+      "loss": 0.7448,
+      "step": 5050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9619802165861892e-05,
+      "loss": 0.3217,
+      "step": 5055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9619333364586752e-05,
+      "loss": 0.1442,
+      "step": 5060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9618864563311616e-05,
+      "loss": 0.2075,
+      "step": 5065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9618395762036476e-05,
+      "loss": 0.2914,
+      "step": 5070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9617926960761335e-05,
+      "loss": 0.3173,
+      "step": 5075
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9617458159486195e-05,
+      "loss": 0.367,
+      "step": 5080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9616989358211055e-05,
+      "loss": 0.3039,
+      "step": 5085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9616520556935915e-05,
+      "loss": 0.4082,
+      "step": 5090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9616051755660775e-05,
+      "loss": 0.4816,
+      "step": 5095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961558295438564e-05,
+      "loss": 0.8166,
+      "step": 5100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96151141531105e-05,
+      "loss": 0.3055,
+      "step": 5105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961464535183536e-05,
+      "loss": 0.1351,
+      "step": 5110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9614176550560218e-05,
+      "loss": 0.2588,
+      "step": 5115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961370774928508e-05,
+      "loss": 0.2485,
+      "step": 5120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961323894800994e-05,
+      "loss": 0.3697,
+      "step": 5125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96127701467348e-05,
+      "loss": 0.3124,
+      "step": 5130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961230134545966e-05,
+      "loss": 0.2263,
+      "step": 5135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961183254418452e-05,
+      "loss": 0.4653,
+      "step": 5140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961136374290938e-05,
+      "loss": 0.4411,
+      "step": 5145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.961089494163424e-05,
+      "loss": 0.7592,
+      "step": 5150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96104261403591e-05,
+      "loss": 0.3478,
+      "step": 5155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9609957339083964e-05,
+      "loss": 0.2249,
+      "step": 5160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9609488537808824e-05,
+      "loss": 0.164,
+      "step": 5165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9609019736533684e-05,
+      "loss": 0.2617,
+      "step": 5170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9608550935258544e-05,
+      "loss": 0.3287,
+      "step": 5175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9608082133983408e-05,
+      "loss": 0.3823,
+      "step": 5180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9607613332708268e-05,
+      "loss": 0.3231,
+      "step": 5185
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9607144531433127e-05,
+      "loss": 0.2896,
+      "step": 5190
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9606675730157987e-05,
+      "loss": 0.4749,
+      "step": 5195
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9606206928882847e-05,
+      "loss": 0.5387,
+      "step": 5200
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960573812760771e-05,
+      "loss": 0.2887,
+      "step": 5205
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960526932633257e-05,
+      "loss": 0.1177,
+      "step": 5210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960480052505743e-05,
+      "loss": 0.2773,
+      "step": 5215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960433172378229e-05,
+      "loss": 0.2782,
+      "step": 5220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960386292250715e-05,
+      "loss": 0.317,
+      "step": 5225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960339412123201e-05,
+      "loss": 0.2609,
+      "step": 5230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960292531995687e-05,
+      "loss": 0.3552,
+      "step": 5235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.960245651868173e-05,
+      "loss": 0.3464,
+      "step": 5240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601987717406594e-05,
+      "loss": 0.5567,
+      "step": 5245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601518916131453e-05,
+      "loss": 0.6603,
+      "step": 5250
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9601050114856313e-05,
+      "loss": 0.4007,
+      "step": 5255
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9600581313581177e-05,
+      "loss": 0.1196,
+      "step": 5260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9600112512306037e-05,
+      "loss": 0.1901,
+      "step": 5265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9599643711030897e-05,
+      "loss": 0.2177,
+      "step": 5270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9599174909755757e-05,
+      "loss": 0.2707,
+      "step": 5275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9598706108480616e-05,
+      "loss": 0.2405,
+      "step": 5280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9598237307205476e-05,
+      "loss": 0.249,
+      "step": 5285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9597768505930336e-05,
+      "loss": 0.3312,
+      "step": 5290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95972997046552e-05,
+      "loss": 0.4266,
+      "step": 5295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959683090338006e-05,
+      "loss": 0.6443,
+      "step": 5300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959636210210492e-05,
+      "loss": 0.373,
+      "step": 5305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959589330082978e-05,
+      "loss": 0.1447,
+      "step": 5310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959542449955464e-05,
+      "loss": 0.2002,
+      "step": 5315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95949556982795e-05,
+      "loss": 0.1767,
+      "step": 5320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9594486897004363e-05,
+      "loss": 0.2046,
+      "step": 5325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9594018095729223e-05,
+      "loss": 0.2941,
+      "step": 5330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9593549294454082e-05,
+      "loss": 0.2813,
+      "step": 5335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9593080493178942e-05,
+      "loss": 0.3738,
+      "step": 5340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9592611691903806e-05,
+      "loss": 0.6161,
+      "step": 5345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9592142890628666e-05,
+      "loss": 0.8597,
+      "step": 5350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9591674089353526e-05,
+      "loss": 0.3059,
+      "step": 5355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9591205288078386e-05,
+      "loss": 0.1577,
+      "step": 5360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590736486803245e-05,
+      "loss": 0.2368,
+      "step": 5365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590267685528105e-05,
+      "loss": 0.2593,
+      "step": 5370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9589798884252965e-05,
+      "loss": 0.3036,
+      "step": 5375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9589330082977825e-05,
+      "loss": 0.3428,
+      "step": 5380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9588861281702685e-05,
+      "loss": 0.3202,
+      "step": 5385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958839248042755e-05,
+      "loss": 0.3252,
+      "step": 5390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958792367915241e-05,
+      "loss": 0.4546,
+      "step": 5395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958745487787727e-05,
+      "loss": 0.7591,
+      "step": 5400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9586986076602132e-05,
+      "loss": 0.3765,
+      "step": 5405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958651727532699e-05,
+      "loss": 0.1274,
+      "step": 5410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958604847405185e-05,
+      "loss": 0.2058,
+      "step": 5415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958557967277671e-05,
+      "loss": 0.3033,
+      "step": 5420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958511087150157e-05,
+      "loss": 0.1879,
+      "step": 5425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958464207022643e-05,
+      "loss": 0.3354,
+      "step": 5430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9584173268951295e-05,
+      "loss": 0.312,
+      "step": 5435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9583704467676155e-05,
+      "loss": 0.3897,
+      "step": 5440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9583235666401015e-05,
+      "loss": 0.5298,
+      "step": 5445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9582766865125875e-05,
+      "loss": 0.8369,
+      "step": 5450
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9582298063850734e-05,
+      "loss": 0.2992,
+      "step": 5455
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9581829262575594e-05,
+      "loss": 0.1595,
+      "step": 5460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9581360461300454e-05,
+      "loss": 0.1288,
+      "step": 5465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9580891660025318e-05,
+      "loss": 0.2173,
+      "step": 5470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9580422858750178e-05,
+      "loss": 0.2598,
+      "step": 5475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9579954057475038e-05,
+      "loss": 0.3008,
+      "step": 5480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.95794852561999e-05,
+      "loss": 0.4085,
+      "step": 5485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957901645492476e-05,
+      "loss": 0.3574,
+      "step": 5490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957854765364962e-05,
+      "loss": 0.5726,
+      "step": 5495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957807885237448e-05,
+      "loss": 0.7328,
+      "step": 5500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957761005109934e-05,
+      "loss": 0.3389,
+      "step": 5505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.95771412498242e-05,
+      "loss": 0.1435,
+      "step": 5510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957667244854906e-05,
+      "loss": 0.1596,
+      "step": 5515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957620364727392e-05,
+      "loss": 0.1642,
+      "step": 5520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957573484599878e-05,
+      "loss": 0.2219,
+      "step": 5525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9575266044723644e-05,
+      "loss": 0.2159,
+      "step": 5530
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9574797243448504e-05,
+      "loss": 0.2886,
+      "step": 5535
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9574328442173363e-05,
+      "loss": 0.2399,
+      "step": 5540
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9573859640898223e-05,
+      "loss": 0.3391,
+      "step": 5545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9573390839623087e-05,
+      "loss": 0.7245,
+      "step": 5550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9572922038347947e-05,
+      "loss": 0.3067,
+      "step": 5555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9572453237072807e-05,
+      "loss": 0.0993,
+      "step": 5560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9571984435797667e-05,
+      "loss": 0.24,
+      "step": 5565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9571515634522526e-05,
+      "loss": 0.1977,
+      "step": 5570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957104683324739e-05,
+      "loss": 0.2716,
+      "step": 5575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957057803197225e-05,
+      "loss": 0.3497,
+      "step": 5580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.957010923069711e-05,
+      "loss": 0.2877,
+      "step": 5585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.956964042942197e-05,
+      "loss": 0.3118,
+      "step": 5590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.956917162814683e-05,
+      "loss": 0.4374,
+      "step": 5595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.956870282687169e-05,
+      "loss": 0.7082,
+      "step": 5600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.956823402559655e-05,
+      "loss": 0.2997,
+      "step": 5605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9567765224321413e-05,
+      "loss": 0.1337,
+      "step": 5610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9567296423046273e-05,
+      "loss": 0.1612,
+      "step": 5615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9566827621771133e-05,
+      "loss": 0.1726,
+      "step": 5620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9566358820495996e-05,
+      "loss": 0.2525,
+      "step": 5625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9565890019220856e-05,
+      "loss": 0.2697,
+      "step": 5630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9565421217945716e-05,
+      "loss": 0.3576,
+      "step": 5635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9564952416670576e-05,
+      "loss": 0.3216,
+      "step": 5640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9564483615395436e-05,
+      "loss": 0.4422,
+      "step": 5645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9564014814120296e-05,
+      "loss": 0.6881,
+      "step": 5650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9563546012845156e-05,
+      "loss": 0.341,
+      "step": 5655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9563077211570015e-05,
+      "loss": 0.2462,
+      "step": 5660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9562608410294875e-05,
+      "loss": 0.1833,
+      "step": 5665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.956213960901974e-05,
+      "loss": 0.1736,
+      "step": 5670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.95616708077446e-05,
+      "loss": 0.3108,
+      "step": 5675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.956120200646946e-05,
+      "loss": 0.1955,
+      "step": 5680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.956073320519432e-05,
+      "loss": 0.3466,
+      "step": 5685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9560264403919182e-05,
+      "loss": 0.3074,
+      "step": 5690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9559795602644042e-05,
+      "loss": 0.4682,
+      "step": 5695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.95593268013689e-05,
+      "loss": 0.6463,
+      "step": 5700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955885800009376e-05,
+      "loss": 0.327,
+      "step": 5705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955838919881862e-05,
+      "loss": 0.17,
+      "step": 5710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9557920397543485e-05,
+      "loss": 0.1442,
+      "step": 5715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9557451596268345e-05,
+      "loss": 0.2012,
+      "step": 5720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9556982794993205e-05,
+      "loss": 0.2722,
+      "step": 5725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9556513993718065e-05,
+      "loss": 0.2928,
+      "step": 5730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9556045192442925e-05,
+      "loss": 0.3275,
+      "step": 5735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9555576391167785e-05,
+      "loss": 0.462,
+      "step": 5740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9555107589892644e-05,
+      "loss": 0.38,
+      "step": 5745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9554638788617504e-05,
+      "loss": 0.7203,
+      "step": 5750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9554169987342368e-05,
+      "loss": 0.3981,
+      "step": 5755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9553701186067228e-05,
+      "loss": 0.1104,
+      "step": 5760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9553232384792088e-05,
+      "loss": 0.2247,
+      "step": 5765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955276358351695e-05,
+      "loss": 0.1969,
+      "step": 5770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955229478224181e-05,
+      "loss": 0.2883,
+      "step": 5775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955182598096667e-05,
+      "loss": 0.2871,
+      "step": 5780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955135717969153e-05,
+      "loss": 0.2452,
+      "step": 5785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955088837841639e-05,
+      "loss": 0.3934,
+      "step": 5790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.955041957714125e-05,
+      "loss": 0.4091,
+      "step": 5795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954995077586611e-05,
+      "loss": 0.7059,
+      "step": 5800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954948197459097e-05,
+      "loss": 0.3207,
+      "step": 5805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9549013173315834e-05,
+      "loss": 0.1067,
+      "step": 5810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9548544372040694e-05,
+      "loss": 0.1733,
+      "step": 5815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9548075570765554e-05,
+      "loss": 0.1885,
+      "step": 5820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9547606769490414e-05,
+      "loss": 0.1855,
+      "step": 5825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9547137968215274e-05,
+      "loss": 0.2796,
+      "step": 5830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9546669166940137e-05,
+      "loss": 0.3381,
+      "step": 5835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9546200365664997e-05,
+      "loss": 0.4255,
+      "step": 5840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9545731564389857e-05,
+      "loss": 0.3635,
+      "step": 5845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9545262763114717e-05,
+      "loss": 0.6507,
+      "step": 5850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954479396183958e-05,
+      "loss": 0.3462,
+      "step": 5855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954432516056444e-05,
+      "loss": 0.1309,
+      "step": 5860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.95438563592893e-05,
+      "loss": 0.1158,
+      "step": 5865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954338755801416e-05,
+      "loss": 0.1644,
+      "step": 5870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954291875673902e-05,
+      "loss": 0.2895,
+      "step": 5875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954244995546388e-05,
+      "loss": 0.3314,
+      "step": 5880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.954198115418874e-05,
+      "loss": 0.2619,
+      "step": 5885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.95415123529136e-05,
+      "loss": 0.5068,
+      "step": 5890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.954104355163846e-05,
+      "loss": 0.3092,
+      "step": 5895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9540574750363323e-05,
+      "loss": 0.6608,
+      "step": 5900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9540105949088183e-05,
+      "loss": 0.3003,
+      "step": 5905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9539637147813043e-05,
+      "loss": 0.1948,
+      "step": 5910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9539168346537906e-05,
+      "loss": 0.2106,
+      "step": 5915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9538699545262766e-05,
+      "loss": 0.1636,
+      "step": 5920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9538230743987626e-05,
+      "loss": 0.2327,
+      "step": 5925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9537761942712486e-05,
+      "loss": 0.3433,
+      "step": 5930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9537293141437346e-05,
+      "loss": 0.2788,
+      "step": 5935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9536824340162206e-05,
+      "loss": 0.3844,
+      "step": 5940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9536355538887066e-05,
+      "loss": 0.5153,
+      "step": 5945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.953588673761193e-05,
+      "loss": 0.8266,
+      "step": 5950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.953541793633679e-05,
+      "loss": 0.3474,
+      "step": 5955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.953494913506165e-05,
+      "loss": 0.1229,
+      "step": 5960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.953448033378651e-05,
+      "loss": 0.2005,
+      "step": 5965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.953401153251137e-05,
+      "loss": 0.1992,
+      "step": 5970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9533542731236232e-05,
+      "loss": 0.2129,
+      "step": 5975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9533073929961092e-05,
+      "loss": 0.2354,
+      "step": 5980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9532605128685952e-05,
+      "loss": 0.3105,
+      "step": 5985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9532136327410812e-05,
+      "loss": 0.3057,
+      "step": 5990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9531667526135675e-05,
+      "loss": 0.4136,
+      "step": 5995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9531198724860535e-05,
+      "loss": 0.6449,
+      "step": 6000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9530729923585395e-05,
+      "loss": 0.2958,
+      "step": 6005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9530261122310255e-05,
+      "loss": 0.1949,
+      "step": 6010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9529792321035115e-05,
+      "loss": 0.2033,
+      "step": 6015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9529323519759975e-05,
+      "loss": 0.1695,
+      "step": 6020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9528854718484835e-05,
+      "loss": 0.2202,
+      "step": 6025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9528385917209695e-05,
+      "loss": 0.4026,
+      "step": 6030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9527917115934555e-05,
+      "loss": 0.2653,
+      "step": 6035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9527448314659418e-05,
+      "loss": 0.4601,
+      "step": 6040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9526979513384278e-05,
+      "loss": 0.4971,
+      "step": 6045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9526510712109138e-05,
+      "loss": 0.6803,
+      "step": 6050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9526041910834e-05,
+      "loss": 0.3679,
+      "step": 6055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.952557310955886e-05,
+      "loss": 0.0853,
+      "step": 6060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.952510430828372e-05,
+      "loss": 0.2384,
+      "step": 6065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.952463550700858e-05,
+      "loss": 0.1462,
+      "step": 6070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.952416670573344e-05,
+      "loss": 0.2165,
+      "step": 6075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.95236979044583e-05,
+      "loss": 0.2811,
+      "step": 6080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9523229103183164e-05,
+      "loss": 0.3491,
+      "step": 6085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9522760301908024e-05,
+      "loss": 0.4301,
+      "step": 6090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9522291500632884e-05,
+      "loss": 0.4114,
+      "step": 6095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9521822699357744e-05,
+      "loss": 0.5665,
+      "step": 6100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9521353898082604e-05,
+      "loss": 0.3065,
+      "step": 6105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9520885096807464e-05,
+      "loss": 0.1459,
+      "step": 6110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9520416295532324e-05,
+      "loss": 0.1675,
+      "step": 6115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9519947494257187e-05,
+      "loss": 0.1782,
+      "step": 6120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9519478692982047e-05,
+      "loss": 0.2657,
+      "step": 6125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9519009891706907e-05,
+      "loss": 0.2721,
+      "step": 6130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951854109043177e-05,
+      "loss": 0.2553,
+      "step": 6135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951807228915663e-05,
+      "loss": 0.3509,
+      "step": 6140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951760348788149e-05,
+      "loss": 0.5497,
+      "step": 6145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951713468660635e-05,
+      "loss": 0.8548,
+      "step": 6150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951666588533121e-05,
+      "loss": 0.3924,
+      "step": 6155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951619708405607e-05,
+      "loss": 0.2242,
+      "step": 6160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951572828278093e-05,
+      "loss": 0.1475,
+      "step": 6165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951525948150579e-05,
+      "loss": 0.2086,
+      "step": 6170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951479068023065e-05,
+      "loss": 0.1798,
+      "step": 6175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9514321878955513e-05,
+      "loss": 0.269,
+      "step": 6180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9513853077680373e-05,
+      "loss": 0.2895,
+      "step": 6185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9513384276405233e-05,
+      "loss": 0.3678,
+      "step": 6190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9512915475130093e-05,
+      "loss": 0.3819,
+      "step": 6195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9512446673854956e-05,
+      "loss": 0.4726,
+      "step": 6200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9511977872579816e-05,
+      "loss": 0.309,
+      "step": 6205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9511509071304676e-05,
+      "loss": 0.1438,
+      "step": 6210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9511040270029536e-05,
+      "loss": 0.2037,
+      "step": 6215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9510571468754396e-05,
+      "loss": 0.2466,
+      "step": 6220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.951010266747926e-05,
+      "loss": 0.2742,
+      "step": 6225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950963386620412e-05,
+      "loss": 0.26,
+      "step": 6230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950916506492898e-05,
+      "loss": 0.2899,
+      "step": 6235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950869626365384e-05,
+      "loss": 0.4501,
+      "step": 6240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.95082274623787e-05,
+      "loss": 0.5071,
+      "step": 6245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950775866110356e-05,
+      "loss": 0.6465,
+      "step": 6250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950728985982842e-05,
+      "loss": 0.3001,
+      "step": 6255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.950682105855328e-05,
+      "loss": 0.136,
+      "step": 6260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9506352257278142e-05,
+      "loss": 0.1448,
+      "step": 6265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9505883456003002e-05,
+      "loss": 0.2306,
+      "step": 6270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9505414654727862e-05,
+      "loss": 0.1849,
+      "step": 6275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9504945853452725e-05,
+      "loss": 0.217,
+      "step": 6280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9504477052177585e-05,
+      "loss": 0.3257,
+      "step": 6285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9504008250902445e-05,
+      "loss": 0.4219,
+      "step": 6290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9503539449627305e-05,
+      "loss": 0.4743,
+      "step": 6295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9503070648352165e-05,
+      "loss": 0.6075,
+      "step": 6300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9502601847077025e-05,
+      "loss": 0.3695,
+      "step": 6305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9502133045801885e-05,
+      "loss": 0.1424,
+      "step": 6310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9501664244526745e-05,
+      "loss": 0.2252,
+      "step": 6315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9501195443251608e-05,
+      "loss": 0.2399,
+      "step": 6320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9500726641976468e-05,
+      "loss": 0.247,
+      "step": 6325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9500257840701328e-05,
+      "loss": 0.3604,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9499789039426188e-05,
+      "loss": 0.4467,
+      "step": 6335
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9499320238151048e-05,
+      "loss": 0.4687,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949885143687591e-05,
+      "loss": 0.3686,
+      "step": 6345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949838263560077e-05,
+      "loss": 0.8263,
+      "step": 6350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949791383432563e-05,
+      "loss": 0.3793,
+      "step": 6355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949744503305049e-05,
+      "loss": 0.1265,
+      "step": 6360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9496976231775354e-05,
+      "loss": 0.1601,
+      "step": 6365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9496507430500214e-05,
+      "loss": 0.2208,
+      "step": 6370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9496038629225074e-05,
+      "loss": 0.2892,
+      "step": 6375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9495569827949934e-05,
+      "loss": 0.2399,
+      "step": 6380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9495101026674794e-05,
+      "loss": 0.4323,
+      "step": 6385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9494632225399654e-05,
+      "loss": 0.4213,
+      "step": 6390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9494163424124514e-05,
+      "loss": 0.6043,
+      "step": 6395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9493694622849374e-05,
+      "loss": 0.8731,
+      "step": 6400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9493225821574237e-05,
+      "loss": 0.2651,
+      "step": 6405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9492757020299097e-05,
+      "loss": 0.1896,
+      "step": 6410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9492288219023957e-05,
+      "loss": 0.2098,
+      "step": 6415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9491819417748817e-05,
+      "loss": 0.2555,
+      "step": 6420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949135061647368e-05,
+      "loss": 0.2872,
+      "step": 6425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.949088181519854e-05,
+      "loss": 0.306,
+      "step": 6430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.94904130139234e-05,
+      "loss": 0.3411,
+      "step": 6435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948994421264826e-05,
+      "loss": 0.3097,
+      "step": 6440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948947541137312e-05,
+      "loss": 0.4879,
+      "step": 6445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948900661009798e-05,
+      "loss": 0.6127,
+      "step": 6450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948853780882284e-05,
+      "loss": 0.3599,
+      "step": 6455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9488069007547703e-05,
+      "loss": 0.1394,
+      "step": 6460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9487600206272563e-05,
+      "loss": 0.2307,
+      "step": 6465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9487131404997423e-05,
+      "loss": 0.2525,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9486662603722283e-05,
+      "loss": 0.1769,
+      "step": 6475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9486193802447143e-05,
+      "loss": 0.303,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9485725001172006e-05,
+      "loss": 0.3339,
+      "step": 6485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9485256199896866e-05,
+      "loss": 0.3536,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9484787398621726e-05,
+      "loss": 0.4037,
+      "step": 6495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9484318597346586e-05,
+      "loss": 0.5179,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948384979607145e-05,
+      "loss": 0.2462,
+      "step": 6505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948338099479631e-05,
+      "loss": 0.1519,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948291219352117e-05,
+      "loss": 0.2413,
+      "step": 6515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948244339224603e-05,
+      "loss": 0.1308,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948197459097089e-05,
+      "loss": 0.3107,
+      "step": 6525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948150578969575e-05,
+      "loss": 0.3151,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.948103698842061e-05,
+      "loss": 0.3446,
+      "step": 6535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.948056818714547e-05,
+      "loss": 0.3298,
+      "step": 6540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.948009938587033e-05,
+      "loss": 0.3325,
+      "step": 6545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9479630584595192e-05,
+      "loss": 0.8412,
+      "step": 6550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9479161783320052e-05,
+      "loss": 0.2412,
+      "step": 6555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9478692982044912e-05,
+      "loss": 0.1557,
+      "step": 6560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9478224180769775e-05,
+      "loss": 0.2324,
+      "step": 6565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9477755379494635e-05,
+      "loss": 0.2358,
+      "step": 6570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9477286578219495e-05,
+      "loss": 0.2939,
+      "step": 6575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9476817776944355e-05,
+      "loss": 0.2902,
+      "step": 6580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9476348975669215e-05,
+      "loss": 0.271,
+      "step": 6585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9475880174394075e-05,
+      "loss": 0.2787,
+      "step": 6590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9475411373118935e-05,
+      "loss": 0.4382,
+      "step": 6595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9474942571843798e-05,
+      "loss": 0.5563,
+      "step": 6600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9474473770568658e-05,
+      "loss": 0.3581,
+      "step": 6605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9474004969293518e-05,
+      "loss": 0.1465,
+      "step": 6610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9473536168018378e-05,
+      "loss": 0.1957,
+      "step": 6615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9473067366743238e-05,
+      "loss": 0.2329,
+      "step": 6620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9472598565468098e-05,
+      "loss": 0.2191,
+      "step": 6625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.947212976419296e-05,
+      "loss": 0.2638,
+      "step": 6630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.947166096291782e-05,
+      "loss": 0.3851,
+      "step": 6635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.947119216164268e-05,
+      "loss": 0.421,
+      "step": 6640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9470723360367544e-05,
+      "loss": 0.3695,
+      "step": 6645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9470254559092404e-05,
+      "loss": 0.7012,
+      "step": 6650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9469785757817264e-05,
+      "loss": 0.3262,
+      "step": 6655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9469316956542124e-05,
+      "loss": 0.0764,
+      "step": 6660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9468848155266984e-05,
+      "loss": 0.2014,
+      "step": 6665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9468379353991844e-05,
+      "loss": 0.2009,
+      "step": 6670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9467910552716704e-05,
+      "loss": 0.2012,
+      "step": 6675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9467441751441564e-05,
+      "loss": 0.2679,
+      "step": 6680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9466972950166424e-05,
+      "loss": 0.2762,
+      "step": 6685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9466504148891287e-05,
+      "loss": 0.4076,
+      "step": 6690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9466035347616147e-05,
+      "loss": 0.3921,
+      "step": 6695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9465566546341007e-05,
+      "loss": 0.6968,
+      "step": 6700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9465097745065867e-05,
+      "loss": 0.3596,
+      "step": 6705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.946462894379073e-05,
+      "loss": 0.0913,
+      "step": 6710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.946416014251559e-05,
+      "loss": 0.1434,
+      "step": 6715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.946369134124045e-05,
+      "loss": 0.1815,
+      "step": 6720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.946322253996531e-05,
+      "loss": 0.2293,
+      "step": 6725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.946275373869017e-05,
+      "loss": 0.2578,
+      "step": 6730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9462284937415033e-05,
+      "loss": 0.3253,
+      "step": 6735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9461816136139893e-05,
+      "loss": 0.4554,
+      "step": 6740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9461347334864753e-05,
+      "loss": 0.433,
+      "step": 6745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9460878533589613e-05,
+      "loss": 0.7031,
+      "step": 6750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9460409732314473e-05,
+      "loss": 0.3681,
+      "step": 6755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9459940931039333e-05,
+      "loss": 0.1622,
+      "step": 6760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9459472129764193e-05,
+      "loss": 0.1286,
+      "step": 6765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9459003328489053e-05,
+      "loss": 0.2374,
+      "step": 6770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9458534527213916e-05,
+      "loss": 0.2842,
+      "step": 6775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9458065725938776e-05,
+      "loss": 0.2651,
+      "step": 6780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9457596924663636e-05,
+      "loss": 0.402,
+      "step": 6785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.94571281233885e-05,
+      "loss": 0.3272,
+      "step": 6790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945665932211336e-05,
+      "loss": 0.4942,
+      "step": 6795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945619052083822e-05,
+      "loss": 0.571,
+      "step": 6800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945572171956308e-05,
+      "loss": 0.3568,
+      "step": 6805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945525291828794e-05,
+      "loss": 0.1872,
+      "step": 6810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.94547841170128e-05,
+      "loss": 0.2116,
+      "step": 6815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945431531573766e-05,
+      "loss": 0.1498,
+      "step": 6820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945384651446252e-05,
+      "loss": 0.2026,
+      "step": 6825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9453377713187382e-05,
+      "loss": 0.2388,
+      "step": 6830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9452908911912242e-05,
+      "loss": 0.3062,
+      "step": 6835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9452440110637102e-05,
+      "loss": 0.4513,
+      "step": 6840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9451971309361962e-05,
+      "loss": 0.4118,
+      "step": 6845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9451502508086822e-05,
+      "loss": 0.5351,
+      "step": 6850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9451033706811685e-05,
+      "loss": 0.2974,
+      "step": 6855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9450564905536545e-05,
+      "loss": 0.1387,
+      "step": 6860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9450096104261405e-05,
+      "loss": 0.1921,
+      "step": 6865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9449627302986265e-05,
+      "loss": 0.2271,
+      "step": 6870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944915850171113e-05,
+      "loss": 0.2264,
+      "step": 6875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944868970043599e-05,
+      "loss": 0.3337,
+      "step": 6880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9448220899160848e-05,
+      "loss": 0.34,
+      "step": 6885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9447752097885708e-05,
+      "loss": 0.4425,
+      "step": 6890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9447283296610568e-05,
+      "loss": 0.5421,
+      "step": 6895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9446814495335428e-05,
+      "loss": 0.5687,
+      "step": 6900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9446345694060288e-05,
+      "loss": 0.4261,
+      "step": 6905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9445876892785148e-05,
+      "loss": 0.0878,
+      "step": 6910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944540809151001e-05,
+      "loss": 0.1664,
+      "step": 6915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944493929023487e-05,
+      "loss": 0.173,
+      "step": 6920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944447048895973e-05,
+      "loss": 0.1802,
+      "step": 6925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.944400168768459e-05,
+      "loss": 0.3044,
+      "step": 6930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9443532886409454e-05,
+      "loss": 0.2697,
+      "step": 6935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9443064085134314e-05,
+      "loss": 0.3823,
+      "step": 6940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9442595283859174e-05,
+      "loss": 0.5836,
+      "step": 6945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9442126482584034e-05,
+      "loss": 0.8075,
+      "step": 6950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9441657681308894e-05,
+      "loss": 0.3048,
+      "step": 6955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9441188880033754e-05,
+      "loss": 0.178,
+      "step": 6960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9440720078758614e-05,
+      "loss": 0.142,
+      "step": 6965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9440251277483477e-05,
+      "loss": 0.1452,
+      "step": 6970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9439782476208337e-05,
+      "loss": 0.2158,
+      "step": 6975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9439313674933197e-05,
+      "loss": 0.2264,
+      "step": 6980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9438844873658057e-05,
+      "loss": 0.3913,
+      "step": 6985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9438376072382917e-05,
+      "loss": 0.3618,
+      "step": 6990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.943790727110778e-05,
+      "loss": 0.4981,
+      "step": 6995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.943743846983264e-05,
+      "loss": 0.5207,
+      "step": 7000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.94369696685575e-05,
+      "loss": 0.2584,
+      "step": 7005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.943650086728236e-05,
+      "loss": 0.1643,
+      "step": 7010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9436032066007223e-05,
+      "loss": 0.162,
+      "step": 7015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9435563264732083e-05,
+      "loss": 0.2558,
+      "step": 7020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9435094463456943e-05,
+      "loss": 0.24,
+      "step": 7025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9434625662181803e-05,
+      "loss": 0.2614,
+      "step": 7030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9434156860906663e-05,
+      "loss": 0.2827,
+      "step": 7035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9433688059631523e-05,
+      "loss": 0.313,
+      "step": 7040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9433219258356383e-05,
+      "loss": 0.5073,
+      "step": 7045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9432750457081243e-05,
+      "loss": 0.6594,
+      "step": 7050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9432281655806103e-05,
+      "loss": 0.2966,
+      "step": 7055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9431812854530966e-05,
+      "loss": 0.1413,
+      "step": 7060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9431344053255826e-05,
+      "loss": 0.1812,
+      "step": 7065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9430875251980686e-05,
+      "loss": 0.1497,
+      "step": 7070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.943040645070555e-05,
+      "loss": 0.1938,
+      "step": 7075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942993764943041e-05,
+      "loss": 0.3134,
+      "step": 7080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942946884815527e-05,
+      "loss": 0.2142,
+      "step": 7085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942900004688013e-05,
+      "loss": 0.4456,
+      "step": 7090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942853124560499e-05,
+      "loss": 0.3444,
+      "step": 7095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942806244432985e-05,
+      "loss": 0.8402,
+      "step": 7100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942759364305471e-05,
+      "loss": 0.3128,
+      "step": 7105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9427124841779572e-05,
+      "loss": 0.1323,
+      "step": 7110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9426656040504432e-05,
+      "loss": 0.1779,
+      "step": 7115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9426187239229292e-05,
+      "loss": 0.1896,
+      "step": 7120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9425718437954152e-05,
+      "loss": 0.2415,
+      "step": 7125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9425249636679012e-05,
+      "loss": 0.2888,
+      "step": 7130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9424780835403872e-05,
+      "loss": 0.2641,
+      "step": 7135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9424312034128735e-05,
+      "loss": 0.3465,
+      "step": 7140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9423843232853595e-05,
+      "loss": 0.4948,
+      "step": 7145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9423374431578455e-05,
+      "loss": 0.6909,
+      "step": 7150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942290563030332e-05,
+      "loss": 0.3321,
+      "step": 7155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942243682902818e-05,
+      "loss": 0.1793,
+      "step": 7160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942196802775304e-05,
+      "loss": 0.1224,
+      "step": 7165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.94214992264779e-05,
+      "loss": 0.2602,
+      "step": 7170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9421030425202758e-05,
+      "loss": 0.1669,
+      "step": 7175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9420561623927618e-05,
+      "loss": 0.1957,
+      "step": 7180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9420092822652478e-05,
+      "loss": 0.3104,
+      "step": 7185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9419624021377338e-05,
+      "loss": 0.2715,
+      "step": 7190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9419155220102198e-05,
+      "loss": 0.3714,
+      "step": 7195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941868641882706e-05,
+      "loss": 0.6902,
+      "step": 7200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941821761755192e-05,
+      "loss": 0.2193,
+      "step": 7205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941774881627678e-05,
+      "loss": 0.1142,
+      "step": 7210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941728001500164e-05,
+      "loss": 0.1894,
+      "step": 7215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9416811213726504e-05,
+      "loss": 0.2033,
+      "step": 7220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9416342412451364e-05,
+      "loss": 0.251,
+      "step": 7225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9415873611176224e-05,
+      "loss": 0.3118,
+      "step": 7230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9415404809901084e-05,
+      "loss": 0.29,
+      "step": 7235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9414936008625944e-05,
+      "loss": 0.331,
+      "step": 7240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9414467207350804e-05,
+      "loss": 0.441,
+      "step": 7245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9413998406075667e-05,
+      "loss": 0.6936,
+      "step": 7250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9413529604800527e-05,
+      "loss": 0.2578,
+      "step": 7255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9413060803525387e-05,
+      "loss": 0.1108,
+      "step": 7260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9412592002250247e-05,
+      "loss": 0.1718,
+      "step": 7265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9412123200975107e-05,
+      "loss": 0.2078,
+      "step": 7270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9411654399699967e-05,
+      "loss": 0.237,
+      "step": 7275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9411185598424827e-05,
+      "loss": 0.2431,
+      "step": 7280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941071679714969e-05,
+      "loss": 0.2729,
+      "step": 7285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.941024799587455e-05,
+      "loss": 0.3672,
+      "step": 7290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940977919459941e-05,
+      "loss": 0.5002,
+      "step": 7295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9409310393324274e-05,
+      "loss": 0.7313,
+      "step": 7300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9408841592049134e-05,
+      "loss": 0.328,
+      "step": 7305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9408372790773993e-05,
+      "loss": 0.2048,
+      "step": 7310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9407903989498853e-05,
+      "loss": 0.2032,
+      "step": 7315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9407435188223713e-05,
+      "loss": 0.2235,
+      "step": 7320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9406966386948573e-05,
+      "loss": 0.198,
+      "step": 7325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9406497585673433e-05,
+      "loss": 0.2556,
+      "step": 7330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9406028784398293e-05,
+      "loss": 0.358,
+      "step": 7335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9405559983123156e-05,
+      "loss": 0.4997,
+      "step": 7340
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9405091181848016e-05,
+      "loss": 0.4498,
+      "step": 7345
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9404622380572876e-05,
+      "loss": 0.5937,
+      "step": 7350
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9404153579297736e-05,
+      "loss": 0.2874,
+      "step": 7355
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.94036847780226e-05,
+      "loss": 0.1371,
+      "step": 7360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940321597674746e-05,
+      "loss": 0.1512,
+      "step": 7365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940274717547232e-05,
+      "loss": 0.1584,
+      "step": 7370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940227837419718e-05,
+      "loss": 0.2193,
+      "step": 7375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.940180957292204e-05,
+      "loss": 0.2825,
+      "step": 7380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9401340771646903e-05,
+      "loss": 0.3116,
+      "step": 7385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9400871970371763e-05,
+      "loss": 0.3652,
+      "step": 7390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9400403169096622e-05,
+      "loss": 0.4405,
+      "step": 7395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9399934367821482e-05,
+      "loss": 0.6158,
+      "step": 7400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9399465566546342e-05,
+      "loss": 0.3158,
+      "step": 7405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9398996765271202e-05,
+      "loss": 0.0907,
+      "step": 7410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9398527963996062e-05,
+      "loss": 0.3189,
+      "step": 7415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9398059162720922e-05,
+      "loss": 0.17,
+      "step": 7420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9397590361445785e-05,
+      "loss": 0.1948,
+      "step": 7425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9397121560170645e-05,
+      "loss": 0.2213,
+      "step": 7430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9396652758895505e-05,
+      "loss": 0.2708,
+      "step": 7435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939618395762037e-05,
+      "loss": 0.3346,
+      "step": 7440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939571515634523e-05,
+      "loss": 0.3433,
+      "step": 7445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939524635507009e-05,
+      "loss": 0.6586,
+      "step": 7450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939477755379495e-05,
+      "loss": 0.2602,
+      "step": 7455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939430875251981e-05,
+      "loss": 0.1247,
+      "step": 7460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939383995124467e-05,
+      "loss": 0.1999,
+      "step": 7465
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9393371149969528e-05,
+      "loss": 0.2082,
+      "step": 7470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9392902348694388e-05,
+      "loss": 0.1804,
+      "step": 7475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939243354741925e-05,
+      "loss": 0.3072,
+      "step": 7480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939196474614411e-05,
+      "loss": 0.2737,
+      "step": 7485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939149594486897e-05,
+      "loss": 0.2939,
+      "step": 7490
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939102714359383e-05,
+      "loss": 0.3612,
+      "step": 7495
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.939055834231869e-05,
+      "loss": 0.591,
+      "step": 7500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9390089541043555e-05,
+      "loss": 0.2915,
+      "step": 7505
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9389620739768415e-05,
+      "loss": 0.1465,
+      "step": 7510
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9389151938493274e-05,
+      "loss": 0.1969,
+      "step": 7515
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9388683137218134e-05,
+      "loss": 0.203,
+      "step": 7520
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9388214335942998e-05,
+      "loss": 0.1947,
+      "step": 7525
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9387745534667858e-05,
+      "loss": 0.2782,
+      "step": 7530
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9387276733392718e-05,
+      "loss": 0.2536,
+      "step": 7535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9386807932117578e-05,
+      "loss": 0.2968,
+      "step": 7540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9386339130842437e-05,
+      "loss": 0.4721,
+      "step": 7545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9385870329567297e-05,
+      "loss": 0.6284,
+      "step": 7550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9385401528292157e-05,
+      "loss": 0.2729,
+      "step": 7555
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9384932727017017e-05,
+      "loss": 0.1275,
+      "step": 7560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9384463925741877e-05,
+      "loss": 0.2018,
+      "step": 7565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.938399512446674e-05,
+      "loss": 0.1743,
+      "step": 7570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.93835263231916e-05,
+      "loss": 0.354,
+      "step": 7575
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.938305752191646e-05,
+      "loss": 0.1568,
+      "step": 7580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9382588720641324e-05,
+      "loss": 0.2309,
+      "step": 7585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9382119919366184e-05,
+      "loss": 0.3072,
+      "step": 7590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9381651118091044e-05,
+      "loss": 0.3901,
+      "step": 7595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9381182316815903e-05,
+      "loss": 0.6689,
+      "step": 7600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9380713515540763e-05,
+      "loss": 0.3356,
+      "step": 7605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9380244714265623e-05,
+      "loss": 0.0993,
+      "step": 7610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9379775912990483e-05,
+      "loss": 0.1436,
+      "step": 7615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9379307111715347e-05,
+      "loss": 0.1633,
+      "step": 7620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9378838310440207e-05,
+      "loss": 0.2219,
+      "step": 7625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9378369509165066e-05,
+      "loss": 0.1863,
+      "step": 7630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9377900707889926e-05,
+      "loss": 0.382,
+      "step": 7635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9377431906614786e-05,
+      "loss": 0.3374,
+      "step": 7640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9376963105339646e-05,
+      "loss": 0.3538,
+      "step": 7645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.937649430406451e-05,
+      "loss": 0.5143,
+      "step": 7650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.937602550278937e-05,
+      "loss": 0.3543,
+      "step": 7655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.937555670151423e-05,
+      "loss": 0.1521,
+      "step": 7660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9375087900239093e-05,
+      "loss": 0.2513,
+      "step": 7665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9374619098963953e-05,
+      "loss": 0.1515,
+      "step": 7670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9374150297688813e-05,
+      "loss": 0.2925,
+      "step": 7675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9373681496413673e-05,
+      "loss": 0.3807,
+      "step": 7680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9373212695138533e-05,
+      "loss": 0.381,
+      "step": 7685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9372743893863392e-05,
+      "loss": 0.2791,
+      "step": 7690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9372275092588252e-05,
+      "loss": 0.4091,
+      "step": 7695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9371806291313112e-05,
+      "loss": 0.7086,
+      "step": 7700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9371337490037972e-05,
+      "loss": 0.2184,
+      "step": 7705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9370868688762836e-05,
+      "loss": 0.1839,
+      "step": 7710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9370399887487696e-05,
+      "loss": 0.1395,
+      "step": 7715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9369931086212555e-05,
+      "loss": 0.1088,
+      "step": 7720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9369462284937415e-05,
+      "loss": 0.1739,
+      "step": 7725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936899348366228e-05,
+      "loss": 0.2314,
+      "step": 7730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936852468238714e-05,
+      "loss": 0.3432,
+      "step": 7735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9368055881112e-05,
+      "loss": 0.3118,
+      "step": 7740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936758707983686e-05,
+      "loss": 0.4382,
+      "step": 7745
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936711827856172e-05,
+      "loss": 0.4952,
+      "step": 7750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936664947728658e-05,
+      "loss": 0.3685,
+      "step": 7755
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936618067601144e-05,
+      "loss": 0.13,
+      "step": 7760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.93657118747363e-05,
+      "loss": 0.1563,
+      "step": 7765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936524307346116e-05,
+      "loss": 0.2089,
+      "step": 7770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936477427218602e-05,
+      "loss": 0.2659,
+      "step": 7775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936430547091088e-05,
+      "loss": 0.1983,
+      "step": 7780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.936383666963574e-05,
+      "loss": 0.2818,
+      "step": 7785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9363367868360605e-05,
+      "loss": 0.418,
+      "step": 7790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9362899067085465e-05,
+      "loss": 0.4263,
+      "step": 7795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9362430265810325e-05,
+      "loss": 0.6848,
+      "step": 7800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9361961464535184e-05,
+      "loss": 0.3445,
+      "step": 7805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9361492663260048e-05,
+      "loss": 0.1231,
+      "step": 7810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9361023861984908e-05,
+      "loss": 0.1814,
+      "step": 7815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9360555060709768e-05,
+      "loss": 0.2129,
+      "step": 7820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9360086259434628e-05,
+      "loss": 0.185,
+      "step": 7825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9359617458159488e-05,
+      "loss": 0.3591,
+      "step": 7830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9359148656884347e-05,
+      "loss": 0.4372,
+      "step": 7835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9358679855609207e-05,
+      "loss": 0.3435,
+      "step": 7840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9358211054334067e-05,
+      "loss": 0.3974,
+      "step": 7845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.935774225305893e-05,
+      "loss": 0.6837,
+      "step": 7850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.935727345178379e-05,
+      "loss": 0.3493,
+      "step": 7855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.935680465050865e-05,
+      "loss": 0.1234,
+      "step": 7860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.935633584923351e-05,
+      "loss": 0.1565,
+      "step": 7865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9355867047958374e-05,
+      "loss": 0.1613,
+      "step": 7870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9355398246683234e-05,
+      "loss": 0.2633,
+      "step": 7875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9354929445408094e-05,
+      "loss": 0.1702,
+      "step": 7880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9354460644132954e-05,
+      "loss": 0.3533,
+      "step": 7885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9353991842857814e-05,
+      "loss": 0.2938,
+      "step": 7890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9353523041582673e-05,
+      "loss": 0.3936,
+      "step": 7895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9353054240307537e-05,
+      "loss": 0.5828,
+      "step": 7900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9352585439032397e-05,
+      "loss": 0.2816,
+      "step": 7905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9352116637757257e-05,
+      "loss": 0.1281,
+      "step": 7910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9351647836482117e-05,
+      "loss": 0.1108,
+      "step": 7915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9351179035206977e-05,
+      "loss": 0.1505,
+      "step": 7920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9350710233931836e-05,
+      "loss": 0.2456,
+      "step": 7925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9350241432656696e-05,
+      "loss": 0.2419,
+      "step": 7930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934977263138156e-05,
+      "loss": 0.2355,
+      "step": 7935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934930383010642e-05,
+      "loss": 0.3848,
+      "step": 7940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934883502883128e-05,
+      "loss": 0.508,
+      "step": 7945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9348366227556143e-05,
+      "loss": 0.688,
+      "step": 7950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9347897426281003e-05,
+      "loss": 0.2389,
+      "step": 7955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9347428625005863e-05,
+      "loss": 0.097,
+      "step": 7960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9346959823730723e-05,
+      "loss": 0.1886,
+      "step": 7965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9346491022455583e-05,
+      "loss": 0.2037,
+      "step": 7970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9346022221180443e-05,
+      "loss": 0.1939,
+      "step": 7975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9345553419905302e-05,
+      "loss": 0.2666,
+      "step": 7980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9345084618630162e-05,
+      "loss": 0.2793,
+      "step": 7985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9344615817355026e-05,
+      "loss": 0.3058,
+      "step": 7990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9344147016079886e-05,
+      "loss": 0.3803,
+      "step": 7995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9343678214804746e-05,
+      "loss": 0.7639,
+      "step": 8000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9343209413529606e-05,
+      "loss": 0.3895,
+      "step": 8005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9342740612254465e-05,
+      "loss": 0.075,
+      "step": 8010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934227181097933e-05,
+      "loss": 0.1631,
+      "step": 8015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934180300970419e-05,
+      "loss": 0.2211,
+      "step": 8020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934133420842905e-05,
+      "loss": 0.1265,
+      "step": 8025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.934086540715391e-05,
+      "loss": 0.2052,
+      "step": 8030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9340396605878772e-05,
+      "loss": 0.2435,
+      "step": 8035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9339927804603632e-05,
+      "loss": 0.3957,
+      "step": 8040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9339459003328492e-05,
+      "loss": 0.3356,
+      "step": 8045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9338990202053352e-05,
+      "loss": 0.7917,
+      "step": 8050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933852140077821e-05,
+      "loss": 0.3205,
+      "step": 8055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933805259950307e-05,
+      "loss": 0.1124,
+      "step": 8060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933758379822793e-05,
+      "loss": 0.1538,
+      "step": 8065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933711499695279e-05,
+      "loss": 0.2043,
+      "step": 8070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933664619567765e-05,
+      "loss": 0.1682,
+      "step": 8075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9336177394402515e-05,
+      "loss": 0.2458,
+      "step": 8080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9335708593127375e-05,
+      "loss": 0.402,
+      "step": 8085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9335239791852235e-05,
+      "loss": 0.4126,
+      "step": 8090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9334770990577098e-05,
+      "loss": 0.4594,
+      "step": 8095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9334302189301958e-05,
+      "loss": 0.6692,
+      "step": 8100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9333833388026818e-05,
+      "loss": 0.3356,
+      "step": 8105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9333364586751678e-05,
+      "loss": 0.1183,
+      "step": 8110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9332895785476538e-05,
+      "loss": 0.22,
+      "step": 8115
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9332426984201398e-05,
+      "loss": 0.1996,
+      "step": 8120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9331958182926257e-05,
+      "loss": 0.273,
+      "step": 8125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933148938165112e-05,
+      "loss": 0.2092,
+      "step": 8130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933102058037598e-05,
+      "loss": 0.2448,
+      "step": 8135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.933055177910084e-05,
+      "loss": 0.2191,
+      "step": 8140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.93300829778257e-05,
+      "loss": 0.4742,
+      "step": 8145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932961417655056e-05,
+      "loss": 0.8307,
+      "step": 8150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932914537527542e-05,
+      "loss": 0.2751,
+      "step": 8155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9328676574000284e-05,
+      "loss": 0.176,
+      "step": 8160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9328207772725144e-05,
+      "loss": 0.1733,
+      "step": 8165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9327738971450004e-05,
+      "loss": 0.2006,
+      "step": 8170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9327270170174867e-05,
+      "loss": 0.2388,
+      "step": 8175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9326801368899727e-05,
+      "loss": 0.3035,
+      "step": 8180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9326332567624587e-05,
+      "loss": 0.2985,
+      "step": 8185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9325863766349447e-05,
+      "loss": 0.3295,
+      "step": 8190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9325394965074307e-05,
+      "loss": 0.4557,
+      "step": 8195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9324926163799167e-05,
+      "loss": 0.5938,
+      "step": 8200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9324457362524027e-05,
+      "loss": 0.3562,
+      "step": 8205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9323988561248887e-05,
+      "loss": 0.1529,
+      "step": 8210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9323519759973746e-05,
+      "loss": 0.1482,
+      "step": 8215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932305095869861e-05,
+      "loss": 0.2249,
+      "step": 8220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932258215742347e-05,
+      "loss": 0.2227,
+      "step": 8225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932211335614833e-05,
+      "loss": 0.2618,
+      "step": 8230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.932164455487319e-05,
+      "loss": 0.2611,
+      "step": 8235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9321175753598053e-05,
+      "loss": 0.2406,
+      "step": 8240
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9320706952322913e-05,
+      "loss": 0.3629,
+      "step": 8245
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9320238151047773e-05,
+      "loss": 0.4992,
+      "step": 8250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9319769349772633e-05,
+      "loss": 0.3151,
+      "step": 8255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9319300548497493e-05,
+      "loss": 0.1345,
+      "step": 8260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9318831747222353e-05,
+      "loss": 0.1521,
+      "step": 8265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9318362945947216e-05,
+      "loss": 0.1733,
+      "step": 8270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9317894144672076e-05,
+      "loss": 0.2129,
+      "step": 8275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9317425343396936e-05,
+      "loss": 0.2378,
+      "step": 8280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9316956542121796e-05,
+      "loss": 0.2658,
+      "step": 8285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9316487740846656e-05,
+      "loss": 0.287,
+      "step": 8290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9316018939571516e-05,
+      "loss": 0.3836,
+      "step": 8295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931555013829638e-05,
+      "loss": 0.6966,
+      "step": 8300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931508133702124e-05,
+      "loss": 0.3483,
+      "step": 8305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.93146125357461e-05,
+      "loss": 0.1619,
+      "step": 8310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931414373447096e-05,
+      "loss": 0.1387,
+      "step": 8315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9313674933195822e-05,
+      "loss": 0.1587,
+      "step": 8320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9313206131920682e-05,
+      "loss": 0.1447,
+      "step": 8325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9312737330645542e-05,
+      "loss": 0.2997,
+      "step": 8330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9312268529370402e-05,
+      "loss": 0.2873,
+      "step": 8335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9311799728095262e-05,
+      "loss": 0.3927,
+      "step": 8340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931133092682012e-05,
+      "loss": 0.3392,
+      "step": 8345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931086212554498e-05,
+      "loss": 0.3994,
+      "step": 8350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.931039332426984e-05,
+      "loss": 0.2752,
+      "step": 8355
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9309924522994705e-05,
+      "loss": 0.1263,
+      "step": 8360
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9309455721719565e-05,
+      "loss": 0.1493,
+      "step": 8365
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9308986920444425e-05,
+      "loss": 0.19,
+      "step": 8370
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9308518119169285e-05,
+      "loss": 0.2094,
+      "step": 8375
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9308049317894148e-05,
+      "loss": 0.3174,
+      "step": 8380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9307580516619008e-05,
+      "loss": 0.2557,
+      "step": 8385
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9307111715343868e-05,
+      "loss": 0.4019,
+      "step": 8390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9306642914068728e-05,
+      "loss": 0.3886,
+      "step": 8395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9306174112793588e-05,
+      "loss": 0.5243,
+      "step": 8400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9305705311518448e-05,
+      "loss": 0.2447,
+      "step": 8405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930523651024331e-05,
+      "loss": 0.1114,
+      "step": 8410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930476770896817e-05,
+      "loss": 0.1295,
+      "step": 8415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930429890769303e-05,
+      "loss": 0.2123,
+      "step": 8420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930383010641789e-05,
+      "loss": 0.2388,
+      "step": 8425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930336130514275e-05,
+      "loss": 0.194,
+      "step": 8430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930289250386761e-05,
+      "loss": 0.2023,
+      "step": 8435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.930242370259247e-05,
+      "loss": 0.3366,
+      "step": 8440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9301954901317334e-05,
+      "loss": 0.4688,
+      "step": 8445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9301486100042194e-05,
+      "loss": 0.5275,
+      "step": 8450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9301017298767054e-05,
+      "loss": 0.198,
+      "step": 8455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9300548497491917e-05,
+      "loss": 0.1555,
+      "step": 8460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9300079696216777e-05,
+      "loss": 0.1445,
+      "step": 8465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9299610894941637e-05,
+      "loss": 0.1536,
+      "step": 8470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9299142093666497e-05,
+      "loss": 0.2828,
+      "step": 8475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9298673292391357e-05,
+      "loss": 0.2771,
+      "step": 8480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9298204491116217e-05,
+      "loss": 0.2566,
+      "step": 8485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9297735689841077e-05,
+      "loss": 0.2707,
+      "step": 8490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9297266888565937e-05,
+      "loss": 0.3241,
+      "step": 8495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.92967980872908e-05,
+      "loss": 0.7655,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.929632928601566e-05,
+      "loss": 0.2513,
+      "step": 8505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.929586048474052e-05,
+      "loss": 0.1115,
+      "step": 8510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.929539168346538e-05,
+      "loss": 0.1718,
+      "step": 8515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.929492288219024e-05,
+      "loss": 0.1965,
+      "step": 8520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9294454080915103e-05,
+      "loss": 0.2988,
+      "step": 8525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9293985279639963e-05,
+      "loss": 0.243,
+      "step": 8530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9293516478364823e-05,
+      "loss": 0.2191,
+      "step": 8535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9293047677089683e-05,
+      "loss": 0.3439,
+      "step": 8540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9292578875814543e-05,
+      "loss": 0.422,
+      "step": 8545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9292110074539406e-05,
+      "loss": 0.6352,
+      "step": 8550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9291641273264266e-05,
+      "loss": 0.2785,
+      "step": 8555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9291172471989126e-05,
+      "loss": 0.1242,
+      "step": 8560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9290703670713986e-05,
+      "loss": 0.1312,
+      "step": 8565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9290234869438846e-05,
+      "loss": 0.1535,
+      "step": 8570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9289766068163706e-05,
+      "loss": 0.2025,
+      "step": 8575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9289297266888566e-05,
+      "loss": 0.2068,
+      "step": 8580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9288828465613426e-05,
+      "loss": 0.344,
+      "step": 8585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.928835966433829e-05,
+      "loss": 0.2893,
+      "step": 8590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.928789086306315e-05,
+      "loss": 0.5662,
+      "step": 8595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.928742206178801e-05,
+      "loss": 0.5428,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9286953260512872e-05,
+      "loss": 0.227,
+      "step": 8605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9286484459237732e-05,
+      "loss": 0.1013,
+      "step": 8610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9286015657962592e-05,
+      "loss": 0.1645,
+      "step": 8615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9285546856687452e-05,
+      "loss": 0.1979,
+      "step": 8620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9285078055412312e-05,
+      "loss": 0.2604,
+      "step": 8625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9284609254137172e-05,
+      "loss": 0.2365,
+      "step": 8630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9284140452862032e-05,
+      "loss": 0.2302,
+      "step": 8635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9283671651586895e-05,
+      "loss": 0.2653,
+      "step": 8640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9283202850311755e-05,
+      "loss": 0.3308,
+      "step": 8645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9282734049036615e-05,
+      "loss": 0.5864,
+      "step": 8650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9282265247761475e-05,
+      "loss": 0.3617,
+      "step": 8655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9281796446486335e-05,
+      "loss": 0.1467,
+      "step": 8660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9281327645211195e-05,
+      "loss": 0.1168,
+      "step": 8665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9280858843936058e-05,
+      "loss": 0.2285,
+      "step": 8670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9280390042660918e-05,
+      "loss": 0.2232,
+      "step": 8675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9279921241385778e-05,
+      "loss": 0.1808,
+      "step": 8680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927945244011064e-05,
+      "loss": 0.2971,
+      "step": 8685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.92789836388355e-05,
+      "loss": 0.2418,
+      "step": 8690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927851483756036e-05,
+      "loss": 0.3725,
+      "step": 8695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927804603628522e-05,
+      "loss": 0.6266,
+      "step": 8700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927757723501008e-05,
+      "loss": 0.3189,
+      "step": 8705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927710843373494e-05,
+      "loss": 0.0855,
+      "step": 8710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.92766396324598e-05,
+      "loss": 0.1166,
+      "step": 8715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927617083118466e-05,
+      "loss": 0.2014,
+      "step": 8720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927570202990952e-05,
+      "loss": 0.257,
+      "step": 8725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9275233228634384e-05,
+      "loss": 0.3014,
+      "step": 8730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9274764427359244e-05,
+      "loss": 0.2917,
+      "step": 8735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9274295626084104e-05,
+      "loss": 0.3215,
+      "step": 8740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9273826824808964e-05,
+      "loss": 0.3349,
+      "step": 8745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9273358023533827e-05,
+      "loss": 0.5432,
+      "step": 8750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9272889222258687e-05,
+      "loss": 0.3159,
+      "step": 8755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9272420420983547e-05,
+      "loss": 0.1048,
+      "step": 8760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9271951619708407e-05,
+      "loss": 0.1627,
+      "step": 8765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9271482818433267e-05,
+      "loss": 0.1638,
+      "step": 8770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9271014017158127e-05,
+      "loss": 0.1727,
+      "step": 8775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927054521588299e-05,
+      "loss": 0.1882,
+      "step": 8780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.927007641460785e-05,
+      "loss": 0.214,
+      "step": 8785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926960761333271e-05,
+      "loss": 0.2846,
+      "step": 8790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926913881205757e-05,
+      "loss": 0.4149,
+      "step": 8795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926867001078243e-05,
+      "loss": 0.5301,
+      "step": 8800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926820120950729e-05,
+      "loss": 0.2375,
+      "step": 8805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9267732408232153e-05,
+      "loss": 0.1747,
+      "step": 8810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9267263606957013e-05,
+      "loss": 0.1291,
+      "step": 8815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9266794805681873e-05,
+      "loss": 0.1782,
+      "step": 8820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9266326004406736e-05,
+      "loss": 0.1866,
+      "step": 8825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9265857203131596e-05,
+      "loss": 0.2627,
+      "step": 8830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9265388401856456e-05,
+      "loss": 0.3373,
+      "step": 8835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9264919600581316e-05,
+      "loss": 0.3501,
+      "step": 8840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9264450799306176e-05,
+      "loss": 0.3043,
+      "step": 8845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9263981998031036e-05,
+      "loss": 0.5005,
+      "step": 8850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9263513196755896e-05,
+      "loss": 0.2497,
+      "step": 8855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9263044395480756e-05,
+      "loss": 0.1362,
+      "step": 8860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9262575594205616e-05,
+      "loss": 0.1199,
+      "step": 8865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9262106792930476e-05,
+      "loss": 0.1285,
+      "step": 8870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926163799165534e-05,
+      "loss": 0.1813,
+      "step": 8875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.92611691903802e-05,
+      "loss": 0.1687,
+      "step": 8880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.926070038910506e-05,
+      "loss": 0.2707,
+      "step": 8885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9260231587829922e-05,
+      "loss": 0.3641,
+      "step": 8890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9259762786554782e-05,
+      "loss": 0.4102,
+      "step": 8895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9259293985279642e-05,
+      "loss": 0.5151,
+      "step": 8900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9258825184004502e-05,
+      "loss": 0.3642,
+      "step": 8905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9258356382729362e-05,
+      "loss": 0.102,
+      "step": 8910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9257887581454222e-05,
+      "loss": 0.1298,
+      "step": 8915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9257418780179085e-05,
+      "loss": 0.1646,
+      "step": 8920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9256949978903945e-05,
+      "loss": 0.2703,
+      "step": 8925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9256481177628805e-05,
+      "loss": 0.2339,
+      "step": 8930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9256012376353665e-05,
+      "loss": 0.2508,
+      "step": 8935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9255543575078525e-05,
+      "loss": 0.2826,
+      "step": 8940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9255074773803385e-05,
+      "loss": 0.4002,
+      "step": 8945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9254605972528245e-05,
+      "loss": 0.5675,
+      "step": 8950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9254137171253108e-05,
+      "loss": 0.2835,
+      "step": 8955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9253668369977968e-05,
+      "loss": 0.0916,
+      "step": 8960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9253199568702828e-05,
+      "loss": 0.1194,
+      "step": 8965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925273076742769e-05,
+      "loss": 0.1408,
+      "step": 8970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925226196615255e-05,
+      "loss": 0.1439,
+      "step": 8975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925179316487741e-05,
+      "loss": 0.2504,
+      "step": 8980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925132436360227e-05,
+      "loss": 0.3225,
+      "step": 8985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925085556232713e-05,
+      "loss": 0.3019,
+      "step": 8990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.925038676105199e-05,
+      "loss": 0.5128,
+      "step": 8995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924991795977685e-05,
+      "loss": 0.744,
+      "step": 9000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924944915850171e-05,
+      "loss": 0.3016,
+      "step": 9005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9248980357226574e-05,
+      "loss": 0.1325,
+      "step": 9010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9248511555951434e-05,
+      "loss": 0.125,
+      "step": 9015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9248042754676294e-05,
+      "loss": 0.1749,
+      "step": 9020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9247573953401154e-05,
+      "loss": 0.2229,
+      "step": 9025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9247105152126014e-05,
+      "loss": 0.1848,
+      "step": 9030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9246636350850877e-05,
+      "loss": 0.2923,
+      "step": 9035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9246167549575737e-05,
+      "loss": 0.3329,
+      "step": 9040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9245698748300597e-05,
+      "loss": 0.3743,
+      "step": 9045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9245229947025457e-05,
+      "loss": 0.6587,
+      "step": 9050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9244761145750317e-05,
+      "loss": 0.2159,
+      "step": 9055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924429234447518e-05,
+      "loss": 0.1817,
+      "step": 9060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924382354320004e-05,
+      "loss": 0.1316,
+      "step": 9065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.92433547419249e-05,
+      "loss": 0.1602,
+      "step": 9070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924288594064976e-05,
+      "loss": 0.1972,
+      "step": 9075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924241713937462e-05,
+      "loss": 0.2425,
+      "step": 9080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924194833809948e-05,
+      "loss": 0.3043,
+      "step": 9085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.924147953682434e-05,
+      "loss": 0.3386,
+      "step": 9090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.92410107355492e-05,
+      "loss": 0.5037,
+      "step": 9095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9240541934274063e-05,
+      "loss": 0.8734,
+      "step": 9100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9240073132998923e-05,
+      "loss": 0.34,
+      "step": 9105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9239604331723783e-05,
+      "loss": 0.0871,
+      "step": 9110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9239135530448646e-05,
+      "loss": 0.1388,
+      "step": 9115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9238666729173506e-05,
+      "loss": 0.1693,
+      "step": 9120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9238197927898366e-05,
+      "loss": 0.2853,
+      "step": 9125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9237729126623226e-05,
+      "loss": 0.313,
+      "step": 9130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9237260325348086e-05,
+      "loss": 0.2605,
+      "step": 9135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9236791524072946e-05,
+      "loss": 0.2986,
+      "step": 9140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9236322722797806e-05,
+      "loss": 0.4476,
+      "step": 9145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923585392152267e-05,
+      "loss": 0.6561,
+      "step": 9150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923538512024753e-05,
+      "loss": 0.3252,
+      "step": 9155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923491631897239e-05,
+      "loss": 0.0688,
+      "step": 9160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923444751769725e-05,
+      "loss": 0.1332,
+      "step": 9165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.923397871642211e-05,
+      "loss": 0.1325,
+      "step": 9170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9233509915146972e-05,
+      "loss": 0.127,
+      "step": 9175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9233041113871832e-05,
+      "loss": 0.1931,
+      "step": 9180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9232572312596692e-05,
+      "loss": 0.1826,
+      "step": 9185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9232103511321552e-05,
+      "loss": 0.3144,
+      "step": 9190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9231634710046412e-05,
+      "loss": 0.3538,
+      "step": 9195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9231165908771275e-05,
+      "loss": 0.5988,
+      "step": 9200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9230697107496135e-05,
+      "loss": 0.3227,
+      "step": 9205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9230228306220995e-05,
+      "loss": 0.1147,
+      "step": 9210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9229759504945855e-05,
+      "loss": 0.133,
+      "step": 9215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9229290703670715e-05,
+      "loss": 0.2695,
+      "step": 9220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9228821902395575e-05,
+      "loss": 0.2266,
+      "step": 9225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9228353101120435e-05,
+      "loss": 0.216,
+      "step": 9230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9227884299845295e-05,
+      "loss": 0.3179,
+      "step": 9235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9227415498570158e-05,
+      "loss": 0.2371,
+      "step": 9240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9226946697295018e-05,
+      "loss": 0.4098,
+      "step": 9245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9226477896019878e-05,
+      "loss": 0.533,
+      "step": 9250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922600909474474e-05,
+      "loss": 0.2309,
+      "step": 9255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.92255402934696e-05,
+      "loss": 0.1296,
+      "step": 9260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922507149219446e-05,
+      "loss": 0.1396,
+      "step": 9265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922460269091932e-05,
+      "loss": 0.2446,
+      "step": 9270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922413388964418e-05,
+      "loss": 0.1899,
+      "step": 9275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.922366508836904e-05,
+      "loss": 0.285,
+      "step": 9280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.92231962870939e-05,
+      "loss": 0.28,
+      "step": 9285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9222727485818764e-05,
+      "loss": 0.2611,
+      "step": 9290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9222258684543624e-05,
+      "loss": 0.424,
+      "step": 9295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9221789883268484e-05,
+      "loss": 0.615,
+      "step": 9300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9221321081993344e-05,
+      "loss": 0.2705,
+      "step": 9305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9220852280718204e-05,
+      "loss": 0.0816,
+      "step": 9310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9220383479443064e-05,
+      "loss": 0.1157,
+      "step": 9315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9219914678167927e-05,
+      "loss": 0.1803,
+      "step": 9320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9219445876892787e-05,
+      "loss": 0.1273,
+      "step": 9325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9218977075617647e-05,
+      "loss": 0.269,
+      "step": 9330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921850827434251e-05,
+      "loss": 0.1694,
+      "step": 9335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921803947306737e-05,
+      "loss": 0.303,
+      "step": 9340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921757067179223e-05,
+      "loss": 0.4144,
+      "step": 9345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921710187051709e-05,
+      "loss": 0.4884,
+      "step": 9350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921663306924195e-05,
+      "loss": 0.2585,
+      "step": 9355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921616426796681e-05,
+      "loss": 0.0678,
+      "step": 9360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921569546669167e-05,
+      "loss": 0.1916,
+      "step": 9365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921522666541653e-05,
+      "loss": 0.1597,
+      "step": 9370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921475786414139e-05,
+      "loss": 0.2067,
+      "step": 9375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.921428906286625e-05,
+      "loss": 0.1931,
+      "step": 9380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9213820261591113e-05,
+      "loss": 0.427,
+      "step": 9385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9213351460315973e-05,
+      "loss": 0.3432,
+      "step": 9390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9212882659040833e-05,
+      "loss": 0.3622,
+      "step": 9395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9212413857765696e-05,
+      "loss": 0.4917,
+      "step": 9400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9211945056490556e-05,
+      "loss": 0.2831,
+      "step": 9405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9211476255215416e-05,
+      "loss": 0.1006,
+      "step": 9410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9211007453940276e-05,
+      "loss": 0.1673,
+      "step": 9415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9210538652665136e-05,
+      "loss": 0.2262,
+      "step": 9420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9210069851389996e-05,
+      "loss": 0.156,
+      "step": 9425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920960105011486e-05,
+      "loss": 0.1828,
+      "step": 9430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920913224883972e-05,
+      "loss": 0.1871,
+      "step": 9435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920866344756458e-05,
+      "loss": 0.3073,
+      "step": 9440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920819464628944e-05,
+      "loss": 0.4646,
+      "step": 9445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92077258450143e-05,
+      "loss": 0.5954,
+      "step": 9450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920725704373916e-05,
+      "loss": 0.2833,
+      "step": 9455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920678824246402e-05,
+      "loss": 0.1231,
+      "step": 9460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9206319441188882e-05,
+      "loss": 0.1253,
+      "step": 9465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9205850639913742e-05,
+      "loss": 0.1852,
+      "step": 9470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9205381838638602e-05,
+      "loss": 0.2212,
+      "step": 9475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9204913037363466e-05,
+      "loss": 0.1587,
+      "step": 9480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9204444236088325e-05,
+      "loss": 0.3125,
+      "step": 9485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9203975434813185e-05,
+      "loss": 0.3626,
+      "step": 9490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9203506633538045e-05,
+      "loss": 0.4767,
+      "step": 9495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9203037832262905e-05,
+      "loss": 0.6986,
+      "step": 9500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9202569030987765e-05,
+      "loss": 0.2779,
+      "step": 9505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9202100229712625e-05,
+      "loss": 0.095,
+      "step": 9510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9201631428437485e-05,
+      "loss": 0.1261,
+      "step": 9515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9201162627162345e-05,
+      "loss": 0.1483,
+      "step": 9520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.920069382588721e-05,
+      "loss": 0.1876,
+      "step": 9525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9200225024612068e-05,
+      "loss": 0.236,
+      "step": 9530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9199756223336928e-05,
+      "loss": 0.2127,
+      "step": 9535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9199287422061788e-05,
+      "loss": 0.2471,
+      "step": 9540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919881862078665e-05,
+      "loss": 0.4438,
+      "step": 9545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919834981951151e-05,
+      "loss": 0.5967,
+      "step": 9550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919788101823637e-05,
+      "loss": 0.3444,
+      "step": 9555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919741221696123e-05,
+      "loss": 0.142,
+      "step": 9560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919694341568609e-05,
+      "loss": 0.0761,
+      "step": 9565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9196474614410955e-05,
+      "loss": 0.2179,
+      "step": 9570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9196005813135814e-05,
+      "loss": 0.1785,
+      "step": 9575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9195537011860674e-05,
+      "loss": 0.2536,
+      "step": 9580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9195068210585534e-05,
+      "loss": 0.1959,
+      "step": 9585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9194599409310394e-05,
+      "loss": 0.2637,
+      "step": 9590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9194130608035254e-05,
+      "loss": 0.4235,
+      "step": 9595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9193661806760114e-05,
+      "loss": 0.3954,
+      "step": 9600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9193193005484977e-05,
+      "loss": 0.2325,
+      "step": 9605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9192724204209837e-05,
+      "loss": 0.1367,
+      "step": 9610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9192255402934697e-05,
+      "loss": 0.1429,
+      "step": 9615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9191786601659557e-05,
+      "loss": 0.1687,
+      "step": 9620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919131780038442e-05,
+      "loss": 0.263,
+      "step": 9625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919084899910928e-05,
+      "loss": 0.2288,
+      "step": 9630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.919038019783414e-05,
+      "loss": 0.2932,
+      "step": 9635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9189911396559e-05,
+      "loss": 0.5055,
+      "step": 9640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918944259528386e-05,
+      "loss": 0.5668,
+      "step": 9645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918897379400872e-05,
+      "loss": 0.5284,
+      "step": 9650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918850499273358e-05,
+      "loss": 0.3792,
+      "step": 9655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9188036191458443e-05,
+      "loss": 0.1393,
+      "step": 9660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9187567390183303e-05,
+      "loss": 0.147,
+      "step": 9665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9187098588908163e-05,
+      "loss": 0.1659,
+      "step": 9670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9186629787633023e-05,
+      "loss": 0.1626,
+      "step": 9675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9186160986357883e-05,
+      "loss": 0.1592,
+      "step": 9680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9185692185082747e-05,
+      "loss": 0.2789,
+      "step": 9685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9185223383807606e-05,
+      "loss": 0.2428,
+      "step": 9690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9184754582532466e-05,
+      "loss": 0.3408,
+      "step": 9695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9184285781257326e-05,
+      "loss": 0.6366,
+      "step": 9700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9183816979982186e-05,
+      "loss": 0.2249,
+      "step": 9705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918334817870705e-05,
+      "loss": 0.1138,
+      "step": 9710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918287937743191e-05,
+      "loss": 0.1484,
+      "step": 9715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918241057615677e-05,
+      "loss": 0.1745,
+      "step": 9720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918194177488163e-05,
+      "loss": 0.1442,
+      "step": 9725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918147297360649e-05,
+      "loss": 0.2449,
+      "step": 9730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918100417233135e-05,
+      "loss": 0.2956,
+      "step": 9735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918053537105621e-05,
+      "loss": 0.3274,
+      "step": 9740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.918006656978107e-05,
+      "loss": 0.3507,
+      "step": 9745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9179597768505932e-05,
+      "loss": 0.6847,
+      "step": 9750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9179128967230792e-05,
+      "loss": 0.2816,
+      "step": 9755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9178660165955652e-05,
+      "loss": 0.1874,
+      "step": 9760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9178191364680516e-05,
+      "loss": 0.1852,
+      "step": 9765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177722563405376e-05,
+      "loss": 0.1936,
+      "step": 9770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9177253762130236e-05,
+      "loss": 0.148,
+      "step": 9775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9176784960855095e-05,
+      "loss": 0.2847,
+      "step": 9780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9176316159579955e-05,
+      "loss": 0.3086,
+      "step": 9785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9175847358304815e-05,
+      "loss": 0.2333,
+      "step": 9790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9175378557029675e-05,
+      "loss": 0.4765,
+      "step": 9795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917490975575454e-05,
+      "loss": 0.5674,
+      "step": 9800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.91744409544794e-05,
+      "loss": 0.2181,
+      "step": 9805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917397215320426e-05,
+      "loss": 0.065,
+      "step": 9810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917350335192912e-05,
+      "loss": 0.1668,
+      "step": 9815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9173034550653978e-05,
+      "loss": 0.1574,
+      "step": 9820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9172565749378838e-05,
+      "loss": 0.1636,
+      "step": 9825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.91720969481037e-05,
+      "loss": 0.2269,
+      "step": 9830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917162814682856e-05,
+      "loss": 0.2636,
+      "step": 9835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917115934555342e-05,
+      "loss": 0.1976,
+      "step": 9840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.917069054427828e-05,
+      "loss": 0.4929,
+      "step": 9845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9170221743003145e-05,
+      "loss": 0.6642,
+      "step": 9850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9169752941728005e-05,
+      "loss": 0.2061,
+      "step": 9855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9169284140452865e-05,
+      "loss": 0.106,
+      "step": 9860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9168815339177724e-05,
+      "loss": 0.1755,
+      "step": 9865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9168346537902584e-05,
+      "loss": 0.1511,
+      "step": 9870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9167877736627444e-05,
+      "loss": 0.2493,
+      "step": 9875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9167408935352304e-05,
+      "loss": 0.309,
+      "step": 9880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9166940134077164e-05,
+      "loss": 0.2733,
+      "step": 9885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9166471332802024e-05,
+      "loss": 0.288,
+      "step": 9890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9166002531526887e-05,
+      "loss": 0.3824,
+      "step": 9895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9165533730251747e-05,
+      "loss": 0.6902,
+      "step": 9900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9165064928976607e-05,
+      "loss": 0.2577,
+      "step": 9905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916459612770147e-05,
+      "loss": 0.1099,
+      "step": 9910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916412732642633e-05,
+      "loss": 0.1004,
+      "step": 9915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916365852515119e-05,
+      "loss": 0.1233,
+      "step": 9920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916318972387605e-05,
+      "loss": 0.1185,
+      "step": 9925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916272092260091e-05,
+      "loss": 0.2188,
+      "step": 9930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.916225212132577e-05,
+      "loss": 0.2539,
+      "step": 9935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9161783320050634e-05,
+      "loss": 0.2565,
+      "step": 9940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9161314518775494e-05,
+      "loss": 0.4717,
+      "step": 9945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9160845717500354e-05,
+      "loss": 0.493,
+      "step": 9950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9160376916225213e-05,
+      "loss": 0.2629,
+      "step": 9955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9159908114950073e-05,
+      "loss": 0.07,
+      "step": 9960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9159439313674933e-05,
+      "loss": 0.1622,
+      "step": 9965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158970512399793e-05,
+      "loss": 0.2093,
+      "step": 9970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158501711124657e-05,
+      "loss": 0.1833,
+      "step": 9975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158032909849516e-05,
+      "loss": 0.2507,
+      "step": 9980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157564108574376e-05,
+      "loss": 0.3487,
+      "step": 9985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915709530729924e-05,
+      "loss": 0.3129,
+      "step": 9990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91566265060241e-05,
+      "loss": 0.4036,
+      "step": 9995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915615770474896e-05,
+      "loss": 0.5754,
+      "step": 10000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915568890347382e-05,
+      "loss": 0.2006,
+      "step": 10005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915522010219868e-05,
+      "loss": 0.1141,
+      "step": 10010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915475130092354e-05,
+      "loss": 0.1682,
+      "step": 10015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91542824996484e-05,
+      "loss": 0.1504,
+      "step": 10020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915381369837326e-05,
+      "loss": 0.2555,
+      "step": 10025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915334489709812e-05,
+      "loss": 0.2481,
+      "step": 10030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152876095822983e-05,
+      "loss": 0.1897,
+      "step": 10035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9152407294547842e-05,
+      "loss": 0.2689,
+      "step": 10040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151938493272702e-05,
+      "loss": 0.4531,
+      "step": 10045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151469691997562e-05,
+      "loss": 0.6491,
+      "step": 10050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151000890722426e-05,
+      "loss": 0.2868,
+      "step": 10055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150532089447286e-05,
+      "loss": 0.0885,
+      "step": 10060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150063288172146e-05,
+      "loss": 0.1029,
+      "step": 10065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149594486897005e-05,
+      "loss": 0.1705,
+      "step": 10070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9149125685621865e-05,
+      "loss": 0.2066,
+      "step": 10075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914865688434673e-05,
+      "loss": 0.2518,
+      "step": 10080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914818808307159e-05,
+      "loss": 0.1692,
+      "step": 10085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914771928179645e-05,
+      "loss": 0.3126,
+      "step": 10090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914725048052131e-05,
+      "loss": 0.5543,
+      "step": 10095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914678167924617e-05,
+      "loss": 0.8418,
+      "step": 10100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914631287797103e-05,
+      "loss": 0.2604,
+      "step": 10105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914584407669589e-05,
+      "loss": 0.0586,
+      "step": 10110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914537527542075e-05,
+      "loss": 0.1234,
+      "step": 10115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914490647414561e-05,
+      "loss": 0.1425,
+      "step": 10120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914443767287047e-05,
+      "loss": 0.2744,
+      "step": 10125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914396887159533e-05,
+      "loss": 0.2341,
+      "step": 10130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9143500070320195e-05,
+      "loss": 0.3112,
+      "step": 10135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9143031269045055e-05,
+      "loss": 0.301,
+      "step": 10140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9142562467769915e-05,
+      "loss": 0.4034,
+      "step": 10145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9142093666494775e-05,
+      "loss": 0.4897,
+      "step": 10150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9141624865219635e-05,
+      "loss": 0.2841,
+      "step": 10155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9141156063944494e-05,
+      "loss": 0.1272,
+      "step": 10160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9140687262669354e-05,
+      "loss": 0.095,
+      "step": 10165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9140218461394214e-05,
+      "loss": 0.2431,
+      "step": 10170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9139749660119078e-05,
+      "loss": 0.185,
+      "step": 10175
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139280858843938e-05,
+      "loss": 0.1765,
+      "step": 10180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138812057568797e-05,
+      "loss": 0.1876,
+      "step": 10185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138343256293657e-05,
+      "loss": 0.205,
+      "step": 10190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913787445501852e-05,
+      "loss": 0.2962,
+      "step": 10195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913740565374338e-05,
+      "loss": 0.6509,
+      "step": 10200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913693685246824e-05,
+      "loss": 0.2717,
+      "step": 10205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91364680511931e-05,
+      "loss": 0.1069,
+      "step": 10210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913599924991796e-05,
+      "loss": 0.2321,
+      "step": 10215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135530448642824e-05,
+      "loss": 0.1316,
+      "step": 10220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135061647367684e-05,
+      "loss": 0.1441,
+      "step": 10225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134592846092544e-05,
+      "loss": 0.2482,
+      "step": 10230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134124044817404e-05,
+      "loss": 0.3006,
+      "step": 10235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133655243542264e-05,
+      "loss": 0.4242,
+      "step": 10240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133186442267123e-05,
+      "loss": 0.3968,
+      "step": 10245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132717640991983e-05,
+      "loss": 0.5792,
+      "step": 10250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132248839716843e-05,
+      "loss": 0.3714,
+      "step": 10255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131780038441707e-05,
+      "loss": 0.1612,
+      "step": 10260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131311237166567e-05,
+      "loss": 0.1666,
+      "step": 10265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130842435891427e-05,
+      "loss": 0.2056,
+      "step": 10270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913037363461629e-05,
+      "loss": 0.2004,
+      "step": 10275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912990483334115e-05,
+      "loss": 0.2485,
+      "step": 10280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912943603206601e-05,
+      "loss": 0.2626,
+      "step": 10285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912896723079087e-05,
+      "loss": 0.3242,
+      "step": 10290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912849842951573e-05,
+      "loss": 0.3341,
+      "step": 10295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912802962824059e-05,
+      "loss": 0.5659,
+      "step": 10300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912756082696545e-05,
+      "loss": 0.2671,
+      "step": 10305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127092025690313e-05,
+      "loss": 0.1295,
+      "step": 10310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126623224415173e-05,
+      "loss": 0.1669,
+      "step": 10315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126154423140033e-05,
+      "loss": 0.1559,
+      "step": 10320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125685621864893e-05,
+      "loss": 0.1648,
+      "step": 10325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125216820589753e-05,
+      "loss": 0.2274,
+      "step": 10330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124748019314612e-05,
+      "loss": 0.3051,
+      "step": 10335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124279218039476e-05,
+      "loss": 0.3135,
+      "step": 10340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123810416764336e-05,
+      "loss": 0.4214,
+      "step": 10345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123341615489196e-05,
+      "loss": 0.4696,
+      "step": 10350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9122872814214056e-05,
+      "loss": 0.2314,
+      "step": 10355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912240401293892e-05,
+      "loss": 0.0987,
+      "step": 10360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912193521166378e-05,
+      "loss": 0.103,
+      "step": 10365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912146641038864e-05,
+      "loss": 0.1721,
+      "step": 10370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91209976091135e-05,
+      "loss": 0.1067,
+      "step": 10375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912052880783836e-05,
+      "loss": 0.2847,
+      "step": 10380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912006000656322e-05,
+      "loss": 0.2213,
+      "step": 10385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911959120528808e-05,
+      "loss": 0.2882,
+      "step": 10390
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911912240401294e-05,
+      "loss": 0.4464,
+      "step": 10395
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.91186536027378e-05,
+      "loss": 0.6176,
+      "step": 10400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911818480146266e-05,
+      "loss": 0.294,
+      "step": 10405
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911771600018752e-05,
+      "loss": 0.1381,
+      "step": 10410
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911724719891238e-05,
+      "loss": 0.1054,
+      "step": 10415
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9116778397637245e-05,
+      "loss": 0.1755,
+      "step": 10420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9116309596362105e-05,
+      "loss": 0.13,
+      "step": 10425
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9115840795086965e-05,
+      "loss": 0.1956,
+      "step": 10430
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9115371993811825e-05,
+      "loss": 0.272,
+      "step": 10435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9114903192536685e-05,
+      "loss": 0.325,
+      "step": 10440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9114434391261545e-05,
+      "loss": 0.4834,
+      "step": 10445
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9113965589986408e-05,
+      "loss": 0.4972,
+      "step": 10450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9113496788711268e-05,
+      "loss": 0.31,
+      "step": 10455
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9113027987436128e-05,
+      "loss": 0.0929,
+      "step": 10460
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9112559186160988e-05,
+      "loss": 0.1527,
+      "step": 10465
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9112090384885848e-05,
+      "loss": 0.2049,
+      "step": 10470
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9111621583610708e-05,
+      "loss": 0.2407,
+      "step": 10475
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9111152782335567e-05,
+      "loss": 0.2558,
+      "step": 10480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911068398106043e-05,
+      "loss": 0.2731,
+      "step": 10485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.911021517978529e-05,
+      "loss": 0.3283,
+      "step": 10490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910974637851015e-05,
+      "loss": 0.4684,
+      "step": 10495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9109277577235014e-05,
+      "loss": 0.4627,
+      "step": 10500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9108808775959874e-05,
+      "loss": 0.2122,
+      "step": 10505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9108339974684734e-05,
+      "loss": 0.1003,
+      "step": 10510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9107871173409594e-05,
+      "loss": 0.1256,
+      "step": 10515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9107402372134454e-05,
+      "loss": 0.0787,
+      "step": 10520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9106933570859314e-05,
+      "loss": 0.1799,
+      "step": 10525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9106464769584174e-05,
+      "loss": 0.2717,
+      "step": 10530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9105995968309034e-05,
+      "loss": 0.2664,
+      "step": 10535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9105527167033893e-05,
+      "loss": 0.3668,
+      "step": 10540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9105058365758757e-05,
+      "loss": 0.3952,
+      "step": 10545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9104589564483617e-05,
+      "loss": 0.5244,
+      "step": 10550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9104120763208477e-05,
+      "loss": 0.257,
+      "step": 10555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910365196193334e-05,
+      "loss": 0.1311,
+      "step": 10560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.91031831606582e-05,
+      "loss": 0.1379,
+      "step": 10565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910271435938306e-05,
+      "loss": 0.2639,
+      "step": 10570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910224555810792e-05,
+      "loss": 0.1975,
+      "step": 10575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910177675683278e-05,
+      "loss": 0.1877,
+      "step": 10580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910130795555764e-05,
+      "loss": 0.1974,
+      "step": 10585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100839154282503e-05,
+      "loss": 0.356,
+      "step": 10590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100370353007363e-05,
+      "loss": 0.3649,
+      "step": 10595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099901551732223e-05,
+      "loss": 0.6513,
+      "step": 10600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099432750457083e-05,
+      "loss": 0.219,
+      "step": 10605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9098963949181943e-05,
+      "loss": 0.1712,
+      "step": 10610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9098495147906803e-05,
+      "loss": 0.1444,
+      "step": 10615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9098026346631663e-05,
+      "loss": 0.1519,
+      "step": 10620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9097557545356526e-05,
+      "loss": 0.1516,
+      "step": 10625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9097088744081386e-05,
+      "loss": 0.2533,
+      "step": 10630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9096619942806246e-05,
+      "loss": 0.2046,
+      "step": 10635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909615114153111e-05,
+      "loss": 0.2561,
+      "step": 10640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909568234025597e-05,
+      "loss": 0.4102,
+      "step": 10645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909521353898083e-05,
+      "loss": 0.4769,
+      "step": 10650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909474473770569e-05,
+      "loss": 0.3088,
+      "step": 10655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909427593643055e-05,
+      "loss": 0.1282,
+      "step": 10660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909380713515541e-05,
+      "loss": 0.1414,
+      "step": 10665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909333833388027e-05,
+      "loss": 0.1356,
+      "step": 10670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909286953260513e-05,
+      "loss": 0.1508,
+      "step": 10675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909240073132999e-05,
+      "loss": 0.2121,
+      "step": 10680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9091931930054852e-05,
+      "loss": 0.2012,
+      "step": 10685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9091463128779712e-05,
+      "loss": 0.4462,
+      "step": 10690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9090994327504572e-05,
+      "loss": 0.3469,
+      "step": 10695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.909052552622943e-05,
+      "loss": 0.5109,
+      "step": 10700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9090056724954295e-05,
+      "loss": 0.2304,
+      "step": 10705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9089587923679155e-05,
+      "loss": 0.1118,
+      "step": 10710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9089119122404015e-05,
+      "loss": 0.1731,
+      "step": 10715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9088650321128875e-05,
+      "loss": 0.1422,
+      "step": 10720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9088181519853735e-05,
+      "loss": 0.2179,
+      "step": 10725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9087712718578598e-05,
+      "loss": 0.2427,
+      "step": 10730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9087243917303458e-05,
+      "loss": 0.2411,
+      "step": 10735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9086775116028318e-05,
+      "loss": 0.2579,
+      "step": 10740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9086306314753178e-05,
+      "loss": 0.3481,
+      "step": 10745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9085837513478038e-05,
+      "loss": 0.5099,
+      "step": 10750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9085368712202898e-05,
+      "loss": 0.2352,
+      "step": 10755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9084899910927758e-05,
+      "loss": 0.1045,
+      "step": 10760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9084431109652618e-05,
+      "loss": 0.2044,
+      "step": 10765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.908396230837748e-05,
+      "loss": 0.1178,
+      "step": 10770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.908349350710234e-05,
+      "loss": 0.2247,
+      "step": 10775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90830247058272e-05,
+      "loss": 0.1583,
+      "step": 10780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9082555904552064e-05,
+      "loss": 0.2375,
+      "step": 10785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9082087103276924e-05,
+      "loss": 0.3239,
+      "step": 10790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9081618302001784e-05,
+      "loss": 0.3176,
+      "step": 10795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9081149500726644e-05,
+      "loss": 0.5501,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9080680699451504e-05,
+      "loss": 0.2226,
+      "step": 10805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9080211898176364e-05,
+      "loss": 0.085,
+      "step": 10810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9079743096901224e-05,
+      "loss": 0.1627,
+      "step": 10815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9079274295626084e-05,
+      "loss": 0.0776,
+      "step": 10820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9078805494350947e-05,
+      "loss": 0.2069,
+      "step": 10825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9078336693075807e-05,
+      "loss": 0.2966,
+      "step": 10830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9077867891800667e-05,
+      "loss": 0.3082,
+      "step": 10835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9077399090525527e-05,
+      "loss": 0.2768,
+      "step": 10840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9076930289250387e-05,
+      "loss": 0.3287,
+      "step": 10845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.907646148797525e-05,
+      "loss": 0.6915,
+      "step": 10850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.907599268670011e-05,
+      "loss": 0.2431,
+      "step": 10855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.907552388542497e-05,
+      "loss": 0.1133,
+      "step": 10860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.907505508414983e-05,
+      "loss": 0.1717,
+      "step": 10865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9074586282874693e-05,
+      "loss": 0.1157,
+      "step": 10870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9074117481599553e-05,
+      "loss": 0.2148,
+      "step": 10875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9073648680324413e-05,
+      "loss": 0.3085,
+      "step": 10880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9073179879049273e-05,
+      "loss": 0.2192,
+      "step": 10885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9072711077774133e-05,
+      "loss": 0.2851,
+      "step": 10890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9072242276498993e-05,
+      "loss": 0.2993,
+      "step": 10895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9071773475223853e-05,
+      "loss": 0.3583,
+      "step": 10900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9071304673948713e-05,
+      "loss": 0.3118,
+      "step": 10905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9070835872673576e-05,
+      "loss": 0.0893,
+      "step": 10910
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9070367071398436e-05,
+      "loss": 0.0958,
+      "step": 10915
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9069898270123296e-05,
+      "loss": 0.1406,
+      "step": 10920
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9069429468848156e-05,
+      "loss": 0.1843,
+      "step": 10925
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.906896066757302e-05,
+      "loss": 0.2012,
+      "step": 10930
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.906849186629788e-05,
+      "loss": 0.2509,
+      "step": 10935
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.906802306502274e-05,
+      "loss": 0.3919,
+      "step": 10940
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.90675542637476e-05,
+      "loss": 0.3983,
+      "step": 10945
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.906708546247246e-05,
+      "loss": 0.5589,
+      "step": 10950
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.906661666119732e-05,
+      "loss": 0.2583,
+      "step": 10955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9066147859922182e-05,
+      "loss": 0.1028,
+      "step": 10960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9065679058647042e-05,
+      "loss": 0.1177,
+      "step": 10965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9065210257371902e-05,
+      "loss": 0.2102,
+      "step": 10970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9064741456096762e-05,
+      "loss": 0.1336,
+      "step": 10975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9064272654821622e-05,
+      "loss": 0.2197,
+      "step": 10980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9063803853546482e-05,
+      "loss": 0.2254,
+      "step": 10985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9063335052271345e-05,
+      "loss": 0.3272,
+      "step": 10990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9062866250996205e-05,
+      "loss": 0.3669,
+      "step": 10995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9062397449721065e-05,
+      "loss": 0.5502,
+      "step": 11000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9061928648445925e-05,
+      "loss": 0.2126,
+      "step": 11005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9061459847170788e-05,
+      "loss": 0.1083,
+      "step": 11010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9060991045895648e-05,
+      "loss": 0.1021,
+      "step": 11015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9060522244620508e-05,
+      "loss": 0.2316,
+      "step": 11020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9060053443345368e-05,
+      "loss": 0.1626,
+      "step": 11025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9059584642070228e-05,
+      "loss": 0.1842,
+      "step": 11030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9059115840795088e-05,
+      "loss": 0.2737,
+      "step": 11035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9058647039519948e-05,
+      "loss": 0.2535,
+      "step": 11040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9058178238244808e-05,
+      "loss": 0.49,
+      "step": 11045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9057709436969668e-05,
+      "loss": 0.7787,
+      "step": 11050
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905724063569453e-05,
+      "loss": 0.2668,
+      "step": 11055
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905677183441939e-05,
+      "loss": 0.097,
+      "step": 11060
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.905630303314425e-05,
+      "loss": 0.0949,
+      "step": 11065
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9055834231869114e-05,
+      "loss": 0.1723,
+      "step": 11070
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9055365430593974e-05,
+      "loss": 0.2328,
+      "step": 11075
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9054896629318834e-05,
+      "loss": 0.2321,
+      "step": 11080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9054427828043694e-05,
+      "loss": 0.3256,
+      "step": 11085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9053959026768554e-05,
+      "loss": 0.331,
+      "step": 11090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9053490225493414e-05,
+      "loss": 0.3254,
+      "step": 11095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9053021424218277e-05,
+      "loss": 0.6373,
+      "step": 11100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9052552622943137e-05,
+      "loss": 0.2815,
+      "step": 11105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9052083821667997e-05,
+      "loss": 0.1203,
+      "step": 11110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9051615020392857e-05,
+      "loss": 0.2007,
+      "step": 11115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9051146219117717e-05,
+      "loss": 0.1206,
+      "step": 11120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9050677417842577e-05,
+      "loss": 0.2319,
+      "step": 11125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9050208616567437e-05,
+      "loss": 0.1269,
+      "step": 11130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.90497398152923e-05,
+      "loss": 0.2155,
+      "step": 11135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904927101401716e-05,
+      "loss": 0.1693,
+      "step": 11140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904880221274202e-05,
+      "loss": 0.4148,
+      "step": 11145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9048333411466883e-05,
+      "loss": 0.6929,
+      "step": 11150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9047864610191743e-05,
+      "loss": 0.2697,
+      "step": 11155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9047395808916603e-05,
+      "loss": 0.115,
+      "step": 11160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9046927007641463e-05,
+      "loss": 0.0827,
+      "step": 11165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9046458206366323e-05,
+      "loss": 0.132,
+      "step": 11170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9045989405091183e-05,
+      "loss": 0.2029,
+      "step": 11175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9045520603816043e-05,
+      "loss": 0.1555,
+      "step": 11180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9045051802540903e-05,
+      "loss": 0.2248,
+      "step": 11185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9044583001265763e-05,
+      "loss": 0.3005,
+      "step": 11190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9044114199990626e-05,
+      "loss": 0.335,
+      "step": 11195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9043645398715486e-05,
+      "loss": 0.636,
+      "step": 11200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9043176597440346e-05,
+      "loss": 0.2538,
+      "step": 11205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9042707796165206e-05,
+      "loss": 0.1232,
+      "step": 11210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904223899489007e-05,
+      "loss": 0.0881,
+      "step": 11215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904177019361493e-05,
+      "loss": 0.1117,
+      "step": 11220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904130139233979e-05,
+      "loss": 0.1845,
+      "step": 11225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904083259106465e-05,
+      "loss": 0.2521,
+      "step": 11230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.904036378978951e-05,
+      "loss": 0.2299,
+      "step": 11235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9039894988514372e-05,
+      "loss": 0.2237,
+      "step": 11240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9039426187239232e-05,
+      "loss": 0.4258,
+      "step": 11245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9038957385964092e-05,
+      "loss": 0.6981,
+      "step": 11250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9038488584688952e-05,
+      "loss": 0.2667,
+      "step": 11255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9038019783413812e-05,
+      "loss": 0.093,
+      "step": 11260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9037550982138672e-05,
+      "loss": 0.1913,
+      "step": 11265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9037082180863532e-05,
+      "loss": 0.1173,
+      "step": 11270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9036613379588392e-05,
+      "loss": 0.2151,
+      "step": 11275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9036144578313255e-05,
+      "loss": 0.1969,
+      "step": 11280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9035675777038115e-05,
+      "loss": 0.2601,
+      "step": 11285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9035206975762975e-05,
+      "loss": 0.2812,
+      "step": 11290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9034738174487838e-05,
+      "loss": 0.3009,
+      "step": 11295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9034269373212698e-05,
+      "loss": 0.5065,
+      "step": 11300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9033800571937558e-05,
+      "loss": 0.2912,
+      "step": 11305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9033331770662418e-05,
+      "loss": 0.0743,
+      "step": 11310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9032862969387278e-05,
+      "loss": 0.1384,
+      "step": 11315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9032394168112138e-05,
+      "loss": 0.1069,
+      "step": 11320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9031925366836998e-05,
+      "loss": 0.1452,
+      "step": 11325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9031456565561858e-05,
+      "loss": 0.162,
+      "step": 11330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903098776428672e-05,
+      "loss": 0.2265,
+      "step": 11335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903051896301158e-05,
+      "loss": 0.277,
+      "step": 11340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.903005016173644e-05,
+      "loss": 0.3428,
+      "step": 11345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.90295813604613e-05,
+      "loss": 0.4802,
+      "step": 11350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902911255918616e-05,
+      "loss": 0.2706,
+      "step": 11355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9028643757911024e-05,
+      "loss": 0.087,
+      "step": 11360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9028174956635884e-05,
+      "loss": 0.1064,
+      "step": 11365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9027706155360744e-05,
+      "loss": 0.0903,
+      "step": 11370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9027237354085604e-05,
+      "loss": 0.2627,
+      "step": 11375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9026768552810467e-05,
+      "loss": 0.1548,
+      "step": 11380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9026299751535327e-05,
+      "loss": 0.2691,
+      "step": 11385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9025830950260187e-05,
+      "loss": 0.2731,
+      "step": 11390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9025362148985047e-05,
+      "loss": 0.3839,
+      "step": 11395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9024893347709907e-05,
+      "loss": 0.6679,
+      "step": 11400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9024424546434767e-05,
+      "loss": 0.3018,
+      "step": 11405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9023955745159627e-05,
+      "loss": 0.1024,
+      "step": 11410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9023486943884487e-05,
+      "loss": 0.1274,
+      "step": 11415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902301814260935e-05,
+      "loss": 0.1693,
+      "step": 11420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902254934133421e-05,
+      "loss": 0.1957,
+      "step": 11425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902208054005907e-05,
+      "loss": 0.1639,
+      "step": 11430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.902161173878393e-05,
+      "loss": 0.2524,
+      "step": 11435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9021142937508793e-05,
+      "loss": 0.4686,
+      "step": 11440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9020674136233653e-05,
+      "loss": 0.4812,
+      "step": 11445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9020205334958513e-05,
+      "loss": 0.4917,
+      "step": 11450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9019736533683373e-05,
+      "loss": 0.2595,
+      "step": 11455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9019267732408233e-05,
+      "loss": 0.0769,
+      "step": 11460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9018798931133093e-05,
+      "loss": 0.1518,
+      "step": 11465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9018330129857953e-05,
+      "loss": 0.1158,
+      "step": 11470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9017861328582816e-05,
+      "loss": 0.1847,
+      "step": 11475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9017392527307676e-05,
+      "loss": 0.1882,
+      "step": 11480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9016923726032536e-05,
+      "loss": 0.2294,
+      "step": 11485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9016454924757396e-05,
+      "loss": 0.371,
+      "step": 11490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9015986123482256e-05,
+      "loss": 0.3808,
+      "step": 11495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.901551732220712e-05,
+      "loss": 0.5649,
+      "step": 11500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.901504852093198e-05,
+      "loss": 0.3712,
+      "step": 11505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.901457971965684e-05,
+      "loss": 0.1052,
+      "step": 11510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.90141109183817e-05,
+      "loss": 0.1216,
+      "step": 11515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9013642117106562e-05,
+      "loss": 0.1673,
+      "step": 11520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9013173315831422e-05,
+      "loss": 0.1775,
+      "step": 11525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9012704514556282e-05,
+      "loss": 0.2124,
+      "step": 11530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9012235713281142e-05,
+      "loss": 0.2195,
+      "step": 11535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9011766912006002e-05,
+      "loss": 0.2557,
+      "step": 11540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9011298110730862e-05,
+      "loss": 0.3398,
+      "step": 11545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9010829309455722e-05,
+      "loss": 0.6744,
+      "step": 11550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9010360508180582e-05,
+      "loss": 0.2358,
+      "step": 11555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9009891706905442e-05,
+      "loss": 0.0978,
+      "step": 11560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9009422905630305e-05,
+      "loss": 0.1722,
+      "step": 11565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9008954104355165e-05,
+      "loss": 0.1497,
+      "step": 11570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9008485303080025e-05,
+      "loss": 0.2723,
+      "step": 11575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900801650180489e-05,
+      "loss": 0.1787,
+      "step": 11580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900754770052975e-05,
+      "loss": 0.3611,
+      "step": 11585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9007078899254608e-05,
+      "loss": 0.2891,
+      "step": 11590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9006610097979468e-05,
+      "loss": 0.3841,
+      "step": 11595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9006141296704328e-05,
+      "loss": 0.6541,
+      "step": 11600
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9005672495429188e-05,
+      "loss": 0.234,
+      "step": 11605
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900520369415405e-05,
+      "loss": 0.1288,
+      "step": 11610
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900473489287891e-05,
+      "loss": 0.1206,
+      "step": 11615
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900426609160377e-05,
+      "loss": 0.176,
+      "step": 11620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900379729032863e-05,
+      "loss": 0.2343,
+      "step": 11625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900332848905349e-05,
+      "loss": 0.2604,
+      "step": 11630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900285968777835e-05,
+      "loss": 0.2572,
+      "step": 11635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.900239088650321e-05,
+      "loss": 0.2942,
+      "step": 11640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9001922085228074e-05,
+      "loss": 0.4468,
+      "step": 11645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9001453283952934e-05,
+      "loss": 0.7121,
+      "step": 11650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9000984482677794e-05,
+      "loss": 0.2138,
+      "step": 11655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9000515681402657e-05,
+      "loss": 0.1269,
+      "step": 11660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9000046880127517e-05,
+      "loss": 0.1191,
+      "step": 11665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8999578078852377e-05,
+      "loss": 0.2002,
+      "step": 11670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8999109277577237e-05,
+      "loss": 0.1856,
+      "step": 11675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8998640476302097e-05,
+      "loss": 0.1776,
+      "step": 11680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8998171675026957e-05,
+      "loss": 0.3287,
+      "step": 11685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8997702873751817e-05,
+      "loss": 0.2521,
+      "step": 11690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8997234072476677e-05,
+      "loss": 0.2936,
+      "step": 11695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8996765271201537e-05,
+      "loss": 0.5431,
+      "step": 11700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.89962964699264e-05,
+      "loss": 0.2955,
+      "step": 11705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.899582766865126e-05,
+      "loss": 0.067,
+      "step": 11710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.899535886737612e-05,
+      "loss": 0.1617,
+      "step": 11715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.899489006610098e-05,
+      "loss": 0.2038,
+      "step": 11720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8994421264825843e-05,
+      "loss": 0.259,
+      "step": 11725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8993952463550703e-05,
+      "loss": 0.18,
+      "step": 11730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8993483662275563e-05,
+      "loss": 0.2509,
+      "step": 11735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8993014861000423e-05,
+      "loss": 0.3614,
+      "step": 11740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8992546059725283e-05,
+      "loss": 0.4022,
+      "step": 11745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8992077258450146e-05,
+      "loss": 0.5694,
+      "step": 11750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8991608457175006e-05,
+      "loss": 0.2532,
+      "step": 11755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8991139655899866e-05,
+      "loss": 0.1058,
+      "step": 11760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8990670854624726e-05,
+      "loss": 0.133,
+      "step": 11765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8990202053349586e-05,
+      "loss": 0.1465,
+      "step": 11770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8989733252074446e-05,
+      "loss": 0.1479,
+      "step": 11775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8989264450799306e-05,
+      "loss": 0.2112,
+      "step": 11780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8988795649524166e-05,
+      "loss": 0.229,
+      "step": 11785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.898832684824903e-05,
+      "loss": 0.2436,
+      "step": 11790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.898785804697389e-05,
+      "loss": 0.3649,
+      "step": 11795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.898738924569875e-05,
+      "loss": 0.7018,
+      "step": 11800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8986920444423613e-05,
+      "loss": 0.255,
+      "step": 11805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8986451643148472e-05,
+      "loss": 0.13,
+      "step": 11810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8985982841873332e-05,
+      "loss": 0.1292,
+      "step": 11815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8985514040598192e-05,
+      "loss": 0.1515,
+      "step": 11820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8985045239323052e-05,
+      "loss": 0.1737,
+      "step": 11825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8984576438047912e-05,
+      "loss": 0.1838,
+      "step": 11830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8984107636772772e-05,
+      "loss": 0.2843,
+      "step": 11835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8983638835497632e-05,
+      "loss": 0.309,
+      "step": 11840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8983170034222495e-05,
+      "loss": 0.3991,
+      "step": 11845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8982701232947355e-05,
+      "loss": 0.7293,
+      "step": 11850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8982232431672215e-05,
+      "loss": 0.2043,
+      "step": 11855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8981763630397075e-05,
+      "loss": 0.1122,
+      "step": 11860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8981294829121935e-05,
+      "loss": 0.1696,
+      "step": 11865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.89808260278468e-05,
+      "loss": 0.1471,
+      "step": 11870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.898035722657166e-05,
+      "loss": 0.2066,
+      "step": 11875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8979888425296518e-05,
+      "loss": 0.2047,
+      "step": 11880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8979419624021378e-05,
+      "loss": 0.2603,
+      "step": 11885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.897895082274624e-05,
+      "loss": 0.3069,
+      "step": 11890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.89784820214711e-05,
+      "loss": 0.4019,
+      "step": 11895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897801322019596e-05,
+      "loss": 0.5205,
+      "step": 11900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897754441892082e-05,
+      "loss": 0.2621,
+      "step": 11905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897707561764568e-05,
+      "loss": 0.1128,
+      "step": 11910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897660681637054e-05,
+      "loss": 0.1462,
+      "step": 11915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.89761380150954e-05,
+      "loss": 0.1285,
+      "step": 11920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897566921382026e-05,
+      "loss": 0.1582,
+      "step": 11925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8975200412545124e-05,
+      "loss": 0.2148,
+      "step": 11930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8974731611269984e-05,
+      "loss": 0.2385,
+      "step": 11935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8974262809994844e-05,
+      "loss": 0.2819,
+      "step": 11940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8973794008719704e-05,
+      "loss": 0.2586,
+      "step": 11945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8973325207444568e-05,
+      "loss": 0.7001,
+      "step": 11950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8972856406169427e-05,
+      "loss": 0.2394,
+      "step": 11955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8972387604894287e-05,
+      "loss": 0.0715,
+      "step": 11960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8971918803619147e-05,
+      "loss": 0.1559,
+      "step": 11965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8971450002344007e-05,
+      "loss": 0.164,
+      "step": 11970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8970981201068867e-05,
+      "loss": 0.1327,
+      "step": 11975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8970512399793727e-05,
+      "loss": 0.2868,
+      "step": 11980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.897004359851859e-05,
+      "loss": 0.2215,
+      "step": 11985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896957479724345e-05,
+      "loss": 0.3209,
+      "step": 11990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896910599596831e-05,
+      "loss": 0.3683,
+      "step": 11995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896863719469317e-05,
+      "loss": 0.6778,
+      "step": 12000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896816839341803e-05,
+      "loss": 0.2626,
+      "step": 12005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8967699592142894e-05,
+      "loss": 0.1251,
+      "step": 12010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8967230790867753e-05,
+      "loss": 0.134,
+      "step": 12015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8966761989592613e-05,
+      "loss": 0.1713,
+      "step": 12020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8966293188317473e-05,
+      "loss": 0.1434,
+      "step": 12025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8965824387042337e-05,
+      "loss": 0.2269,
+      "step": 12030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8965355585767197e-05,
+      "loss": 0.1525,
+      "step": 12035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8964886784492056e-05,
+      "loss": 0.2387,
+      "step": 12040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8964417983216916e-05,
+      "loss": 0.3101,
+      "step": 12045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8963949181941776e-05,
+      "loss": 0.5783,
+      "step": 12050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8963480380666636e-05,
+      "loss": 0.2279,
+      "step": 12055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8963011579391496e-05,
+      "loss": 0.1041,
+      "step": 12060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8962542778116356e-05,
+      "loss": 0.1346,
+      "step": 12065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8962073976841216e-05,
+      "loss": 0.1,
+      "step": 12070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896160517556608e-05,
+      "loss": 0.1159,
+      "step": 12075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.896113637429094e-05,
+      "loss": 0.2763,
+      "step": 12080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.89606675730158e-05,
+      "loss": 0.2358,
+      "step": 12085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8960198771740663e-05,
+      "loss": 0.3385,
+      "step": 12090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8959729970465523e-05,
+      "loss": 0.3224,
+      "step": 12095
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8959261169190382e-05,
+      "loss": 0.6675,
+      "step": 12100
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8958792367915242e-05,
+      "loss": 0.2995,
+      "step": 12105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8958323566640102e-05,
+      "loss": 0.0414,
+      "step": 12110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8957854765364962e-05,
+      "loss": 0.112,
+      "step": 12115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8957385964089822e-05,
+      "loss": 0.1313,
+      "step": 12120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8956917162814686e-05,
+      "loss": 0.1316,
+      "step": 12125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8956448361539545e-05,
+      "loss": 0.1418,
+      "step": 12130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8955979560264405e-05,
+      "loss": 0.2418,
+      "step": 12135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8955510758989265e-05,
+      "loss": 0.3003,
+      "step": 12140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8955041957714125e-05,
+      "loss": 0.4785,
+      "step": 12145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8954573156438985e-05,
+      "loss": 0.5624,
+      "step": 12150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895410435516385e-05,
+      "loss": 0.2527,
+      "step": 12155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895363555388871e-05,
+      "loss": 0.0963,
+      "step": 12160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895316675261357e-05,
+      "loss": 0.0813,
+      "step": 12165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8952697951338432e-05,
+      "loss": 0.1403,
+      "step": 12170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895222915006329e-05,
+      "loss": 0.1686,
+      "step": 12175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895176034878815e-05,
+      "loss": 0.1835,
+      "step": 12180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895129154751301e-05,
+      "loss": 0.1565,
+      "step": 12185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895082274623787e-05,
+      "loss": 0.383,
+      "step": 12190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.895035394496273e-05,
+      "loss": 0.375,
+      "step": 12195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894988514368759e-05,
+      "loss": 0.8233,
+      "step": 12200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894941634241245e-05,
+      "loss": 0.2411,
+      "step": 12205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894894754113731e-05,
+      "loss": 0.0966,
+      "step": 12210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8948478739862174e-05,
+      "loss": 0.1386,
+      "step": 12215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8948009938587034e-05,
+      "loss": 0.2216,
+      "step": 12220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8947541137311894e-05,
+      "loss": 0.181,
+      "step": 12225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8947072336036754e-05,
+      "loss": 0.1556,
+      "step": 12230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8946603534761618e-05,
+      "loss": 0.2458,
+      "step": 12235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8946134733486478e-05,
+      "loss": 0.2611,
+      "step": 12240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8945665932211337e-05,
+      "loss": 0.305,
+      "step": 12245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8945197130936197e-05,
+      "loss": 0.5079,
+      "step": 12250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8944728329661057e-05,
+      "loss": 0.2764,
+      "step": 12255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894425952838592e-05,
+      "loss": 0.0815,
+      "step": 12260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894379072711078e-05,
+      "loss": 0.0799,
+      "step": 12265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894332192583564e-05,
+      "loss": 0.1726,
+      "step": 12270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.89428531245605e-05,
+      "loss": 0.1347,
+      "step": 12275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894238432328536e-05,
+      "loss": 0.1555,
+      "step": 12280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894191552201022e-05,
+      "loss": 0.2513,
+      "step": 12285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.894144672073508e-05,
+      "loss": 0.3376,
+      "step": 12290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8940977919459944e-05,
+      "loss": 0.3014,
+      "step": 12295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8940509118184804e-05,
+      "loss": 0.5171,
+      "step": 12300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8940040316909663e-05,
+      "loss": 0.2599,
+      "step": 12305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8939571515634523e-05,
+      "loss": 0.0997,
+      "step": 12310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8939102714359387e-05,
+      "loss": 0.1469,
+      "step": 12315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8938633913084247e-05,
+      "loss": 0.1405,
+      "step": 12320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8938165111809107e-05,
+      "loss": 0.2392,
+      "step": 12325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8937696310533967e-05,
+      "loss": 0.1594,
+      "step": 12330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8937227509258826e-05,
+      "loss": 0.181,
+      "step": 12335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8936758707983686e-05,
+      "loss": 0.2841,
+      "step": 12340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8936289906708546e-05,
+      "loss": 0.3794,
+      "step": 12345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8935821105433406e-05,
+      "loss": 0.4712,
+      "step": 12350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.893535230415827e-05,
+      "loss": 0.2406,
+      "step": 12355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.893488350288313e-05,
+      "loss": 0.1092,
+      "step": 12360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.893441470160799e-05,
+      "loss": 0.0954,
+      "step": 12365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.893394590033285e-05,
+      "loss": 0.1485,
+      "step": 12370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8933477099057713e-05,
+      "loss": 0.2261,
+      "step": 12375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8933008297782573e-05,
+      "loss": 0.2628,
+      "step": 12380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8932539496507433e-05,
+      "loss": 0.2334,
+      "step": 12385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8932070695232293e-05,
+      "loss": 0.2887,
+      "step": 12390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8931601893957152e-05,
+      "loss": 0.2995,
+      "step": 12395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8931133092682016e-05,
+      "loss": 0.6158,
+      "step": 12400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8930664291406876e-05,
+      "loss": 0.2373,
+      "step": 12405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8930195490131736e-05,
+      "loss": 0.078,
+      "step": 12410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8929726688856596e-05,
+      "loss": 0.119,
+      "step": 12415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8929257887581455e-05,
+      "loss": 0.0879,
+      "step": 12420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8928789086306315e-05,
+      "loss": 0.1112,
+      "step": 12425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8928320285031175e-05,
+      "loss": 0.1409,
+      "step": 12430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8927851483756035e-05,
+      "loss": 0.2855,
+      "step": 12435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89273826824809e-05,
+      "loss": 0.1745,
+      "step": 12440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892691388120576e-05,
+      "loss": 0.344,
+      "step": 12445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892644507993062e-05,
+      "loss": 0.4228,
+      "step": 12450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8925976278655482e-05,
+      "loss": 0.2344,
+      "step": 12455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8925507477380342e-05,
+      "loss": 0.0847,
+      "step": 12460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89250386761052e-05,
+      "loss": 0.1617,
+      "step": 12465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892456987483006e-05,
+      "loss": 0.1262,
+      "step": 12470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892410107355492e-05,
+      "loss": 0.1367,
+      "step": 12475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892363227227978e-05,
+      "loss": 0.2337,
+      "step": 12480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.892316347100464e-05,
+      "loss": 0.2775,
+      "step": 12485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.89226946697295e-05,
+      "loss": 0.3666,
+      "step": 12490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8922225868454365e-05,
+      "loss": 0.3938,
+      "step": 12495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8921757067179225e-05,
+      "loss": 0.6377,
+      "step": 12500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8921288265904085e-05,
+      "loss": 0.2246,
+      "step": 12505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8920819464628944e-05,
+      "loss": 0.074,
+      "step": 12510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8920350663353804e-05,
+      "loss": 0.1685,
+      "step": 12515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8919881862078668e-05,
+      "loss": 0.1246,
+      "step": 12520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8919413060803528e-05,
+      "loss": 0.1052,
+      "step": 12525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8918944259528388e-05,
+      "loss": 0.2717,
+      "step": 12530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8918475458253248e-05,
+      "loss": 0.1854,
+      "step": 12535
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891800665697811e-05,
+      "loss": 0.3528,
+      "step": 12540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891753785570297e-05,
+      "loss": 0.1803,
+      "step": 12545
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891706905442783e-05,
+      "loss": 0.5755,
+      "step": 12550
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891660025315269e-05,
+      "loss": 0.2829,
+      "step": 12555
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891613145187755e-05,
+      "loss": 0.0912,
+      "step": 12560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891566265060241e-05,
+      "loss": 0.1055,
+      "step": 12565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891519384932727e-05,
+      "loss": 0.1751,
+      "step": 12570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891472504805213e-05,
+      "loss": 0.1399,
+      "step": 12575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.891425624677699e-05,
+      "loss": 0.1935,
+      "step": 12580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8913787445501854e-05,
+      "loss": 0.2453,
+      "step": 12585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8913318644226714e-05,
+      "loss": 0.1864,
+      "step": 12590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8912849842951574e-05,
+      "loss": 0.3226,
+      "step": 12595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8912381041676437e-05,
+      "loss": 0.3684,
+      "step": 12600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8911912240401297e-05,
+      "loss": 0.2648,
+      "step": 12605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8911443439126157e-05,
+      "loss": 0.061,
+      "step": 12610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8910974637851017e-05,
+      "loss": 0.1064,
+      "step": 12615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8910505836575877e-05,
+      "loss": 0.1771,
+      "step": 12620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8910037035300736e-05,
+      "loss": 0.1259,
+      "step": 12625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8909568234025596e-05,
+      "loss": 0.1778,
+      "step": 12630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890909943275046e-05,
+      "loss": 0.1552,
+      "step": 12635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890863063147532e-05,
+      "loss": 0.2755,
+      "step": 12640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890816183020018e-05,
+      "loss": 0.3597,
+      "step": 12645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890769302892504e-05,
+      "loss": 0.6109,
+      "step": 12650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.89072242276499e-05,
+      "loss": 0.2731,
+      "step": 12655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890675542637476e-05,
+      "loss": 0.1214,
+      "step": 12660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8906286625099623e-05,
+      "loss": 0.1173,
+      "step": 12665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8905817823824483e-05,
+      "loss": 0.1032,
+      "step": 12670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8905349022549343e-05,
+      "loss": 0.2092,
+      "step": 12675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8904880221274206e-05,
+      "loss": 0.2052,
+      "step": 12680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8904411419999066e-05,
+      "loss": 0.3043,
+      "step": 12685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8903942618723926e-05,
+      "loss": 0.3043,
+      "step": 12690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8903473817448786e-05,
+      "loss": 0.3598,
+      "step": 12695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8903005016173646e-05,
+      "loss": 0.5739,
+      "step": 12700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8902536214898506e-05,
+      "loss": 0.1942,
+      "step": 12705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8902067413623366e-05,
+      "loss": 0.0969,
+      "step": 12710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8901598612348225e-05,
+      "loss": 0.121,
+      "step": 12715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8901129811073085e-05,
+      "loss": 0.1946,
+      "step": 12720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890066100979795e-05,
+      "loss": 0.1054,
+      "step": 12725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.890019220852281e-05,
+      "loss": 0.2131,
+      "step": 12730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.889972340724767e-05,
+      "loss": 0.3432,
+      "step": 12735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.889925460597253e-05,
+      "loss": 0.3533,
+      "step": 12740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8898785804697392e-05,
+      "loss": 0.2782,
+      "step": 12745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8898317003422252e-05,
+      "loss": 1.0246,
+      "step": 12750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8897848202147112e-05,
+      "loss": 0.2875,
+      "step": 12755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889737940087197e-05,
+      "loss": 0.0689,
+      "step": 12760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889691059959683e-05,
+      "loss": 0.1025,
+      "step": 12765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889644179832169e-05,
+      "loss": 0.1406,
+      "step": 12770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8895972997046555e-05,
+      "loss": 0.1504,
+      "step": 12775
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8895504195771415e-05,
+      "loss": 0.2003,
+      "step": 12780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8895035394496275e-05,
+      "loss": 0.2294,
+      "step": 12785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8894566593221135e-05,
+      "loss": 0.2264,
+      "step": 12790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8894097791945995e-05,
+      "loss": 0.3517,
+      "step": 12795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8893628990670854e-05,
+      "loss": 0.6386,
+      "step": 12800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8893160189395718e-05,
+      "loss": 0.2141,
+      "step": 12805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8892691388120578e-05,
+      "loss": 0.1371,
+      "step": 12810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8892222586845438e-05,
+      "loss": 0.119,
+      "step": 12815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8891753785570298e-05,
+      "loss": 0.1502,
+      "step": 12820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889128498429516e-05,
+      "loss": 0.1665,
+      "step": 12825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889081618302002e-05,
+      "loss": 0.1954,
+      "step": 12830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.889034738174488e-05,
+      "loss": 0.2543,
+      "step": 12835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888987858046974e-05,
+      "loss": 0.2318,
+      "step": 12840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.88894097791946e-05,
+      "loss": 0.3885,
+      "step": 12845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888894097791946e-05,
+      "loss": 0.4542,
+      "step": 12850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888847217664432e-05,
+      "loss": 0.2372,
+      "step": 12855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888800337536918e-05,
+      "loss": 0.1154,
+      "step": 12860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8887534574094044e-05,
+      "loss": 0.0905,
+      "step": 12865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8887065772818904e-05,
+      "loss": 0.1059,
+      "step": 12870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8886596971543764e-05,
+      "loss": 0.2216,
+      "step": 12875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8886128170268624e-05,
+      "loss": 0.1879,
+      "step": 12880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8885659368993487e-05,
+      "loss": 0.2195,
+      "step": 12885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8885190567718347e-05,
+      "loss": 0.3195,
+      "step": 12890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8884721766443207e-05,
+      "loss": 0.2832,
+      "step": 12895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8884252965168067e-05,
+      "loss": 0.4605,
+      "step": 12900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8883784163892927e-05,
+      "loss": 0.2373,
+      "step": 12905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888331536261779e-05,
+      "loss": 0.1402,
+      "step": 12910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888284656134265e-05,
+      "loss": 0.0874,
+      "step": 12915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888237776006751e-05,
+      "loss": 0.1415,
+      "step": 12920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888190895879237e-05,
+      "loss": 0.2131,
+      "step": 12925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888144015751723e-05,
+      "loss": 0.1925,
+      "step": 12930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888097135624209e-05,
+      "loss": 0.1838,
+      "step": 12935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888050255496695e-05,
+      "loss": 0.2667,
+      "step": 12940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.888003375369181e-05,
+      "loss": 0.4606,
+      "step": 12945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8879564952416673e-05,
+      "loss": 0.593,
+      "step": 12950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8879096151141533e-05,
+      "loss": 0.2165,
+      "step": 12955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8878627349866393e-05,
+      "loss": 0.1231,
+      "step": 12960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8878158548591256e-05,
+      "loss": 0.1449,
+      "step": 12965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8877689747316116e-05,
+      "loss": 0.1908,
+      "step": 12970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8877220946040976e-05,
+      "loss": 0.1959,
+      "step": 12975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8876752144765836e-05,
+      "loss": 0.264,
+      "step": 12980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8876283343490696e-05,
+      "loss": 0.2641,
+      "step": 12985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8875814542215556e-05,
+      "loss": 0.2006,
+      "step": 12990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8875345740940416e-05,
+      "loss": 0.3727,
+      "step": 12995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8874876939665276e-05,
+      "loss": 0.6469,
+      "step": 13000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.887440813839014e-05,
+      "loss": 0.1506,
+      "step": 13005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8873939337115e-05,
+      "loss": 0.1014,
+      "step": 13010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.887347053583986e-05,
+      "loss": 0.1381,
+      "step": 13015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.887300173456472e-05,
+      "loss": 0.1419,
+      "step": 13020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.887253293328958e-05,
+      "loss": 0.1822,
+      "step": 13025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8872064132014442e-05,
+      "loss": 0.2727,
+      "step": 13030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8871595330739302e-05,
+      "loss": 0.2228,
+      "step": 13035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8871126529464162e-05,
+      "loss": 0.2866,
+      "step": 13040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8870657728189022e-05,
+      "loss": 0.2962,
+      "step": 13045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8870188926913885e-05,
+      "loss": 0.5063,
+      "step": 13050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8869720125638745e-05,
+      "loss": 0.2182,
+      "step": 13055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8869251324363605e-05,
+      "loss": 0.168,
+      "step": 13060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8868782523088465e-05,
+      "loss": 0.0722,
+      "step": 13065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8868313721813325e-05,
+      "loss": 0.1356,
+      "step": 13070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8867844920538185e-05,
+      "loss": 0.2862,
+      "step": 13075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8867376119263045e-05,
+      "loss": 0.1236,
+      "step": 13080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8866907317987905e-05,
+      "loss": 0.2837,
+      "step": 13085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8866438516712765e-05,
+      "loss": 0.329,
+      "step": 13090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8865969715437628e-05,
+      "loss": 0.3582,
+      "step": 13095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8865500914162488e-05,
+      "loss": 0.5645,
+      "step": 13100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8865032112887348e-05,
+      "loss": 0.2567,
+      "step": 13105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886456331161221e-05,
+      "loss": 0.1308,
+      "step": 13110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886409451033707e-05,
+      "loss": 0.1187,
+      "step": 13115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886362570906193e-05,
+      "loss": 0.1349,
+      "step": 13120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886315690778679e-05,
+      "loss": 0.1611,
+      "step": 13125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886268810651165e-05,
+      "loss": 0.1943,
+      "step": 13130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886221930523651e-05,
+      "loss": 0.1838,
+      "step": 13135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.886175050396137e-05,
+      "loss": 0.2555,
+      "step": 13140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8861281702686234e-05,
+      "loss": 0.3234,
+      "step": 13145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8860812901411094e-05,
+      "loss": 0.5682,
+      "step": 13150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8860344100135954e-05,
+      "loss": 0.2446,
+      "step": 13155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8859875298860814e-05,
+      "loss": 0.1032,
+      "step": 13160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8859406497585674e-05,
+      "loss": 0.0889,
+      "step": 13165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8858937696310534e-05,
+      "loss": 0.2059,
+      "step": 13170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8858468895035397e-05,
+      "loss": 0.1573,
+      "step": 13175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8858000093760257e-05,
+      "loss": 0.1783,
+      "step": 13180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8857531292485117e-05,
+      "loss": 0.2177,
+      "step": 13185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885706249120998e-05,
+      "loss": 0.24,
+      "step": 13190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885659368993484e-05,
+      "loss": 0.362,
+      "step": 13195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.88561248886597e-05,
+      "loss": 0.5683,
+      "step": 13200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885565608738456e-05,
+      "loss": 0.254,
+      "step": 13205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885518728610942e-05,
+      "loss": 0.1012,
+      "step": 13210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885471848483428e-05,
+      "loss": 0.1314,
+      "step": 13215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885424968355914e-05,
+      "loss": 0.1051,
+      "step": 13220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8853780882284e-05,
+      "loss": 0.1469,
+      "step": 13225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.885331208100886e-05,
+      "loss": 0.1632,
+      "step": 13230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8852843279733723e-05,
+      "loss": 0.2104,
+      "step": 13235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8852374478458583e-05,
+      "loss": 0.3246,
+      "step": 13240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8851905677183443e-05,
+      "loss": 0.213,
+      "step": 13245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8851436875908303e-05,
+      "loss": 0.5547,
+      "step": 13250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8850968074633166e-05,
+      "loss": 0.2415,
+      "step": 13255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8850499273358026e-05,
+      "loss": 0.0852,
+      "step": 13260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8850030472082886e-05,
+      "loss": 0.0872,
+      "step": 13265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8849561670807746e-05,
+      "loss": 0.1391,
+      "step": 13270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8849092869532606e-05,
+      "loss": 0.0998,
+      "step": 13275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8848624068257466e-05,
+      "loss": 0.1616,
+      "step": 13280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884815526698233e-05,
+      "loss": 0.1745,
+      "step": 13285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884768646570719e-05,
+      "loss": 0.3633,
+      "step": 13290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884721766443205e-05,
+      "loss": 0.4456,
+      "step": 13295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884674886315691e-05,
+      "loss": 0.6131,
+      "step": 13300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884628006188177e-05,
+      "loss": 0.2623,
+      "step": 13305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.884581126060663e-05,
+      "loss": 0.0944,
+      "step": 13310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8845342459331492e-05,
+      "loss": 0.1549,
+      "step": 13315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8844873658056352e-05,
+      "loss": 0.145,
+      "step": 13320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8844404856781212e-05,
+      "loss": 0.1983,
+      "step": 13325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8843936055506072e-05,
+      "loss": 0.263,
+      "step": 13330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8843467254230935e-05,
+      "loss": 0.2859,
+      "step": 13335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8842998452955795e-05,
+      "loss": 0.2764,
+      "step": 13340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8842529651680655e-05,
+      "loss": 0.3232,
+      "step": 13345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8842060850405515e-05,
+      "loss": 0.5258,
+      "step": 13350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8841592049130375e-05,
+      "loss": 0.2526,
+      "step": 13355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8841123247855235e-05,
+      "loss": 0.0839,
+      "step": 13360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8840654446580095e-05,
+      "loss": 0.1898,
+      "step": 13365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8840185645304955e-05,
+      "loss": 0.1795,
+      "step": 13370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8839716844029818e-05,
+      "loss": 0.1649,
+      "step": 13375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8839248042754678e-05,
+      "loss": 0.2422,
+      "step": 13380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8838779241479538e-05,
+      "loss": 0.2588,
+      "step": 13385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8838310440204398e-05,
+      "loss": 0.2336,
+      "step": 13390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883784163892926e-05,
+      "loss": 0.3311,
+      "step": 13395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883737283765412e-05,
+      "loss": 0.5475,
+      "step": 13400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883690403637898e-05,
+      "loss": 0.3553,
+      "step": 13405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883643523510384e-05,
+      "loss": 0.1288,
+      "step": 13410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.88359664338287e-05,
+      "loss": 0.1275,
+      "step": 13415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883549763255356e-05,
+      "loss": 0.1297,
+      "step": 13420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8835028831278424e-05,
+      "loss": 0.1836,
+      "step": 13425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8834560030003284e-05,
+      "loss": 0.2246,
+      "step": 13430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8834091228728144e-05,
+      "loss": 0.1759,
+      "step": 13435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8833622427453004e-05,
+      "loss": 0.2417,
+      "step": 13440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8833153626177864e-05,
+      "loss": 0.3051,
+      "step": 13445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8832684824902724e-05,
+      "loss": 0.5476,
+      "step": 13450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8832216023627584e-05,
+      "loss": 0.2013,
+      "step": 13455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8831747222352447e-05,
+      "loss": 0.0828,
+      "step": 13460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8831278421077307e-05,
+      "loss": 0.0529,
+      "step": 13465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8830809619802167e-05,
+      "loss": 0.0868,
+      "step": 13470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.883034081852703e-05,
+      "loss": 0.1599,
+      "step": 13475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882987201725189e-05,
+      "loss": 0.2012,
+      "step": 13480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882940321597675e-05,
+      "loss": 0.2599,
+      "step": 13485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882893441470161e-05,
+      "loss": 0.1924,
+      "step": 13490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882846561342647e-05,
+      "loss": 0.4321,
+      "step": 13495
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882799681215133e-05,
+      "loss": 0.6084,
+      "step": 13500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882752801087619e-05,
+      "loss": 0.1664,
+      "step": 13505
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882705920960105e-05,
+      "loss": 0.1051,
+      "step": 13510
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8826590408325913e-05,
+      "loss": 0.1545,
+      "step": 13515
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8826121607050773e-05,
+      "loss": 0.1605,
+      "step": 13520
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8825652805775633e-05,
+      "loss": 0.1598,
+      "step": 13525
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8825184004500493e-05,
+      "loss": 0.2838,
+      "step": 13530
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8824715203225353e-05,
+      "loss": 0.1525,
+      "step": 13535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8824246401950216e-05,
+      "loss": 0.2664,
+      "step": 13540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8823777600675076e-05,
+      "loss": 0.3345,
+      "step": 13545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8823308799399936e-05,
+      "loss": 0.3239,
+      "step": 13550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8822839998124796e-05,
+      "loss": 0.195,
+      "step": 13555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882237119684966e-05,
+      "loss": 0.0748,
+      "step": 13560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882190239557452e-05,
+      "loss": 0.1101,
+      "step": 13565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882143359429938e-05,
+      "loss": 0.1743,
+      "step": 13570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882096479302424e-05,
+      "loss": 0.2923,
+      "step": 13575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.88204959917491e-05,
+      "loss": 0.2246,
+      "step": 13580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.882002719047396e-05,
+      "loss": 0.3307,
+      "step": 13585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881955838919882e-05,
+      "loss": 0.237,
+      "step": 13590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881908958792368e-05,
+      "loss": 0.3441,
+      "step": 13595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.881862078664854e-05,
+      "loss": 0.6305,
+      "step": 13600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8818151985373402e-05,
+      "loss": 0.234,
+      "step": 13605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8817683184098262e-05,
+      "loss": 0.136,
+      "step": 13610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8817214382823122e-05,
+      "loss": 0.1397,
+      "step": 13615
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8816745581547985e-05,
+      "loss": 0.1026,
+      "step": 13620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8816276780272845e-05,
+      "loss": 0.2265,
+      "step": 13625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8815807978997705e-05,
+      "loss": 0.2514,
+      "step": 13630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8815339177722565e-05,
+      "loss": 0.1731,
+      "step": 13635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8814870376447425e-05,
+      "loss": 0.1958,
+      "step": 13640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8814401575172285e-05,
+      "loss": 0.3905,
+      "step": 13645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8813932773897145e-05,
+      "loss": 0.573,
+      "step": 13650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8813463972622008e-05,
+      "loss": 0.2444,
+      "step": 13655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8812995171346868e-05,
+      "loss": 0.0534,
+      "step": 13660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8812526370071728e-05,
+      "loss": 0.0949,
+      "step": 13665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8812057568796588e-05,
+      "loss": 0.1562,
+      "step": 13670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8811588767521448e-05,
+      "loss": 0.1673,
+      "step": 13675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8811119966246308e-05,
+      "loss": 0.2063,
+      "step": 13680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.881065116497117e-05,
+      "loss": 0.1861,
+      "step": 13685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.881018236369603e-05,
+      "loss": 0.2188,
+      "step": 13690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880971356242089e-05,
+      "loss": 0.3303,
+      "step": 13695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8809244761145754e-05,
+      "loss": 0.6072,
+      "step": 13700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8808775959870614e-05,
+      "loss": 0.2465,
+      "step": 13705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8808307158595474e-05,
+      "loss": 0.0536,
+      "step": 13710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8807838357320334e-05,
+      "loss": 0.0992,
+      "step": 13715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8807369556045194e-05,
+      "loss": 0.1814,
+      "step": 13720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8806900754770054e-05,
+      "loss": 0.1639,
+      "step": 13725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8806431953494914e-05,
+      "loss": 0.174,
+      "step": 13730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8805963152219774e-05,
+      "loss": 0.1975,
+      "step": 13735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8805494350944634e-05,
+      "loss": 0.175,
+      "step": 13740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8805025549669497e-05,
+      "loss": 0.3268,
+      "step": 13745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8804556748394357e-05,
+      "loss": 0.582,
+      "step": 13750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8804087947119217e-05,
+      "loss": 0.2182,
+      "step": 13755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880361914584408e-05,
+      "loss": 0.093,
+      "step": 13760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880315034456894e-05,
+      "loss": 0.1654,
+      "step": 13765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.88026815432938e-05,
+      "loss": 0.0881,
+      "step": 13770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880221274201866e-05,
+      "loss": 0.1605,
+      "step": 13775
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880174394074352e-05,
+      "loss": 0.1614,
+      "step": 13780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880127513946838e-05,
+      "loss": 0.1891,
+      "step": 13785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.880080633819324e-05,
+      "loss": 0.2944,
+      "step": 13790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8800337536918103e-05,
+      "loss": 0.4841,
+      "step": 13795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8799868735642963e-05,
+      "loss": 0.5165,
+      "step": 13800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8799399934367823e-05,
+      "loss": 0.2663,
+      "step": 13805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8798931133092683e-05,
+      "loss": 0.067,
+      "step": 13810
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8798462331817543e-05,
+      "loss": 0.1632,
+      "step": 13815
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8797993530542403e-05,
+      "loss": 0.0878,
+      "step": 13820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8797524729267266e-05,
+      "loss": 0.1335,
+      "step": 13825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8797055927992126e-05,
+      "loss": 0.2923,
+      "step": 13830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8796587126716986e-05,
+      "loss": 0.2697,
+      "step": 13835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879611832544185e-05,
+      "loss": 0.2409,
+      "step": 13840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879564952416671e-05,
+      "loss": 0.3278,
+      "step": 13845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879518072289157e-05,
+      "loss": 0.4615,
+      "step": 13850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879471192161643e-05,
+      "loss": 0.2961,
+      "step": 13855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879424312034129e-05,
+      "loss": 0.1467,
+      "step": 13860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879377431906615e-05,
+      "loss": 0.0899,
+      "step": 13865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879330551779101e-05,
+      "loss": 0.1482,
+      "step": 13870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879283671651587e-05,
+      "loss": 0.1441,
+      "step": 13875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.879236791524073e-05,
+      "loss": 0.1818,
+      "step": 13880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8791899113965592e-05,
+      "loss": 0.2261,
+      "step": 13885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8791430312690452e-05,
+      "loss": 0.2463,
+      "step": 13890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8790961511415312e-05,
+      "loss": 0.2855,
+      "step": 13895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8790492710140172e-05,
+      "loss": 0.5662,
+      "step": 13900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8790023908865035e-05,
+      "loss": 0.217,
+      "step": 13905
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8789555107589895e-05,
+      "loss": 0.1254,
+      "step": 13910
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8789086306314755e-05,
+      "loss": 0.1276,
+      "step": 13915
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8788617505039615e-05,
+      "loss": 0.1244,
+      "step": 13920
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8788148703764475e-05,
+      "loss": 0.1612,
+      "step": 13925
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8787679902489335e-05,
+      "loss": 0.2602,
+      "step": 13930
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.87872111012142e-05,
+      "loss": 0.3629,
+      "step": 13935
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8786742299939058e-05,
+      "loss": 0.266,
+      "step": 13940
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8786273498663918e-05,
+      "loss": 0.3049,
+      "step": 13945
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8785804697388778e-05,
+      "loss": 0.5083,
+      "step": 13950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8785335896113638e-05,
+      "loss": 0.2277,
+      "step": 13955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8784867094838498e-05,
+      "loss": 0.1174,
+      "step": 13960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8784398293563358e-05,
+      "loss": 0.0655,
+      "step": 13965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.878392949228822e-05,
+      "loss": 0.1587,
+      "step": 13970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.878346069101308e-05,
+      "loss": 0.1778,
+      "step": 13975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.878299188973794e-05,
+      "loss": 0.2082,
+      "step": 13980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8782523088462804e-05,
+      "loss": 0.1945,
+      "step": 13985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8782054287187664e-05,
+      "loss": 0.2287,
+      "step": 13990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8781585485912524e-05,
+      "loss": 0.2369,
+      "step": 13995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8781116684637384e-05,
+      "loss": 0.5127,
+      "step": 14000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8780647883362244e-05,
+      "loss": 0.2929,
+      "step": 14005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8780179082087104e-05,
+      "loss": 0.1198,
+      "step": 14010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8779710280811964e-05,
+      "loss": 0.0986,
+      "step": 14015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8779241479536824e-05,
+      "loss": 0.1393,
+      "step": 14020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8778772678261687e-05,
+      "loss": 0.1961,
+      "step": 14025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8778303876986547e-05,
+      "loss": 0.2312,
+      "step": 14030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8777835075711407e-05,
+      "loss": 0.2327,
+      "step": 14035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8777366274436267e-05,
+      "loss": 0.258,
+      "step": 14040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8776897473161127e-05,
+      "loss": 0.4388,
+      "step": 14045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.877642867188599e-05,
+      "loss": 0.5567,
+      "step": 14050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.877595987061085e-05,
+      "loss": 0.222,
+      "step": 14055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.877549106933571e-05,
+      "loss": 0.0841,
+      "step": 14060
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.877502226806057e-05,
+      "loss": 0.0847,
+      "step": 14065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.877455346678543e-05,
+      "loss": 0.181,
+      "step": 14070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8774084665510293e-05,
+      "loss": 0.2742,
+      "step": 14075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8773615864235153e-05,
+      "loss": 0.1208,
+      "step": 14080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8773147062960013e-05,
+      "loss": 0.2701,
+      "step": 14085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8772678261684873e-05,
+      "loss": 0.2529,
+      "step": 14090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8772209460409733e-05,
+      "loss": 0.3087,
+      "step": 14095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8771740659134593e-05,
+      "loss": 0.6273,
+      "step": 14100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8771271857859453e-05,
+      "loss": 0.1981,
+      "step": 14105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8770803056584316e-05,
+      "loss": 0.1016,
+      "step": 14110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8770334255309176e-05,
+      "loss": 0.115,
+      "step": 14115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8769865454034036e-05,
+      "loss": 0.07,
+      "step": 14120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8769396652758896e-05,
+      "loss": 0.1167,
+      "step": 14125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876892785148376e-05,
+      "loss": 0.2139,
+      "step": 14130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876845905020862e-05,
+      "loss": 0.2074,
+      "step": 14135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876799024893348e-05,
+      "loss": 0.173,
+      "step": 14140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876752144765834e-05,
+      "loss": 0.3318,
+      "step": 14145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.87670526463832e-05,
+      "loss": 0.5468,
+      "step": 14150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876658384510806e-05,
+      "loss": 0.242,
+      "step": 14155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876611504383292e-05,
+      "loss": 0.0742,
+      "step": 14160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8765646242557782e-05,
+      "loss": 0.0878,
+      "step": 14165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8765177441282642e-05,
+      "loss": 0.157,
+      "step": 14170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8764708640007502e-05,
+      "loss": 0.1693,
+      "step": 14175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8764239838732362e-05,
+      "loss": 0.1899,
+      "step": 14180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8763771037457222e-05,
+      "loss": 0.2461,
+      "step": 14185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8763302236182085e-05,
+      "loss": 0.2921,
+      "step": 14190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8762833434906945e-05,
+      "loss": 0.285,
+      "step": 14195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8762364633631805e-05,
+      "loss": 0.4061,
+      "step": 14200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8761895832356665e-05,
+      "loss": 0.2432,
+      "step": 14205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876142703108153e-05,
+      "loss": 0.1719,
+      "step": 14210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876095822980639e-05,
+      "loss": 0.1674,
+      "step": 14215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876048942853125e-05,
+      "loss": 0.1117,
+      "step": 14220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.876002062725611e-05,
+      "loss": 0.1181,
+      "step": 14225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.875955182598097e-05,
+      "loss": 0.2447,
+      "step": 14230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8759083024705828e-05,
+      "loss": 0.2381,
+      "step": 14235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8758614223430688e-05,
+      "loss": 0.2788,
+      "step": 14240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8758145422155548e-05,
+      "loss": 0.3321,
+      "step": 14245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8757676620880408e-05,
+      "loss": 0.5291,
+      "step": 14250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.875720781960527e-05,
+      "loss": 0.1678,
+      "step": 14255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.875673901833013e-05,
+      "loss": 0.1219,
+      "step": 14260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.875627021705499e-05,
+      "loss": 0.0905,
+      "step": 14265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8755801415779855e-05,
+      "loss": 0.2237,
+      "step": 14270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8755332614504714e-05,
+      "loss": 0.1641,
+      "step": 14275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8754863813229574e-05,
+      "loss": 0.2348,
+      "step": 14280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8754395011954434e-05,
+      "loss": 0.1618,
+      "step": 14285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8753926210679294e-05,
+      "loss": 0.2344,
+      "step": 14290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8753457409404154e-05,
+      "loss": 0.3558,
+      "step": 14295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8752988608129014e-05,
+      "loss": 0.456,
+      "step": 14300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8752519806853877e-05,
+      "loss": 0.2746,
+      "step": 14305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8752051005578737e-05,
+      "loss": 0.1039,
+      "step": 14310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8751582204303597e-05,
+      "loss": 0.1487,
+      "step": 14315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8751113403028457e-05,
+      "loss": 0.0682,
+      "step": 14320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8750644601753317e-05,
+      "loss": 0.2572,
+      "step": 14325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8750175800478177e-05,
+      "loss": 0.2256,
+      "step": 14330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874970699920304e-05,
+      "loss": 0.2427,
+      "step": 14335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.87492381979279e-05,
+      "loss": 0.1959,
+      "step": 14340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874876939665276e-05,
+      "loss": 0.3314,
+      "step": 14345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8748300595377624e-05,
+      "loss": 0.7407,
+      "step": 14350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8747831794102484e-05,
+      "loss": 0.1726,
+      "step": 14355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8747362992827344e-05,
+      "loss": 0.1146,
+      "step": 14360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8746894191552203e-05,
+      "loss": 0.0943,
+      "step": 14365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8746425390277063e-05,
+      "loss": 0.143,
+      "step": 14370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8745956589001923e-05,
+      "loss": 0.1744,
+      "step": 14375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8745487787726783e-05,
+      "loss": 0.1772,
+      "step": 14380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8745018986451643e-05,
+      "loss": 0.1932,
+      "step": 14385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8744550185176503e-05,
+      "loss": 0.2547,
+      "step": 14390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8744081383901363e-05,
+      "loss": 0.2008,
+      "step": 14395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8743612582626226e-05,
+      "loss": 0.5431,
+      "step": 14400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8743143781351086e-05,
+      "loss": 0.2268,
+      "step": 14405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8742674980075946e-05,
+      "loss": 0.0781,
+      "step": 14410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874220617880081e-05,
+      "loss": 0.1206,
+      "step": 14415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874173737752567e-05,
+      "loss": 0.0975,
+      "step": 14420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874126857625053e-05,
+      "loss": 0.193,
+      "step": 14425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874079977497539e-05,
+      "loss": 0.1506,
+      "step": 14430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.874033097370025e-05,
+      "loss": 0.2245,
+      "step": 14435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.873986217242511e-05,
+      "loss": 0.2452,
+      "step": 14440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8739393371149973e-05,
+      "loss": 0.3057,
+      "step": 14445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8738924569874833e-05,
+      "loss": 0.7368,
+      "step": 14450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8738455768599692e-05,
+      "loss": 0.2164,
+      "step": 14455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8737986967324552e-05,
+      "loss": 0.0852,
+      "step": 14460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8737518166049412e-05,
+      "loss": 0.1504,
+      "step": 14465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8737049364774272e-05,
+      "loss": 0.2015,
+      "step": 14470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8736580563499132e-05,
+      "loss": 0.1602,
+      "step": 14475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8736111762223995e-05,
+      "loss": 0.2252,
+      "step": 14480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8735642960948855e-05,
+      "loss": 0.1856,
+      "step": 14485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8735174159673715e-05,
+      "loss": 0.2775,
+      "step": 14490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873470535839858e-05,
+      "loss": 0.2782,
+      "step": 14495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873423655712344e-05,
+      "loss": 0.5278,
+      "step": 14500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.87337677558483e-05,
+      "loss": 0.2166,
+      "step": 14505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873329895457316e-05,
+      "loss": 0.1283,
+      "step": 14510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873283015329802e-05,
+      "loss": 0.1561,
+      "step": 14515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873236135202288e-05,
+      "loss": 0.1354,
+      "step": 14520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8731892550747738e-05,
+      "loss": 0.1407,
+      "step": 14525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8731423749472598e-05,
+      "loss": 0.2432,
+      "step": 14530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8730954948197458e-05,
+      "loss": 0.2267,
+      "step": 14535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873048614692232e-05,
+      "loss": 0.2128,
+      "step": 14540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.873001734564718e-05,
+      "loss": 0.3205,
+      "step": 14545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.872954854437204e-05,
+      "loss": 0.631,
+      "step": 14550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.87290797430969e-05,
+      "loss": 0.2592,
+      "step": 14555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8728610941821765e-05,
+      "loss": 0.1638,
+      "step": 14560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8728142140546625e-05,
+      "loss": 0.1215,
+      "step": 14565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8727673339271484e-05,
+      "loss": 0.2601,
+      "step": 14570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8727204537996344e-05,
+      "loss": 0.2216,
+      "step": 14575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8726735736721204e-05,
+      "loss": 0.1664,
+      "step": 14580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8726266935446068e-05,
+      "loss": 0.1903,
+      "step": 14585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8725798134170928e-05,
+      "loss": 0.2214,
+      "step": 14590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8725329332895788e-05,
+      "loss": 0.384,
+      "step": 14595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8724860531620647e-05,
+      "loss": 0.6086,
+      "step": 14600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8724391730345507e-05,
+      "loss": 0.1693,
+      "step": 14605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8723922929070367e-05,
+      "loss": 0.0831,
+      "step": 14610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8723454127795227e-05,
+      "loss": 0.1812,
+      "step": 14615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.872298532652009e-05,
+      "loss": 0.1165,
+      "step": 14620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.872251652524495e-05,
+      "loss": 0.1394,
+      "step": 14625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.872204772396981e-05,
+      "loss": 0.3595,
+      "step": 14630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.872157892269467e-05,
+      "loss": 0.2451,
+      "step": 14635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8721110121419534e-05,
+      "loss": 0.2951,
+      "step": 14640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8720641320144394e-05,
+      "loss": 0.4274,
+      "step": 14645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8720172518869254e-05,
+      "loss": 0.5065,
+      "step": 14650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8719703717594113e-05,
+      "loss": 0.2857,
+      "step": 14655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8719234916318973e-05,
+      "loss": 0.0604,
+      "step": 14660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8718766115043833e-05,
+      "loss": 0.1006,
+      "step": 14665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8718297313768693e-05,
+      "loss": 0.1257,
+      "step": 14670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8717828512493557e-05,
+      "loss": 0.2122,
+      "step": 14675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8717359711218417e-05,
+      "loss": 0.2196,
+      "step": 14680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8716890909943276e-05,
+      "loss": 0.2033,
+      "step": 14685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8716422108668136e-05,
+      "loss": 0.2635,
+      "step": 14690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8715953307392996e-05,
+      "loss": 0.3508,
+      "step": 14695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.871548450611786e-05,
+      "loss": 0.5574,
+      "step": 14700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.871501570484272e-05,
+      "loss": 0.177,
+      "step": 14705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.871454690356758e-05,
+      "loss": 0.0636,
+      "step": 14710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.871407810229244e-05,
+      "loss": 0.1045,
+      "step": 14715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.87136093010173e-05,
+      "loss": 0.2398,
+      "step": 14720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8713140499742163e-05,
+      "loss": 0.169,
+      "step": 14725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8712671698467023e-05,
+      "loss": 0.1389,
+      "step": 14730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8712202897191883e-05,
+      "loss": 0.1491,
+      "step": 14735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8711734095916743e-05,
+      "loss": 0.2459,
+      "step": 14740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8711265294641602e-05,
+      "loss": 0.3591,
+      "step": 14745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8710796493366462e-05,
+      "loss": 0.639,
+      "step": 14750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8710327692091322e-05,
+      "loss": 0.1863,
+      "step": 14755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8709858890816182e-05,
+      "loss": 0.113,
+      "step": 14760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8709390089541046e-05,
+      "loss": 0.134,
+      "step": 14765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8708921288265906e-05,
+      "loss": 0.1618,
+      "step": 14770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8708452486990765e-05,
+      "loss": 0.1839,
+      "step": 14775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870798368571563e-05,
+      "loss": 0.2527,
+      "step": 14780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870751488444049e-05,
+      "loss": 0.2761,
+      "step": 14785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870704608316535e-05,
+      "loss": 0.2639,
+      "step": 14790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870657728189021e-05,
+      "loss": 0.3858,
+      "step": 14795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870610848061507e-05,
+      "loss": 0.4675,
+      "step": 14800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870563967933993e-05,
+      "loss": 0.2161,
+      "step": 14805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870517087806479e-05,
+      "loss": 0.0642,
+      "step": 14810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8704702076789652e-05,
+      "loss": 0.1011,
+      "step": 14815
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870423327551451e-05,
+      "loss": 0.1166,
+      "step": 14820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870376447423937e-05,
+      "loss": 0.1504,
+      "step": 14825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870329567296423e-05,
+      "loss": 0.2044,
+      "step": 14830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870282687168909e-05,
+      "loss": 0.1632,
+      "step": 14835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.870235807041395e-05,
+      "loss": 0.3202,
+      "step": 14840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8701889269138815e-05,
+      "loss": 0.3161,
+      "step": 14845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8701420467863675e-05,
+      "loss": 0.5072,
+      "step": 14850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8700951666588535e-05,
+      "loss": 0.1994,
+      "step": 14855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8700482865313394e-05,
+      "loss": 0.1046,
+      "step": 14860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8700014064038258e-05,
+      "loss": 0.1125,
+      "step": 14865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8699545262763118e-05,
+      "loss": 0.1324,
+      "step": 14870
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8699076461487978e-05,
+      "loss": 0.1554,
+      "step": 14875
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8698607660212838e-05,
+      "loss": 0.1562,
+      "step": 14880
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8698138858937698e-05,
+      "loss": 0.1357,
+      "step": 14885
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8697670057662557e-05,
+      "loss": 0.1963,
+      "step": 14890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8697201256387417e-05,
+      "loss": 0.2944,
+      "step": 14895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8696732455112277e-05,
+      "loss": 0.5406,
+      "step": 14900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8696263653837137e-05,
+      "loss": 0.225,
+      "step": 14905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8695794852562e-05,
+      "loss": 0.115,
+      "step": 14910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.869532605128686e-05,
+      "loss": 0.0827,
+      "step": 14915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.869485725001172e-05,
+      "loss": 0.1133,
+      "step": 14920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8694388448736584e-05,
+      "loss": 0.1307,
+      "step": 14925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8693919647461444e-05,
+      "loss": 0.1646,
+      "step": 14930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8693450846186304e-05,
+      "loss": 0.1663,
+      "step": 14935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8692982044911164e-05,
+      "loss": 0.208,
+      "step": 14940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8692513243636024e-05,
+      "loss": 0.3248,
+      "step": 14945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8692044442360883e-05,
+      "loss": 0.539,
+      "step": 14950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8691575641085747e-05,
+      "loss": 0.2796,
+      "step": 14955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8691106839810607e-05,
+      "loss": 0.0703,
+      "step": 14960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8690638038535467e-05,
+      "loss": 0.1278,
+      "step": 14965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8690169237260327e-05,
+      "loss": 0.067,
+      "step": 14970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8689700435985187e-05,
+      "loss": 0.192,
+      "step": 14975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8689231634710046e-05,
+      "loss": 0.2147,
+      "step": 14980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8688762833434906e-05,
+      "loss": 0.2016,
+      "step": 14985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868829403215977e-05,
+      "loss": 0.2057,
+      "step": 14990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868782523088463e-05,
+      "loss": 0.2402,
+      "step": 14995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868735642960949e-05,
+      "loss": 0.5123,
+      "step": 15000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8686887628334353e-05,
+      "loss": 0.2725,
+      "step": 15005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8686418827059213e-05,
+      "loss": 0.0648,
+      "step": 15010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8685950025784073e-05,
+      "loss": 0.1405,
+      "step": 15015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8685481224508933e-05,
+      "loss": 0.0819,
+      "step": 15020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8685012423233793e-05,
+      "loss": 0.1102,
+      "step": 15025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8684543621958653e-05,
+      "loss": 0.2164,
+      "step": 15030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8684074820683512e-05,
+      "loss": 0.2653,
+      "step": 15035
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8683606019408372e-05,
+      "loss": 0.338,
+      "step": 15040
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8683137218133232e-05,
+      "loss": 0.3184,
+      "step": 15045
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8682668416858096e-05,
+      "loss": 0.4224,
+      "step": 15050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8682199615582956e-05,
+      "loss": 0.2217,
+      "step": 15055
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8681730814307816e-05,
+      "loss": 0.1586,
+      "step": 15060
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8681262013032675e-05,
+      "loss": 0.0931,
+      "step": 15065
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.868079321175754e-05,
+      "loss": 0.187,
+      "step": 15070
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.86803244104824e-05,
+      "loss": 0.1737,
+      "step": 15075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.867985560920726e-05,
+      "loss": 0.2213,
+      "step": 15080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.867938680793212e-05,
+      "loss": 0.1752,
+      "step": 15085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.867891800665698e-05,
+      "loss": 0.2712,
+      "step": 15090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8678449205381842e-05,
+      "loss": 0.4204,
+      "step": 15095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8677980404106702e-05,
+      "loss": 0.5598,
+      "step": 15100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8677511602831562e-05,
+      "loss": 0.2795,
+      "step": 15105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.867704280155642e-05,
+      "loss": 0.0626,
+      "step": 15110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.867657400028128e-05,
+      "loss": 0.062,
+      "step": 15115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.867610519900614e-05,
+      "loss": 0.0722,
+      "step": 15120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8675636397731e-05,
+      "loss": 0.1323,
+      "step": 15125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8675167596455865e-05,
+      "loss": 0.2438,
+      "step": 15130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8674698795180725e-05,
+      "loss": 0.1662,
+      "step": 15135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8674229993905585e-05,
+      "loss": 0.3017,
+      "step": 15140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8673761192630448e-05,
+      "loss": 0.2221,
+      "step": 15145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8673292391355308e-05,
+      "loss": 0.4873,
+      "step": 15150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8672823590080168e-05,
+      "loss": 0.2781,
+      "step": 15155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8672354788805028e-05,
+      "loss": 0.0843,
+      "step": 15160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8671885987529888e-05,
+      "loss": 0.1082,
+      "step": 15165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8671417186254748e-05,
+      "loss": 0.1206,
+      "step": 15170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8670948384979608e-05,
+      "loss": 0.1604,
+      "step": 15175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8670479583704468e-05,
+      "loss": 0.1387,
+      "step": 15180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8670010782429327e-05,
+      "loss": 0.256,
+      "step": 15185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866954198115419e-05,
+      "loss": 0.2167,
+      "step": 15190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866907317987905e-05,
+      "loss": 0.3426,
+      "step": 15195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866860437860391e-05,
+      "loss": 0.5178,
+      "step": 15200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866813557732877e-05,
+      "loss": 0.2076,
+      "step": 15205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8667666776053634e-05,
+      "loss": 0.0497,
+      "step": 15210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8667197974778494e-05,
+      "loss": 0.1064,
+      "step": 15215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8666729173503354e-05,
+      "loss": 0.2208,
+      "step": 15220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8666260372228214e-05,
+      "loss": 0.1425,
+      "step": 15225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8665791570953074e-05,
+      "loss": 0.1764,
+      "step": 15230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8665322769677937e-05,
+      "loss": 0.1408,
+      "step": 15235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8664853968402797e-05,
+      "loss": 0.1988,
+      "step": 15240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8664385167127657e-05,
+      "loss": 0.3279,
+      "step": 15245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8663916365852517e-05,
+      "loss": 0.4128,
+      "step": 15250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8663447564577377e-05,
+      "loss": 0.2538,
+      "step": 15255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8662978763302237e-05,
+      "loss": 0.101,
+      "step": 15260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8662509962027097e-05,
+      "loss": 0.1287,
+      "step": 15265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8662041160751956e-05,
+      "loss": 0.1088,
+      "step": 15270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866157235947682e-05,
+      "loss": 0.1514,
+      "step": 15275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866110355820168e-05,
+      "loss": 0.2464,
+      "step": 15280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.866063475692654e-05,
+      "loss": 0.2969,
+      "step": 15285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8660165955651403e-05,
+      "loss": 0.2988,
+      "step": 15290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8659697154376263e-05,
+      "loss": 0.3855,
+      "step": 15295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8659228353101123e-05,
+      "loss": 0.6949,
+      "step": 15300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8658759551825983e-05,
+      "loss": 0.2056,
+      "step": 15305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8658290750550843e-05,
+      "loss": 0.0848,
+      "step": 15310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8657821949275703e-05,
+      "loss": 0.1188,
+      "step": 15315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8657353148000563e-05,
+      "loss": 0.0881,
+      "step": 15320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8656884346725426e-05,
+      "loss": 0.1742,
+      "step": 15325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8656415545450286e-05,
+      "loss": 0.1222,
+      "step": 15330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8655946744175146e-05,
+      "loss": 0.1678,
+      "step": 15335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8655477942900006e-05,
+      "loss": 0.1992,
+      "step": 15340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8655009141624866e-05,
+      "loss": 0.3601,
+      "step": 15345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8654540340349726e-05,
+      "loss": 0.5375,
+      "step": 15350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865407153907459e-05,
+      "loss": 0.2414,
+      "step": 15355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865360273779945e-05,
+      "loss": 0.0649,
+      "step": 15360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865313393652431e-05,
+      "loss": 0.0756,
+      "step": 15365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865266513524917e-05,
+      "loss": 0.1562,
+      "step": 15370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8652196333974032e-05,
+      "loss": 0.1215,
+      "step": 15375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8651727532698892e-05,
+      "loss": 0.1505,
+      "step": 15380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8651258731423752e-05,
+      "loss": 0.197,
+      "step": 15385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8650789930148612e-05,
+      "loss": 0.2759,
+      "step": 15390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8650321128873472e-05,
+      "loss": 0.2041,
+      "step": 15395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8649852327598332e-05,
+      "loss": 0.4761,
+      "step": 15400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864938352632319e-05,
+      "loss": 0.1988,
+      "step": 15405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864891472504805e-05,
+      "loss": 0.1376,
+      "step": 15410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864844592377291e-05,
+      "loss": 0.0872,
+      "step": 15415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8647977122497775e-05,
+      "loss": 0.0863,
+      "step": 15420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8647508321222635e-05,
+      "loss": 0.135,
+      "step": 15425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8647039519947495e-05,
+      "loss": 0.0906,
+      "step": 15430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8646570718672358e-05,
+      "loss": 0.2675,
+      "step": 15435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8646101917397218e-05,
+      "loss": 0.1789,
+      "step": 15440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8645633116122078e-05,
+      "loss": 0.3858,
+      "step": 15445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8645164314846938e-05,
+      "loss": 0.6604,
+      "step": 15450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8644695513571798e-05,
+      "loss": 0.1742,
+      "step": 15455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8644226712296658e-05,
+      "loss": 0.1072,
+      "step": 15460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864375791102152e-05,
+      "loss": 0.1201,
+      "step": 15465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864328910974638e-05,
+      "loss": 0.1129,
+      "step": 15470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864282030847124e-05,
+      "loss": 0.073,
+      "step": 15475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.86423515071961e-05,
+      "loss": 0.2668,
+      "step": 15480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864188270592096e-05,
+      "loss": 0.2092,
+      "step": 15485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.864141390464582e-05,
+      "loss": 0.252,
+      "step": 15490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8640945103370684e-05,
+      "loss": 0.2868,
+      "step": 15495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8640476302095544e-05,
+      "loss": 0.6121,
+      "step": 15500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8640007500820404e-05,
+      "loss": 0.1869,
+      "step": 15505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8639538699545264e-05,
+      "loss": 0.0788,
+      "step": 15510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8639069898270127e-05,
+      "loss": 0.0961,
+      "step": 15515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8638601096994987e-05,
+      "loss": 0.1343,
+      "step": 15520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8638132295719847e-05,
+      "loss": 0.1213,
+      "step": 15525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8637663494444707e-05,
+      "loss": 0.2465,
+      "step": 15530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8637194693169567e-05,
+      "loss": 0.2933,
+      "step": 15535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8636725891894427e-05,
+      "loss": 0.205,
+      "step": 15540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8636257090619287e-05,
+      "loss": 0.3036,
+      "step": 15545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8635788289344147e-05,
+      "loss": 0.4355,
+      "step": 15550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8635319488069007e-05,
+      "loss": 0.2686,
+      "step": 15555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.863485068679387e-05,
+      "loss": 0.0397,
+      "step": 15560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.863438188551873e-05,
+      "loss": 0.2227,
+      "step": 15565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.863391308424359e-05,
+      "loss": 0.1172,
+      "step": 15570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8633444282968453e-05,
+      "loss": 0.1019,
+      "step": 15575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8632975481693313e-05,
+      "loss": 0.184,
+      "step": 15580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8632506680418173e-05,
+      "loss": 0.2066,
+      "step": 15585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8632037879143033e-05,
+      "loss": 0.1915,
+      "step": 15590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8631569077867893e-05,
+      "loss": 0.2898,
+      "step": 15595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8631100276592753e-05,
+      "loss": 0.4485,
+      "step": 15600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8630631475317616e-05,
+      "loss": 0.2207,
+      "step": 15605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8630162674042476e-05,
+      "loss": 0.1243,
+      "step": 15610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8629693872767336e-05,
+      "loss": 0.0872,
+      "step": 15615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8629225071492196e-05,
+      "loss": 0.1143,
+      "step": 15620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8628756270217056e-05,
+      "loss": 0.065,
+      "step": 15625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8628287468941916e-05,
+      "loss": 0.2135,
+      "step": 15630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8627818667666776e-05,
+      "loss": 0.2308,
+      "step": 15635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862734986639164e-05,
+      "loss": 0.2845,
+      "step": 15640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.86268810651165e-05,
+      "loss": 0.3182,
+      "step": 15645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862641226384136e-05,
+      "loss": 0.6225,
+      "step": 15650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8625943462566222e-05,
+      "loss": 0.2644,
+      "step": 15655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8625474661291082e-05,
+      "loss": 0.086,
+      "step": 15660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8625005860015942e-05,
+      "loss": 0.157,
+      "step": 15665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8624537058740802e-05,
+      "loss": 0.1244,
+      "step": 15670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8624068257465662e-05,
+      "loss": 0.1208,
+      "step": 15675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8623599456190522e-05,
+      "loss": 0.1678,
+      "step": 15680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8623130654915382e-05,
+      "loss": 0.2024,
+      "step": 15685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8622661853640242e-05,
+      "loss": 0.3057,
+      "step": 15690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.86221930523651e-05,
+      "loss": 0.2436,
+      "step": 15695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8621724251089965e-05,
+      "loss": 0.4505,
+      "step": 15700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8621255449814825e-05,
+      "loss": 0.2134,
+      "step": 15705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8620786648539685e-05,
+      "loss": 0.0503,
+      "step": 15710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8620317847264545e-05,
+      "loss": 0.103,
+      "step": 15715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8619849045989408e-05,
+      "loss": 0.132,
+      "step": 15720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8619380244714268e-05,
+      "loss": 0.2023,
+      "step": 15725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8618911443439128e-05,
+      "loss": 0.1633,
+      "step": 15730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8618442642163988e-05,
+      "loss": 0.2361,
+      "step": 15735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8617973840888848e-05,
+      "loss": 0.2142,
+      "step": 15740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.861750503961371e-05,
+      "loss": 0.2203,
+      "step": 15745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.861703623833857e-05,
+      "loss": 0.4832,
+      "step": 15750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861656743706343e-05,
+      "loss": 0.171,
+      "step": 15755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861609863578829e-05,
+      "loss": 0.0748,
+      "step": 15760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861562983451315e-05,
+      "loss": 0.0726,
+      "step": 15765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861516103323801e-05,
+      "loss": 0.1165,
+      "step": 15770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861469223196287e-05,
+      "loss": 0.1643,
+      "step": 15775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.861422343068773e-05,
+      "loss": 0.2114,
+      "step": 15780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8613754629412594e-05,
+      "loss": 0.1382,
+      "step": 15785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8613285828137454e-05,
+      "loss": 0.1504,
+      "step": 15790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8612817026862314e-05,
+      "loss": 0.3677,
+      "step": 15795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8612348225587177e-05,
+      "loss": 0.7503,
+      "step": 15800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8611879424312037e-05,
+      "loss": 0.1473,
+      "step": 15805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8611410623036897e-05,
+      "loss": 0.0413,
+      "step": 15810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8610941821761757e-05,
+      "loss": 0.1276,
+      "step": 15815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8610473020486617e-05,
+      "loss": 0.12,
+      "step": 15820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8610004219211477e-05,
+      "loss": 0.1935,
+      "step": 15825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8609535417936337e-05,
+      "loss": 0.1424,
+      "step": 15830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8609066616661197e-05,
+      "loss": 0.2974,
+      "step": 15835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860859781538606e-05,
+      "loss": 0.1485,
+      "step": 15840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860812901411092e-05,
+      "loss": 0.3909,
+      "step": 15845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860766021283578e-05,
+      "loss": 0.4972,
+      "step": 15850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860719141156064e-05,
+      "loss": 0.2334,
+      "step": 15855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.86067226102855e-05,
+      "loss": 0.059,
+      "step": 15860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8606253809010363e-05,
+      "loss": 0.0805,
+      "step": 15865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8605785007735223e-05,
+      "loss": 0.1012,
+      "step": 15870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8605316206460083e-05,
+      "loss": 0.0957,
+      "step": 15875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8604847405184943e-05,
+      "loss": 0.1382,
+      "step": 15880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8604378603909806e-05,
+      "loss": 0.2678,
+      "step": 15885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8603909802634666e-05,
+      "loss": 0.2458,
+      "step": 15890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8603441001359526e-05,
+      "loss": 0.3033,
+      "step": 15895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8602972200084386e-05,
+      "loss": 0.5778,
+      "step": 15900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8602503398809246e-05,
+      "loss": 0.2027,
+      "step": 15905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8602034597534106e-05,
+      "loss": 0.0999,
+      "step": 15910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8601565796258966e-05,
+      "loss": 0.0957,
+      "step": 15915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8601096994983826e-05,
+      "loss": 0.2037,
+      "step": 15920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860062819370869e-05,
+      "loss": 0.1079,
+      "step": 15925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.860015939243355e-05,
+      "loss": 0.1333,
+      "step": 15930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.859969059115841e-05,
+      "loss": 0.2305,
+      "step": 15935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.859922178988327e-05,
+      "loss": 0.2528,
+      "step": 15940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8598752988608132e-05,
+      "loss": 0.3822,
+      "step": 15945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8598284187332992e-05,
+      "loss": 0.5118,
+      "step": 15950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8597815386057852e-05,
+      "loss": 0.3077,
+      "step": 15955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8597346584782712e-05,
+      "loss": 0.0664,
+      "step": 15960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8596877783507572e-05,
+      "loss": 0.1011,
+      "step": 15965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8596408982232432e-05,
+      "loss": 0.1759,
+      "step": 15970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8595940180957295e-05,
+      "loss": 0.1561,
+      "step": 15975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8595471379682155e-05,
+      "loss": 0.1241,
+      "step": 15980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8595002578407015e-05,
+      "loss": 0.1386,
+      "step": 15985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8594533777131875e-05,
+      "loss": 0.2541,
+      "step": 15990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8594064975856735e-05,
+      "loss": 0.2858,
+      "step": 15995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8593596174581595e-05,
+      "loss": 0.552,
+      "step": 16000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8593127373306458e-05,
+      "loss": 0.2186,
+      "step": 16005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8592658572031318e-05,
+      "loss": 0.0791,
+      "step": 16010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8592189770756178e-05,
+      "loss": 0.1378,
+      "step": 16015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8591720969481038e-05,
+      "loss": 0.202,
+      "step": 16020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.85912521682059e-05,
+      "loss": 0.1665,
+      "step": 16025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.859078336693076e-05,
+      "loss": 0.2158,
+      "step": 16030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.859031456565562e-05,
+      "loss": 0.2452,
+      "step": 16035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858984576438048e-05,
+      "loss": 0.1764,
+      "step": 16040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858937696310534e-05,
+      "loss": 0.3921,
+      "step": 16045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.85889081618302e-05,
+      "loss": 0.5582,
+      "step": 16050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858843936055506e-05,
+      "loss": 0.1606,
+      "step": 16055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858797055927992e-05,
+      "loss": 0.089,
+      "step": 16060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858750175800478e-05,
+      "loss": 0.0914,
+      "step": 16065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8587032956729644e-05,
+      "loss": 0.1239,
+      "step": 16070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8586564155454504e-05,
+      "loss": 0.1263,
+      "step": 16075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8586095354179364e-05,
+      "loss": 0.1562,
+      "step": 16080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8585626552904227e-05,
+      "loss": 0.3022,
+      "step": 16085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8585157751629087e-05,
+      "loss": 0.2022,
+      "step": 16090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8584688950353947e-05,
+      "loss": 0.3969,
+      "step": 16095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8584220149078807e-05,
+      "loss": 0.5977,
+      "step": 16100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8583751347803667e-05,
+      "loss": 0.2297,
+      "step": 16105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8583282546528527e-05,
+      "loss": 0.0994,
+      "step": 16110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858281374525339e-05,
+      "loss": 0.0737,
+      "step": 16115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858234494397825e-05,
+      "loss": 0.097,
+      "step": 16120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858187614270311e-05,
+      "loss": 0.1379,
+      "step": 16125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858140734142797e-05,
+      "loss": 0.1508,
+      "step": 16130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858093854015283e-05,
+      "loss": 0.2133,
+      "step": 16135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858046973887769e-05,
+      "loss": 0.2874,
+      "step": 16140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.858000093760255e-05,
+      "loss": 0.2308,
+      "step": 16145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8579532136327413e-05,
+      "loss": 0.5109,
+      "step": 16150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8579063335052273e-05,
+      "loss": 0.2218,
+      "step": 16155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8578594533777133e-05,
+      "loss": 0.1128,
+      "step": 16160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8578125732501996e-05,
+      "loss": 0.1386,
+      "step": 16165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8577656931226856e-05,
+      "loss": 0.1299,
+      "step": 16170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8577188129951716e-05,
+      "loss": 0.1277,
+      "step": 16175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8576719328676576e-05,
+      "loss": 0.201,
+      "step": 16180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8576250527401436e-05,
+      "loss": 0.2323,
+      "step": 16185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8575781726126296e-05,
+      "loss": 0.2772,
+      "step": 16190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8575312924851156e-05,
+      "loss": 0.3258,
+      "step": 16195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8574844123576016e-05,
+      "loss": 0.3383,
+      "step": 16200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8574375322300876e-05,
+      "loss": 0.2396,
+      "step": 16205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.857390652102574e-05,
+      "loss": 0.0677,
+      "step": 16210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.85734377197506e-05,
+      "loss": 0.105,
+      "step": 16215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.857296891847546e-05,
+      "loss": 0.1804,
+      "step": 16220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.857250011720032e-05,
+      "loss": 0.1835,
+      "step": 16225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8572031315925182e-05,
+      "loss": 0.1532,
+      "step": 16230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8571562514650042e-05,
+      "loss": 0.2916,
+      "step": 16235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8571093713374902e-05,
+      "loss": 0.3538,
+      "step": 16240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8570624912099762e-05,
+      "loss": 0.3102,
+      "step": 16245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8570156110824622e-05,
+      "loss": 0.5525,
+      "step": 16250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8569687309549485e-05,
+      "loss": 0.2395,
+      "step": 16255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8569218508274345e-05,
+      "loss": 0.0512,
+      "step": 16260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8568749706999205e-05,
+      "loss": 0.0791,
+      "step": 16265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8568280905724065e-05,
+      "loss": 0.139,
+      "step": 16270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8567812104448925e-05,
+      "loss": 0.1549,
+      "step": 16275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8567343303173785e-05,
+      "loss": 0.2231,
+      "step": 16280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8566874501898645e-05,
+      "loss": 0.2823,
+      "step": 16285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8566405700623505e-05,
+      "loss": 0.2111,
+      "step": 16290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8565936899348368e-05,
+      "loss": 0.3525,
+      "step": 16295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8565468098073228e-05,
+      "loss": 0.4608,
+      "step": 16300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8564999296798088e-05,
+      "loss": 0.1706,
+      "step": 16305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856453049552295e-05,
+      "loss": 0.1103,
+      "step": 16310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856406169424781e-05,
+      "loss": 0.0674,
+      "step": 16315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856359289297267e-05,
+      "loss": 0.0725,
+      "step": 16320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856312409169753e-05,
+      "loss": 0.0853,
+      "step": 16325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856265529042239e-05,
+      "loss": 0.2292,
+      "step": 16330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856218648914725e-05,
+      "loss": 0.2606,
+      "step": 16335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856171768787211e-05,
+      "loss": 0.2461,
+      "step": 16340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.856124888659697e-05,
+      "loss": 0.3852,
+      "step": 16345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8560780085321834e-05,
+      "loss": 0.5247,
+      "step": 16350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8560311284046694e-05,
+      "loss": 0.2266,
+      "step": 16355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8559842482771554e-05,
+      "loss": 0.0598,
+      "step": 16360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8559373681496414e-05,
+      "loss": 0.1304,
+      "step": 16365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8558904880221274e-05,
+      "loss": 0.106,
+      "step": 16370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8558436078946137e-05,
+      "loss": 0.1039,
+      "step": 16375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8557967277670997e-05,
+      "loss": 0.1404,
+      "step": 16380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8557498476395857e-05,
+      "loss": 0.2072,
+      "step": 16385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8557029675120717e-05,
+      "loss": 0.2344,
+      "step": 16390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855656087384558e-05,
+      "loss": 0.3281,
+      "step": 16395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855609207257044e-05,
+      "loss": 0.5027,
+      "step": 16400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.85556232712953e-05,
+      "loss": 0.1777,
+      "step": 16405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855515447002016e-05,
+      "loss": 0.0762,
+      "step": 16410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855468566874502e-05,
+      "loss": 0.0743,
+      "step": 16415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855421686746988e-05,
+      "loss": 0.0928,
+      "step": 16420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.855374806619474e-05,
+      "loss": 0.1213,
+      "step": 16425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.85532792649196e-05,
+      "loss": 0.1962,
+      "step": 16430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8552810463644463e-05,
+      "loss": 0.2512,
+      "step": 16435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8552341662369323e-05,
+      "loss": 0.2477,
+      "step": 16440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8551872861094183e-05,
+      "loss": 0.2721,
+      "step": 16445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8551404059819043e-05,
+      "loss": 0.4693,
+      "step": 16450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8550935258543906e-05,
+      "loss": 0.2471,
+      "step": 16455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8550466457268766e-05,
+      "loss": 0.0744,
+      "step": 16460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8549997655993626e-05,
+      "loss": 0.0537,
+      "step": 16465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8549528854718486e-05,
+      "loss": 0.1404,
+      "step": 16470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8549060053443346e-05,
+      "loss": 0.0935,
+      "step": 16475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8548591252168206e-05,
+      "loss": 0.1698,
+      "step": 16480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8548122450893066e-05,
+      "loss": 0.2063,
+      "step": 16485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854765364961793e-05,
+      "loss": 0.2454,
+      "step": 16490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854718484834279e-05,
+      "loss": 0.3434,
+      "step": 16495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854671604706765e-05,
+      "loss": 0.3791,
+      "step": 16500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854624724579251e-05,
+      "loss": 0.2235,
+      "step": 16505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.854577844451737e-05,
+      "loss": 0.0541,
+      "step": 16510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8545309643242232e-05,
+      "loss": 0.0699,
+      "step": 16515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8544840841967092e-05,
+      "loss": 0.0983,
+      "step": 16520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8544372040691952e-05,
+      "loss": 0.1524,
+      "step": 16525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8543903239416812e-05,
+      "loss": 0.2209,
+      "step": 16530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8543434438141676e-05,
+      "loss": 0.2845,
+      "step": 16535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8542965636866535e-05,
+      "loss": 0.1833,
+      "step": 16540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8542496835591395e-05,
+      "loss": 0.2518,
+      "step": 16545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8542028034316255e-05,
+      "loss": 0.4805,
+      "step": 16550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8541559233041115e-05,
+      "loss": 0.2084,
+      "step": 16555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8541090431765975e-05,
+      "loss": 0.0975,
+      "step": 16560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8540621630490835e-05,
+      "loss": 0.0754,
+      "step": 16565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8540152829215695e-05,
+      "loss": 0.1748,
+      "step": 16570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8539684027940555e-05,
+      "loss": 0.1479,
+      "step": 16575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853921522666542e-05,
+      "loss": 0.1724,
+      "step": 16580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8538746425390278e-05,
+      "loss": 0.1805,
+      "step": 16585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8538277624115138e-05,
+      "loss": 0.233,
+      "step": 16590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853780882284e-05,
+      "loss": 0.2265,
+      "step": 16595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853734002156486e-05,
+      "loss": 0.4862,
+      "step": 16600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.853687122028972e-05,
+      "loss": 0.2111,
+      "step": 16605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.853640241901458e-05,
+      "loss": 0.0882,
+      "step": 16610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.853593361773944e-05,
+      "loss": 0.1286,
+      "step": 16615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.85354648164643e-05,
+      "loss": 0.0689,
+      "step": 16620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8534996015189165e-05,
+      "loss": 0.0597,
+      "step": 16625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8534527213914024e-05,
+      "loss": 0.1602,
+      "step": 16630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8534058412638884e-05,
+      "loss": 0.2968,
+      "step": 16635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8533589611363744e-05,
+      "loss": 0.3255,
+      "step": 16640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8533120810088604e-05,
+      "loss": 0.3148,
+      "step": 16645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8532652008813464e-05,
+      "loss": 0.4054,
+      "step": 16650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8532183207538324e-05,
+      "loss": 0.2527,
+      "step": 16655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8531714406263187e-05,
+      "loss": 0.0541,
+      "step": 16660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8531245604988047e-05,
+      "loss": 0.1128,
+      "step": 16665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8530776803712907e-05,
+      "loss": 0.1116,
+      "step": 16670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.853030800243777e-05,
+      "loss": 0.1365,
+      "step": 16675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852983920116263e-05,
+      "loss": 0.1894,
+      "step": 16680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852937039988749e-05,
+      "loss": 0.3003,
+      "step": 16685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852890159861235e-05,
+      "loss": 0.1914,
+      "step": 16690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852843279733721e-05,
+      "loss": 0.3021,
+      "step": 16695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852796399606207e-05,
+      "loss": 0.3965,
+      "step": 16700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852749519478693e-05,
+      "loss": 0.2011,
+      "step": 16705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852702639351179e-05,
+      "loss": 0.0675,
+      "step": 16710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852655759223665e-05,
+      "loss": 0.1059,
+      "step": 16715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8526088790961513e-05,
+      "loss": 0.1146,
+      "step": 16720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8525619989686373e-05,
+      "loss": 0.1235,
+      "step": 16725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8525151188411233e-05,
+      "loss": 0.1721,
+      "step": 16730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8524682387136093e-05,
+      "loss": 0.1825,
+      "step": 16735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8524213585860957e-05,
+      "loss": 0.2841,
+      "step": 16740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8523744784585816e-05,
+      "loss": 0.3064,
+      "step": 16745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8523275983310676e-05,
+      "loss": 0.4229,
+      "step": 16750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8522807182035536e-05,
+      "loss": 0.2078,
+      "step": 16755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8522338380760396e-05,
+      "loss": 0.0681,
+      "step": 16760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852186957948526e-05,
+      "loss": 0.1168,
+      "step": 16765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852140077821012e-05,
+      "loss": 0.0938,
+      "step": 16770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852093197693498e-05,
+      "loss": 0.1566,
+      "step": 16775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.852046317565984e-05,
+      "loss": 0.1815,
+      "step": 16780
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.85199943743847e-05,
+      "loss": 0.2747,
+      "step": 16785
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.851952557310956e-05,
+      "loss": 0.1572,
+      "step": 16790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.851905677183442e-05,
+      "loss": 0.4039,
+      "step": 16795
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.851858797055928e-05,
+      "loss": 0.5064,
+      "step": 16800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8518119169284142e-05,
+      "loss": 0.2327,
+      "step": 16805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8517650368009002e-05,
+      "loss": 0.055,
+      "step": 16810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8517181566733862e-05,
+      "loss": 0.0713,
+      "step": 16815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8516712765458726e-05,
+      "loss": 0.1341,
+      "step": 16820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8516243964183586e-05,
+      "loss": 0.2755,
+      "step": 16825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8515775162908446e-05,
+      "loss": 0.1865,
+      "step": 16830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8515306361633305e-05,
+      "loss": 0.2744,
+      "step": 16835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8514837560358165e-05,
+      "loss": 0.1419,
+      "step": 16840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8514368759083025e-05,
+      "loss": 0.3494,
+      "step": 16845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8513899957807885e-05,
+      "loss": 0.4996,
+      "step": 16850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8513431156532745e-05,
+      "loss": 0.1462,
+      "step": 16855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851296235525761e-05,
+      "loss": 0.1193,
+      "step": 16860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851249355398247e-05,
+      "loss": 0.1351,
+      "step": 16865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851202475270733e-05,
+      "loss": 0.0935,
+      "step": 16870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851155595143219e-05,
+      "loss": 0.1988,
+      "step": 16875
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8511087150157048e-05,
+      "loss": 0.1656,
+      "step": 16880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851061834888191e-05,
+      "loss": 0.2114,
+      "step": 16885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.851014954760677e-05,
+      "loss": 0.2165,
+      "step": 16890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850968074633163e-05,
+      "loss": 0.2662,
+      "step": 16895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850921194505649e-05,
+      "loss": 0.694,
+      "step": 16900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8508743143781355e-05,
+      "loss": 0.2216,
+      "step": 16905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8508274342506215e-05,
+      "loss": 0.0824,
+      "step": 16910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8507805541231075e-05,
+      "loss": 0.0962,
+      "step": 16915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8507336739955934e-05,
+      "loss": 0.1082,
+      "step": 16920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8506867938680794e-05,
+      "loss": 0.1578,
+      "step": 16925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8506399137405654e-05,
+      "loss": 0.1898,
+      "step": 16930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8505930336130514e-05,
+      "loss": 0.1983,
+      "step": 16935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8505461534855374e-05,
+      "loss": 0.2282,
+      "step": 16940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8504992733580238e-05,
+      "loss": 0.2615,
+      "step": 16945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8504523932305097e-05,
+      "loss": 0.4631,
+      "step": 16950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8504055131029957e-05,
+      "loss": 0.2476,
+      "step": 16955
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850358632975482e-05,
+      "loss": 0.0629,
+      "step": 16960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850311752847968e-05,
+      "loss": 0.0986,
+      "step": 16965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850264872720454e-05,
+      "loss": 0.1212,
+      "step": 16970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.85021799259294e-05,
+      "loss": 0.127,
+      "step": 16975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850171112465426e-05,
+      "loss": 0.2048,
+      "step": 16980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850124232337912e-05,
+      "loss": 0.2524,
+      "step": 16985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850077352210398e-05,
+      "loss": 0.2069,
+      "step": 16990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.850030472082884e-05,
+      "loss": 0.2817,
+      "step": 16995
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8499835919553704e-05,
+      "loss": 0.5592,
+      "step": 17000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8499367118278564e-05,
+      "loss": 0.2087,
+      "step": 17005
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8498898317003423e-05,
+      "loss": 0.1038,
+      "step": 17010
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8498429515728283e-05,
+      "loss": 0.1234,
+      "step": 17015
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8497960714453143e-05,
+      "loss": 0.1361,
+      "step": 17020
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8497491913178007e-05,
+      "loss": 0.1025,
+      "step": 17025
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8497023111902867e-05,
+      "loss": 0.1794,
+      "step": 17030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8496554310627727e-05,
+      "loss": 0.2281,
+      "step": 17035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8496085509352586e-05,
+      "loss": 0.2468,
+      "step": 17040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849561670807745e-05,
+      "loss": 0.3502,
+      "step": 17045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849514790680231e-05,
+      "loss": 0.5949,
+      "step": 17050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849467910552717e-05,
+      "loss": 0.2099,
+      "step": 17055
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849421030425203e-05,
+      "loss": 0.0767,
+      "step": 17060
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849374150297689e-05,
+      "loss": 0.142,
+      "step": 17065
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849327270170175e-05,
+      "loss": 0.1833,
+      "step": 17070
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849280390042661e-05,
+      "loss": 0.0711,
+      "step": 17075
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849233509915147e-05,
+      "loss": 0.1719,
+      "step": 17080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.849186629787633e-05,
+      "loss": 0.25,
+      "step": 17085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8491397496601193e-05,
+      "loss": 0.1838,
+      "step": 17090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8490928695326052e-05,
+      "loss": 0.3021,
+      "step": 17095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8490459894050912e-05,
+      "loss": 0.5295,
+      "step": 17100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8489991092775776e-05,
+      "loss": 0.2923,
+      "step": 17105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8489522291500636e-05,
+      "loss": 0.0491,
+      "step": 17110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8489053490225496e-05,
+      "loss": 0.0952,
+      "step": 17115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8488584688950356e-05,
+      "loss": 0.0852,
+      "step": 17120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8488115887675215e-05,
+      "loss": 0.1166,
+      "step": 17125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8487647086400075e-05,
+      "loss": 0.1835,
+      "step": 17130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8487178285124935e-05,
+      "loss": 0.2004,
+      "step": 17135
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.84867094838498e-05,
+      "loss": 0.2276,
+      "step": 17140
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848624068257466e-05,
+      "loss": 0.3723,
+      "step": 17145
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848577188129952e-05,
+      "loss": 0.4231,
+      "step": 17150
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848530308002438e-05,
+      "loss": 0.2236,
+      "step": 17155
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848483427874924e-05,
+      "loss": 0.0599,
+      "step": 17160
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.84843654774741e-05,
+      "loss": 0.1565,
+      "step": 17165
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848389667619896e-05,
+      "loss": 0.1829,
+      "step": 17170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848342787492382e-05,
+      "loss": 0.111,
+      "step": 17175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.848295907364868e-05,
+      "loss": 0.1442,
+      "step": 17180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8482490272373545e-05,
+      "loss": 0.2144,
+      "step": 17185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8482021471098405e-05,
+      "loss": 0.2541,
+      "step": 17190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8481552669823265e-05,
+      "loss": 0.2623,
+      "step": 17195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8481083868548125e-05,
+      "loss": 0.4617,
+      "step": 17200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8480615067272985e-05,
+      "loss": 0.2354,
+      "step": 17205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8480146265997845e-05,
+      "loss": 0.1225,
+      "step": 17210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8479677464722704e-05,
+      "loss": 0.0805,
+      "step": 17215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8479208663447564e-05,
+      "loss": 0.1008,
+      "step": 17220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8478739862172424e-05,
+      "loss": 0.1341,
+      "step": 17225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8478271060897288e-05,
+      "loss": 0.1171,
+      "step": 17230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8477802259622148e-05,
+      "loss": 0.2465,
+      "step": 17235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8477333458347008e-05,
+      "loss": 0.3368,
+      "step": 17240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8476864657071867e-05,
+      "loss": 0.4202,
+      "step": 17245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.847639585579673e-05,
+      "loss": 0.4027,
+      "step": 17250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847592705452159e-05,
+      "loss": 0.2201,
+      "step": 17255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847545825324645e-05,
+      "loss": 0.0969,
+      "step": 17260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847498945197131e-05,
+      "loss": 0.0765,
+      "step": 17265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847452065069617e-05,
+      "loss": 0.198,
+      "step": 17270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8474051849421034e-05,
+      "loss": 0.153,
+      "step": 17275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8473583048145894e-05,
+      "loss": 0.1368,
+      "step": 17280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8473114246870754e-05,
+      "loss": 0.1888,
+      "step": 17285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8472645445595614e-05,
+      "loss": 0.1688,
+      "step": 17290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8472176644320474e-05,
+      "loss": 0.2878,
+      "step": 17295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8471707843045333e-05,
+      "loss": 0.5695,
+      "step": 17300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8471239041770193e-05,
+      "loss": 0.2027,
+      "step": 17305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8470770240495057e-05,
+      "loss": 0.0891,
+      "step": 17310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8470301439219917e-05,
+      "loss": 0.109,
+      "step": 17315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469832637944777e-05,
+      "loss": 0.1004,
+      "step": 17320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469363836669637e-05,
+      "loss": 0.0984,
+      "step": 17325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84688950353945e-05,
+      "loss": 0.19,
+      "step": 17330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846842623411936e-05,
+      "loss": 0.1577,
+      "step": 17335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846795743284422e-05,
+      "loss": 0.3391,
+      "step": 17340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846748863156908e-05,
+      "loss": 0.2577,
+      "step": 17345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846701983029394e-05,
+      "loss": 0.4126,
+      "step": 17350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.84665510290188e-05,
+      "loss": 0.1848,
+      "step": 17355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846608222774366e-05,
+      "loss": 0.1064,
+      "step": 17360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846561342646852e-05,
+      "loss": 0.1071,
+      "step": 17365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8465144625193383e-05,
+      "loss": 0.1296,
+      "step": 17370
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464675823918243e-05,
+      "loss": 0.1236,
+      "step": 17375
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464207022643103e-05,
+      "loss": 0.3178,
+      "step": 17380
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8463738221367963e-05,
+      "loss": 0.3502,
+      "step": 17385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8463269420092826e-05,
+      "loss": 0.3099,
+      "step": 17390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8462800618817686e-05,
+      "loss": 0.2417,
+      "step": 17395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8462331817542546e-05,
+      "loss": 0.4998,
+      "step": 17400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8461863016267406e-05,
+      "loss": 0.2151,
+      "step": 17405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8461394214992266e-05,
+      "loss": 0.0575,
+      "step": 17410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846092541371713e-05,
+      "loss": 0.122,
+      "step": 17415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846045661244199e-05,
+      "loss": 0.1077,
+      "step": 17420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845998781116685e-05,
+      "loss": 0.1329,
+      "step": 17425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845951900989171e-05,
+      "loss": 0.1299,
+      "step": 17430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845905020861657e-05,
+      "loss": 0.2726,
+      "step": 17435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845858140734143e-05,
+      "loss": 0.1639,
+      "step": 17440
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845811260606629e-05,
+      "loss": 0.4012,
+      "step": 17445
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845764380479115e-05,
+      "loss": 0.3916,
+      "step": 17450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8457175003516012e-05,
+      "loss": 0.2054,
+      "step": 17455
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8456706202240872e-05,
+      "loss": 0.0669,
+      "step": 17460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845623740096573e-05,
+      "loss": 0.1326,
+      "step": 17465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8455768599690595e-05,
+      "loss": 0.1697,
+      "step": 17470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8455299798415455e-05,
+      "loss": 0.1535,
+      "step": 17475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8454830997140315e-05,
+      "loss": 0.1638,
+      "step": 17480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8454362195865175e-05,
+      "loss": 0.1457,
+      "step": 17485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8453893394590035e-05,
+      "loss": 0.2595,
+      "step": 17490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8453424593314895e-05,
+      "loss": 0.2966,
+      "step": 17495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8452955792039755e-05,
+      "loss": 0.6608,
+      "step": 17500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8452486990764614e-05,
+      "loss": 0.1939,
+      "step": 17505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8452018189489478e-05,
+      "loss": 0.1323,
+      "step": 17510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8451549388214338e-05,
+      "loss": 0.1026,
+      "step": 17515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8451080586939198e-05,
+      "loss": 0.1229,
+      "step": 17520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8450611785664058e-05,
+      "loss": 0.0851,
+      "step": 17525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8450142984388918e-05,
+      "loss": 0.1832,
+      "step": 17530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844967418311378e-05,
+      "loss": 0.1868,
+      "step": 17535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844920538183864e-05,
+      "loss": 0.2952,
+      "step": 17540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.84487365805635e-05,
+      "loss": 0.2594,
+      "step": 17545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844826777928836e-05,
+      "loss": 0.3857,
+      "step": 17550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8447798978013224e-05,
+      "loss": 0.2327,
+      "step": 17555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8447330176738084e-05,
+      "loss": 0.0328,
+      "step": 17560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8446861375462944e-05,
+      "loss": 0.0574,
+      "step": 17565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8446392574187804e-05,
+      "loss": 0.1277,
+      "step": 17570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8445923772912664e-05,
+      "loss": 0.2752,
+      "step": 17575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8445454971637524e-05,
+      "loss": 0.1907,
+      "step": 17580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8444986170362384e-05,
+      "loss": 0.2263,
+      "step": 17585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8444517369087244e-05,
+      "loss": 0.2014,
+      "step": 17590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8444048567812103e-05,
+      "loss": 0.3681,
+      "step": 17595
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8443579766536967e-05,
+      "loss": 0.53,
+      "step": 17600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8443110965261827e-05,
+      "loss": 0.1759,
+      "step": 17605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8442642163986687e-05,
+      "loss": 0.106,
+      "step": 17610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844217336271155e-05,
+      "loss": 0.0909,
+      "step": 17615
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844170456143641e-05,
+      "loss": 0.0902,
+      "step": 17620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844123576016127e-05,
+      "loss": 0.1365,
+      "step": 17625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844076695888613e-05,
+      "loss": 0.1976,
+      "step": 17630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844029815761099e-05,
+      "loss": 0.2058,
+      "step": 17635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.843982935633585e-05,
+      "loss": 0.1611,
+      "step": 17640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.843936055506071e-05,
+      "loss": 0.2566,
+      "step": 17645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8438891753785573e-05,
+      "loss": 0.5291,
+      "step": 17650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8438422952510433e-05,
+      "loss": 0.1834,
+      "step": 17655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8437954151235293e-05,
+      "loss": 0.1088,
+      "step": 17660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8437485349960153e-05,
+      "loss": 0.0704,
+      "step": 17665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8437016548685013e-05,
+      "loss": 0.1562,
+      "step": 17670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8436547747409873e-05,
+      "loss": 0.1251,
+      "step": 17675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8436078946134736e-05,
+      "loss": 0.11,
+      "step": 17680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8435610144859596e-05,
+      "loss": 0.1469,
+      "step": 17685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8435141343584456e-05,
+      "loss": 0.5929,
+      "step": 17690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843467254230932e-05,
+      "loss": 0.312,
+      "step": 17695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843420374103418e-05,
+      "loss": 0.627,
+      "step": 17700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843373493975904e-05,
+      "loss": 0.2812,
+      "step": 17705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.84332661384839e-05,
+      "loss": 0.05,
+      "step": 17710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843279733720876e-05,
+      "loss": 0.0554,
+      "step": 17715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843232853593362e-05,
+      "loss": 0.1176,
+      "step": 17720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843185973465848e-05,
+      "loss": 0.1333,
+      "step": 17725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.843139093338334e-05,
+      "loss": 0.1865,
+      "step": 17730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.84309221321082e-05,
+      "loss": 0.2019,
+      "step": 17735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8430453330833062e-05,
+      "loss": 0.1893,
+      "step": 17740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8429984529557922e-05,
+      "loss": 0.1795,
+      "step": 17745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8429515728282782e-05,
+      "loss": 0.4059,
+      "step": 17750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.842904692700764e-05,
+      "loss": 0.2472,
+      "step": 17755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8428578125732505e-05,
+      "loss": 0.1022,
+      "step": 17760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8428109324457365e-05,
+      "loss": 0.1209,
+      "step": 17765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8427640523182225e-05,
+      "loss": 0.13,
+      "step": 17770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8427171721907085e-05,
+      "loss": 0.1884,
+      "step": 17775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8426702920631945e-05,
+      "loss": 0.2003,
+      "step": 17780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8426234119356805e-05,
+      "loss": 0.1951,
+      "step": 17785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8425765318081668e-05,
+      "loss": 0.1778,
+      "step": 17790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8425296516806528e-05,
+      "loss": 0.3655,
+      "step": 17795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8424827715531388e-05,
+      "loss": 0.4681,
+      "step": 17800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8424358914256248e-05,
+      "loss": 0.1807,
+      "step": 17805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8423890112981108e-05,
+      "loss": 0.041,
+      "step": 17810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8423421311705968e-05,
+      "loss": 0.0738,
+      "step": 17815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.842295251043083e-05,
+      "loss": 0.1138,
+      "step": 17820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.842248370915569e-05,
+      "loss": 0.1336,
+      "step": 17825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.842201490788055e-05,
+      "loss": 0.2036,
+      "step": 17830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.842154610660541e-05,
+      "loss": 0.166,
+      "step": 17835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8421077305330274e-05,
+      "loss": 0.1972,
+      "step": 17840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8420608504055134e-05,
+      "loss": 0.2675,
+      "step": 17845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8420139702779994e-05,
+      "loss": 0.431,
+      "step": 17850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8419670901504854e-05,
+      "loss": 0.2293,
+      "step": 17855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8419202100229714e-05,
+      "loss": 0.0781,
+      "step": 17860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8418733298954574e-05,
+      "loss": 0.0761,
+      "step": 17865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8418264497679434e-05,
+      "loss": 0.1226,
+      "step": 17870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8417795696404294e-05,
+      "loss": 0.219,
+      "step": 17875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8417326895129157e-05,
+      "loss": 0.2012,
+      "step": 17880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8416858093854017e-05,
+      "loss": 0.159,
+      "step": 17885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8416389292578877e-05,
+      "loss": 0.2031,
+      "step": 17890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8415920491303737e-05,
+      "loss": 0.3211,
+      "step": 17895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.84154516900286e-05,
+      "loss": 0.5196,
+      "step": 17900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.841498288875346e-05,
+      "loss": 0.2361,
+      "step": 17905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.841451408747832e-05,
+      "loss": 0.2127,
+      "step": 17910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.841404528620318e-05,
+      "loss": 0.1421,
+      "step": 17915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.841357648492804e-05,
+      "loss": 0.1517,
+      "step": 17920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8413107683652903e-05,
+      "loss": 0.138,
+      "step": 17925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8412638882377763e-05,
+      "loss": 0.2045,
+      "step": 17930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8412170081102623e-05,
+      "loss": 0.2729,
+      "step": 17935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8411701279827483e-05,
+      "loss": 0.2523,
+      "step": 17940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8411232478552343e-05,
+      "loss": 0.3385,
+      "step": 17945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8410763677277203e-05,
+      "loss": 0.4665,
+      "step": 17950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8410294876002063e-05,
+      "loss": 0.2603,
+      "step": 17955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8409826074726923e-05,
+      "loss": 0.0925,
+      "step": 17960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8409357273451786e-05,
+      "loss": 0.1406,
+      "step": 17965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8408888472176646e-05,
+      "loss": 0.1379,
+      "step": 17970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8408419670901506e-05,
+      "loss": 0.0881,
+      "step": 17975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840795086962637e-05,
+      "loss": 0.158,
+      "step": 17980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840748206835123e-05,
+      "loss": 0.2092,
+      "step": 17985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840701326707609e-05,
+      "loss": 0.349,
+      "step": 17990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840654446580095e-05,
+      "loss": 0.4693,
+      "step": 17995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840607566452581e-05,
+      "loss": 0.4295,
+      "step": 18000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840560686325067e-05,
+      "loss": 0.225,
+      "step": 18005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840513806197553e-05,
+      "loss": 0.1049,
+      "step": 18010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.840466926070039e-05,
+      "loss": 0.1065,
+      "step": 18015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8404200459425252e-05,
+      "loss": 0.1185,
+      "step": 18020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8403731658150112e-05,
+      "loss": 0.2128,
+      "step": 18025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8403262856874972e-05,
+      "loss": 0.1386,
+      "step": 18030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8402794055599832e-05,
+      "loss": 0.2085,
+      "step": 18035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8402325254324692e-05,
+      "loss": 0.2312,
+      "step": 18040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8401856453049555e-05,
+      "loss": 0.3012,
+      "step": 18045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8401387651774415e-05,
+      "loss": 0.5115,
+      "step": 18050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8400918850499275e-05,
+      "loss": 0.2569,
+      "step": 18055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8400450049224135e-05,
+      "loss": 0.1007,
+      "step": 18060
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8399981247948998e-05,
+      "loss": 0.1214,
+      "step": 18065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8399512446673858e-05,
+      "loss": 0.1432,
+      "step": 18070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8399043645398718e-05,
+      "loss": 0.1994,
+      "step": 18075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8398574844123578e-05,
+      "loss": 0.1827,
+      "step": 18080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8398106042848438e-05,
+      "loss": 0.2193,
+      "step": 18085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8397637241573298e-05,
+      "loss": 0.2757,
+      "step": 18090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8397168440298158e-05,
+      "loss": 0.2565,
+      "step": 18095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8396699639023018e-05,
+      "loss": 0.4674,
+      "step": 18100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8396230837747878e-05,
+      "loss": 0.198,
+      "step": 18105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.839576203647274e-05,
+      "loss": 0.0999,
+      "step": 18110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.83952932351976e-05,
+      "loss": 0.0993,
+      "step": 18115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.839482443392246e-05,
+      "loss": 0.1571,
+      "step": 18120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8394355632647324e-05,
+      "loss": 0.0981,
+      "step": 18125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8393886831372184e-05,
+      "loss": 0.1901,
+      "step": 18130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8393418030097044e-05,
+      "loss": 0.248,
+      "step": 18135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8392949228821904e-05,
+      "loss": 0.2168,
+      "step": 18140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8392480427546764e-05,
+      "loss": 0.2339,
+      "step": 18145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8392011626271624e-05,
+      "loss": 0.5106,
+      "step": 18150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8391542824996484e-05,
+      "loss": 0.2561,
+      "step": 18155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8391074023721347e-05,
+      "loss": 0.1187,
+      "step": 18160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8390605222446207e-05,
+      "loss": 0.0982,
+      "step": 18165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8390136421171067e-05,
+      "loss": 0.0904,
+      "step": 18170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8389667619895927e-05,
+      "loss": 0.1419,
+      "step": 18175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8389198818620787e-05,
+      "loss": 0.1005,
+      "step": 18180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8388730017345647e-05,
+      "loss": 0.1154,
+      "step": 18185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.838826121607051e-05,
+      "loss": 0.2446,
+      "step": 18190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.838779241479537e-05,
+      "loss": 0.3814,
+      "step": 18195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.838732361352023e-05,
+      "loss": 0.4632,
+      "step": 18200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8386854812245093e-05,
+      "loss": 0.1842,
+      "step": 18205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8386386010969953e-05,
+      "loss": 0.1311,
+      "step": 18210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8385917209694813e-05,
+      "loss": 0.1317,
+      "step": 18215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8385448408419673e-05,
+      "loss": 0.1859,
+      "step": 18220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8384979607144533e-05,
+      "loss": 0.0955,
+      "step": 18225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8384510805869393e-05,
+      "loss": 0.1593,
+      "step": 18230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8384042004594253e-05,
+      "loss": 0.2589,
+      "step": 18235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8383573203319113e-05,
+      "loss": 0.3052,
+      "step": 18240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8383104402043973e-05,
+      "loss": 0.2756,
+      "step": 18245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8382635600768836e-05,
+      "loss": 0.4911,
+      "step": 18250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8382166799493696e-05,
+      "loss": 0.1806,
+      "step": 18255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8381697998218556e-05,
+      "loss": 0.0599,
+      "step": 18260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8381229196943416e-05,
+      "loss": 0.0742,
+      "step": 18265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.838076039566828e-05,
+      "loss": 0.0986,
+      "step": 18270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.838029159439314e-05,
+      "loss": 0.201,
+      "step": 18275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8379822793118e-05,
+      "loss": 0.2901,
+      "step": 18280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.837935399184286e-05,
+      "loss": 0.1967,
+      "step": 18285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.837888519056772e-05,
+      "loss": 0.2117,
+      "step": 18290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.837841638929258e-05,
+      "loss": 0.2381,
+      "step": 18295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8377947588017442e-05,
+      "loss": 0.473,
+      "step": 18300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8377478786742302e-05,
+      "loss": 0.2497,
+      "step": 18305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8377009985467162e-05,
+      "loss": 0.0581,
+      "step": 18310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8376541184192022e-05,
+      "loss": 0.0962,
+      "step": 18315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8376072382916882e-05,
+      "loss": 0.089,
+      "step": 18320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8375603581641742e-05,
+      "loss": 0.2265,
+      "step": 18325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8375134780366605e-05,
+      "loss": 0.1076,
+      "step": 18330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8374665979091465e-05,
+      "loss": 0.2302,
+      "step": 18335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8374197177816325e-05,
+      "loss": 0.2914,
+      "step": 18340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.837372837654119e-05,
+      "loss": 0.2649,
+      "step": 18345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8373259575266048e-05,
+      "loss": 0.7229,
+      "step": 18350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8372790773990908e-05,
+      "loss": 0.2282,
+      "step": 18355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8372321972715768e-05,
+      "loss": 0.0604,
+      "step": 18360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8371853171440628e-05,
+      "loss": 0.1012,
+      "step": 18365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8371384370165488e-05,
+      "loss": 0.1669,
+      "step": 18370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8370915568890348e-05,
+      "loss": 0.1804,
+      "step": 18375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8370446767615208e-05,
+      "loss": 0.202,
+      "step": 18380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8369977966340068e-05,
+      "loss": 0.1953,
+      "step": 18385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836950916506493e-05,
+      "loss": 0.2295,
+      "step": 18390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836904036378979e-05,
+      "loss": 0.2345,
+      "step": 18395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836857156251465e-05,
+      "loss": 0.4057,
+      "step": 18400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836810276123951e-05,
+      "loss": 0.206,
+      "step": 18405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8367633959964374e-05,
+      "loss": 0.0997,
+      "step": 18410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8367165158689234e-05,
+      "loss": 0.0887,
+      "step": 18415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8366696357414094e-05,
+      "loss": 0.1298,
+      "step": 18420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8366227556138954e-05,
+      "loss": 0.198,
+      "step": 18425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8365758754863814e-05,
+      "loss": 0.2134,
+      "step": 18430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8365289953588674e-05,
+      "loss": 0.1678,
+      "step": 18435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8364821152313537e-05,
+      "loss": 0.2678,
+      "step": 18440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8364352351038397e-05,
+      "loss": 0.2635,
+      "step": 18445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8363883549763257e-05,
+      "loss": 0.6666,
+      "step": 18450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8363414748488117e-05,
+      "loss": 0.1784,
+      "step": 18455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8362945947212977e-05,
+      "loss": 0.0755,
+      "step": 18460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8362477145937837e-05,
+      "loss": 0.1487,
+      "step": 18465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8362008344662697e-05,
+      "loss": 0.0882,
+      "step": 18470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836153954338756e-05,
+      "loss": 0.1553,
+      "step": 18475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836107074211242e-05,
+      "loss": 0.1412,
+      "step": 18480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.836060194083728e-05,
+      "loss": 0.2449,
+      "step": 18485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8360133139562143e-05,
+      "loss": 0.217,
+      "step": 18490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8359664338287003e-05,
+      "loss": 0.3949,
+      "step": 18495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8359195537011863e-05,
+      "loss": 0.4204,
+      "step": 18500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8358726735736723e-05,
+      "loss": 0.1945,
+      "step": 18505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8358257934461583e-05,
+      "loss": 0.0726,
+      "step": 18510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8357789133186443e-05,
+      "loss": 0.1058,
+      "step": 18515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8357320331911303e-05,
+      "loss": 0.0942,
+      "step": 18520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8356851530636163e-05,
+      "loss": 0.1328,
+      "step": 18525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8356382729361026e-05,
+      "loss": 0.1709,
+      "step": 18530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8355913928085886e-05,
+      "loss": 0.1998,
+      "step": 18535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8355445126810746e-05,
+      "loss": 0.3097,
+      "step": 18540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8354976325535606e-05,
+      "loss": 0.3424,
+      "step": 18545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8354507524260466e-05,
+      "loss": 0.4817,
+      "step": 18550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.835403872298533e-05,
+      "loss": 0.1701,
+      "step": 18555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.835356992171019e-05,
+      "loss": 0.0983,
+      "step": 18560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.835310112043505e-05,
+      "loss": 0.1378,
+      "step": 18565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.835263231915991e-05,
+      "loss": 0.1254,
+      "step": 18570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8352163517884772e-05,
+      "loss": 0.1184,
+      "step": 18575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8351694716609632e-05,
+      "loss": 0.1757,
+      "step": 18580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8351225915334492e-05,
+      "loss": 0.1283,
+      "step": 18585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8350757114059352e-05,
+      "loss": 0.259,
+      "step": 18590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8350288312784212e-05,
+      "loss": 0.2993,
+      "step": 18595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8349819511509072e-05,
+      "loss": 0.695,
+      "step": 18600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8349350710233932e-05,
+      "loss": 0.2788,
+      "step": 18605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8348881908958792e-05,
+      "loss": 0.1188,
+      "step": 18610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8348413107683652e-05,
+      "loss": 0.0826,
+      "step": 18615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8347944306408515e-05,
+      "loss": 0.1032,
+      "step": 18620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8347475505133375e-05,
+      "loss": 0.0864,
+      "step": 18625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8347006703858235e-05,
+      "loss": 0.183,
+      "step": 18630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.83465379025831e-05,
+      "loss": 0.1417,
+      "step": 18635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.834606910130796e-05,
+      "loss": 0.179,
+      "step": 18640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8345600300032818e-05,
+      "loss": 0.3551,
+      "step": 18645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8345131498757678e-05,
+      "loss": 0.6553,
+      "step": 18650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8344662697482538e-05,
+      "loss": 0.1851,
+      "step": 18655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8344193896207398e-05,
+      "loss": 0.0574,
+      "step": 18660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8343725094932258e-05,
+      "loss": 0.0999,
+      "step": 18665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.834325629365712e-05,
+      "loss": 0.1026,
+      "step": 18670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.834278749238198e-05,
+      "loss": 0.1747,
+      "step": 18675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.834231869110684e-05,
+      "loss": 0.1283,
+      "step": 18680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.83418498898317e-05,
+      "loss": 0.1925,
+      "step": 18685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.834138108855656e-05,
+      "loss": 0.2752,
+      "step": 18690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8340912287281424e-05,
+      "loss": 0.2445,
+      "step": 18695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8340443486006284e-05,
+      "loss": 0.4561,
+      "step": 18700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8339974684731144e-05,
+      "loss": 0.2342,
+      "step": 18705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8339505883456004e-05,
+      "loss": 0.051,
+      "step": 18710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8339037082180868e-05,
+      "loss": 0.1113,
+      "step": 18715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8338568280905727e-05,
+      "loss": 0.1012,
+      "step": 18720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8338099479630587e-05,
+      "loss": 0.1449,
+      "step": 18725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8337630678355447e-05,
+      "loss": 0.1451,
+      "step": 18730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8337161877080307e-05,
+      "loss": 0.1471,
+      "step": 18735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8336693075805167e-05,
+      "loss": 0.17,
+      "step": 18740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8336224274530027e-05,
+      "loss": 0.1593,
+      "step": 18745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8335755473254887e-05,
+      "loss": 0.4541,
+      "step": 18750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8335286671979747e-05,
+      "loss": 0.2293,
+      "step": 18755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.833481787070461e-05,
+      "loss": 0.0878,
+      "step": 18760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.833434906942947e-05,
+      "loss": 0.1217,
+      "step": 18765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.833388026815433e-05,
+      "loss": 0.1412,
+      "step": 18770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8333411466879193e-05,
+      "loss": 0.1524,
+      "step": 18775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8332942665604053e-05,
+      "loss": 0.1379,
+      "step": 18780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8332473864328913e-05,
+      "loss": 0.3588,
+      "step": 18785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8332005063053773e-05,
+      "loss": 0.2402,
+      "step": 18790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8331536261778633e-05,
+      "loss": 0.3659,
+      "step": 18795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8331067460503493e-05,
+      "loss": 0.4519,
+      "step": 18800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8330598659228353e-05,
+      "loss": 0.1818,
+      "step": 18805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8330129857953216e-05,
+      "loss": 0.0785,
+      "step": 18810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8329661056678076e-05,
+      "loss": 0.1177,
+      "step": 18815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8329192255402936e-05,
+      "loss": 0.0967,
+      "step": 18820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8328723454127796e-05,
+      "loss": 0.0934,
+      "step": 18825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8328254652852656e-05,
+      "loss": 0.1312,
+      "step": 18830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8327785851577516e-05,
+      "loss": 0.3729,
+      "step": 18835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.832731705030238e-05,
+      "loss": 0.1916,
+      "step": 18840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.832684824902724e-05,
+      "loss": 0.3668,
+      "step": 18845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.83263794477521e-05,
+      "loss": 0.5375,
+      "step": 18850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8325910646476963e-05,
+      "loss": 0.2088,
+      "step": 18855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8325441845201823e-05,
+      "loss": 0.0323,
+      "step": 18860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8324973043926682e-05,
+      "loss": 0.1017,
+      "step": 18865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8324504242651542e-05,
+      "loss": 0.0991,
+      "step": 18870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8324035441376402e-05,
+      "loss": 0.1889,
+      "step": 18875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8323566640101262e-05,
+      "loss": 0.0697,
+      "step": 18880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8323097838826122e-05,
+      "loss": 0.2361,
+      "step": 18885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8322629037550982e-05,
+      "loss": 0.2399,
+      "step": 18890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8322160236275842e-05,
+      "loss": 0.3615,
+      "step": 18895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8321691435000705e-05,
+      "loss": 0.45,
+      "step": 18900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8321222633725565e-05,
+      "loss": 0.2541,
+      "step": 18905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8320753832450425e-05,
+      "loss": 0.0357,
+      "step": 18910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8320285031175285e-05,
+      "loss": 0.069,
+      "step": 18915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831981622990015e-05,
+      "loss": 0.1533,
+      "step": 18920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831934742862501e-05,
+      "loss": 0.0688,
+      "step": 18925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831887862734987e-05,
+      "loss": 0.1047,
+      "step": 18930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8318409826074728e-05,
+      "loss": 0.153,
+      "step": 18935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8317941024799588e-05,
+      "loss": 0.2231,
+      "step": 18940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8317472223524448e-05,
+      "loss": 0.4165,
+      "step": 18945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831700342224931e-05,
+      "loss": 0.521,
+      "step": 18950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831653462097417e-05,
+      "loss": 0.2484,
+      "step": 18955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831606581969903e-05,
+      "loss": 0.1186,
+      "step": 18960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.831559701842389e-05,
+      "loss": 0.1492,
+      "step": 18965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831512821714875e-05,
+      "loss": 0.1672,
+      "step": 18970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831465941587361e-05,
+      "loss": 0.2043,
+      "step": 18975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.831419061459847e-05,
+      "loss": 0.1615,
+      "step": 18980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8313721813323334e-05,
+      "loss": 0.1741,
+      "step": 18985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8313253012048194e-05,
+      "loss": 0.1771,
+      "step": 18990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8312784210773054e-05,
+      "loss": 0.2435,
+      "step": 18995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8312315409497918e-05,
+      "loss": 0.4375,
+      "step": 19000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8311846608222778e-05,
+      "loss": 0.1995,
+      "step": 19005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8311377806947637e-05,
+      "loss": 0.0675,
+      "step": 19010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8310909005672497e-05,
+      "loss": 0.1508,
+      "step": 19015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8310440204397357e-05,
+      "loss": 0.1821,
+      "step": 19020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8309971403122217e-05,
+      "loss": 0.2436,
+      "step": 19025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8309502601847077e-05,
+      "loss": 0.1228,
+      "step": 19030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8309033800571937e-05,
+      "loss": 0.2127,
+      "step": 19035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.83085649992968e-05,
+      "loss": 0.1937,
+      "step": 19040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830809619802166e-05,
+      "loss": 0.254,
+      "step": 19045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830762739674652e-05,
+      "loss": 0.5409,
+      "step": 19050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830715859547138e-05,
+      "loss": 0.252,
+      "step": 19055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830668979419624e-05,
+      "loss": 0.0875,
+      "step": 19060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8306220992921104e-05,
+      "loss": 0.1311,
+      "step": 19065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8305752191645963e-05,
+      "loss": 0.1158,
+      "step": 19070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8305283390370823e-05,
+      "loss": 0.1872,
+      "step": 19075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8304814589095683e-05,
+      "loss": 0.2072,
+      "step": 19080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8304345787820543e-05,
+      "loss": 0.2102,
+      "step": 19085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8303876986545407e-05,
+      "loss": 0.2013,
+      "step": 19090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8303408185270267e-05,
+      "loss": 0.2112,
+      "step": 19095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8302939383995126e-05,
+      "loss": 0.6385,
+      "step": 19100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8302470582719986e-05,
+      "loss": 0.1278,
+      "step": 19105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8302001781444846e-05,
+      "loss": 0.0642,
+      "step": 19110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8301532980169706e-05,
+      "loss": 0.0807,
+      "step": 19115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8301064178894566e-05,
+      "loss": 0.1014,
+      "step": 19120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830059537761943e-05,
+      "loss": 0.15,
+      "step": 19125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.830012657634429e-05,
+      "loss": 0.1112,
+      "step": 19130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.829965777506915e-05,
+      "loss": 0.2095,
+      "step": 19135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.829918897379401e-05,
+      "loss": 0.1831,
+      "step": 19140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8298720172518873e-05,
+      "loss": 0.3499,
+      "step": 19145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8298251371243733e-05,
+      "loss": 0.5991,
+      "step": 19150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8297782569968592e-05,
+      "loss": 0.1876,
+      "step": 19155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8297313768693452e-05,
+      "loss": 0.133,
+      "step": 19160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8296844967418312e-05,
+      "loss": 0.1359,
+      "step": 19165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8296376166143172e-05,
+      "loss": 0.1115,
+      "step": 19170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8295907364868032e-05,
+      "loss": 0.1205,
+      "step": 19175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8295438563592896e-05,
+      "loss": 0.1723,
+      "step": 19180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8294969762317755e-05,
+      "loss": 0.2365,
+      "step": 19185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8294500961042615e-05,
+      "loss": 0.3599,
+      "step": 19190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8294032159767475e-05,
+      "loss": 0.3155,
+      "step": 19195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8293563358492335e-05,
+      "loss": 0.5404,
+      "step": 19200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.82930945572172e-05,
+      "loss": 0.24,
+      "step": 19205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.829262575594206e-05,
+      "loss": 0.1003,
+      "step": 19210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.829215695466692e-05,
+      "loss": 0.1443,
+      "step": 19215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.829168815339178e-05,
+      "loss": 0.1095,
+      "step": 19220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8291219352116642e-05,
+      "loss": 0.133,
+      "step": 19225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.82907505508415e-05,
+      "loss": 0.1638,
+      "step": 19230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.829028174956636e-05,
+      "loss": 0.2031,
+      "step": 19235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828981294829122e-05,
+      "loss": 0.3121,
+      "step": 19240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828934414701608e-05,
+      "loss": 0.2641,
+      "step": 19245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828887534574094e-05,
+      "loss": 0.3234,
+      "step": 19250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.82884065444658e-05,
+      "loss": 0.2048,
+      "step": 19255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828793774319066e-05,
+      "loss": 0.0689,
+      "step": 19260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828746894191552e-05,
+      "loss": 0.1006,
+      "step": 19265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8287000140640385e-05,
+      "loss": 0.0911,
+      "step": 19270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8286531339365244e-05,
+      "loss": 0.2028,
+      "step": 19275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8286062538090104e-05,
+      "loss": 0.1203,
+      "step": 19280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8285593736814968e-05,
+      "loss": 0.1986,
+      "step": 19285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8285124935539828e-05,
+      "loss": 0.2303,
+      "step": 19290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8284656134264688e-05,
+      "loss": 0.2265,
+      "step": 19295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8284187332989548e-05,
+      "loss": 0.6002,
+      "step": 19300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8283718531714407e-05,
+      "loss": 0.1912,
+      "step": 19305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8283249730439267e-05,
+      "loss": 0.0834,
+      "step": 19310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8282780929164127e-05,
+      "loss": 0.1226,
+      "step": 19315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828231212788899e-05,
+      "loss": 0.201,
+      "step": 19320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828184332661385e-05,
+      "loss": 0.0701,
+      "step": 19325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828137452533871e-05,
+      "loss": 0.2523,
+      "step": 19330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828090572406357e-05,
+      "loss": 0.1423,
+      "step": 19335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.828043692278843e-05,
+      "loss": 0.2959,
+      "step": 19340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.827996812151329e-05,
+      "loss": 0.3732,
+      "step": 19345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8279499320238154e-05,
+      "loss": 0.4631,
+      "step": 19350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8279030518963014e-05,
+      "loss": 0.1664,
+      "step": 19355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8278561717687873e-05,
+      "loss": 0.1044,
+      "step": 19360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8278092916412737e-05,
+      "loss": 0.1303,
+      "step": 19365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8277624115137597e-05,
+      "loss": 0.1096,
+      "step": 19370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8277155313862457e-05,
+      "loss": 0.1389,
+      "step": 19375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8276686512587317e-05,
+      "loss": 0.1583,
+      "step": 19380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8276217711312177e-05,
+      "loss": 0.2356,
+      "step": 19385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8275748910037036e-05,
+      "loss": 0.2076,
+      "step": 19390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8275280108761896e-05,
+      "loss": 0.2269,
+      "step": 19395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8274811307486756e-05,
+      "loss": 0.3307,
+      "step": 19400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8274342506211616e-05,
+      "loss": 0.2044,
+      "step": 19405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8273873704936476e-05,
+      "loss": 0.0698,
+      "step": 19410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.827340490366134e-05,
+      "loss": 0.1467,
+      "step": 19415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.82729361023862e-05,
+      "loss": 0.1113,
+      "step": 19420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.827246730111106e-05,
+      "loss": 0.1657,
+      "step": 19425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8271998499835923e-05,
+      "loss": 0.1277,
+      "step": 19430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8271529698560783e-05,
+      "loss": 0.1894,
+      "step": 19435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8271060897285643e-05,
+      "loss": 0.1985,
+      "step": 19440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8270592096010503e-05,
+      "loss": 0.3031,
+      "step": 19445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8270123294735362e-05,
+      "loss": 0.5417,
+      "step": 19450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8269654493460222e-05,
+      "loss": 0.1623,
+      "step": 19455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8269185692185086e-05,
+      "loss": 0.0407,
+      "step": 19460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8268716890909946e-05,
+      "loss": 0.0865,
+      "step": 19465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8268248089634806e-05,
+      "loss": 0.1004,
+      "step": 19470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8267779288359666e-05,
+      "loss": 0.1137,
+      "step": 19475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8267310487084525e-05,
+      "loss": 0.1354,
+      "step": 19480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8266841685809385e-05,
+      "loss": 0.2164,
+      "step": 19485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8266372884534245e-05,
+      "loss": 0.2859,
+      "step": 19490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826590408325911e-05,
+      "loss": 0.3575,
+      "step": 19495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826543528198397e-05,
+      "loss": 0.5411,
+      "step": 19500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826496648070883e-05,
+      "loss": 0.2238,
+      "step": 19505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8264497679433692e-05,
+      "loss": 0.0554,
+      "step": 19510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8264028878158552e-05,
+      "loss": 0.1204,
+      "step": 19515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826356007688341e-05,
+      "loss": 0.1493,
+      "step": 19520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826309127560827e-05,
+      "loss": 0.1468,
+      "step": 19525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826262247433313e-05,
+      "loss": 0.2244,
+      "step": 19530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826215367305799e-05,
+      "loss": 0.1374,
+      "step": 19535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826168487178285e-05,
+      "loss": 0.1748,
+      "step": 19540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.826121607050771e-05,
+      "loss": 0.2261,
+      "step": 19545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8260747269232575e-05,
+      "loss": 0.4271,
+      "step": 19550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8260278467957435e-05,
+      "loss": 0.1653,
+      "step": 19555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8259809666682295e-05,
+      "loss": 0.0668,
+      "step": 19560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8259340865407154e-05,
+      "loss": 0.1128,
+      "step": 19565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8258872064132014e-05,
+      "loss": 0.3179,
+      "step": 19570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8258403262856878e-05,
+      "loss": 0.1556,
+      "step": 19575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8257934461581738e-05,
+      "loss": 0.2226,
+      "step": 19580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8257465660306598e-05,
+      "loss": 0.2039,
+      "step": 19585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8256996859031458e-05,
+      "loss": 0.3087,
+      "step": 19590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8256528057756317e-05,
+      "loss": 0.3327,
+      "step": 19595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.825605925648118e-05,
+      "loss": 0.4217,
+      "step": 19600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.825559045520604e-05,
+      "loss": 0.1834,
+      "step": 19605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.82551216539309e-05,
+      "loss": 0.0819,
+      "step": 19610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.825465285265576e-05,
+      "loss": 0.1225,
+      "step": 19615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.825418405138062e-05,
+      "loss": 0.0862,
+      "step": 19620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.825371525010548e-05,
+      "loss": 0.2495,
+      "step": 19625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.825324644883034e-05,
+      "loss": 0.1874,
+      "step": 19630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8252777647555204e-05,
+      "loss": 0.2202,
+      "step": 19635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8252308846280064e-05,
+      "loss": 0.2002,
+      "step": 19640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8251840045004924e-05,
+      "loss": 0.2528,
+      "step": 19645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8251371243729784e-05,
+      "loss": 0.4702,
+      "step": 19650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8250902442454647e-05,
+      "loss": 0.2313,
+      "step": 19655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8250433641179507e-05,
+      "loss": 0.0705,
+      "step": 19660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8249964839904367e-05,
+      "loss": 0.0752,
+      "step": 19665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8249496038629227e-05,
+      "loss": 0.1123,
+      "step": 19670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8249027237354087e-05,
+      "loss": 0.1416,
+      "step": 19675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8248558436078947e-05,
+      "loss": 0.2429,
+      "step": 19680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8248089634803806e-05,
+      "loss": 0.2266,
+      "step": 19685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824762083352867e-05,
+      "loss": 0.2297,
+      "step": 19690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824715203225353e-05,
+      "loss": 0.3014,
+      "step": 19695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824668323097839e-05,
+      "loss": 0.4253,
+      "step": 19700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824621442970325e-05,
+      "loss": 0.2434,
+      "step": 19705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.824574562842811e-05,
+      "loss": 0.0985,
+      "step": 19710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8245276827152973e-05,
+      "loss": 0.1376,
+      "step": 19715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8244808025877833e-05,
+      "loss": 0.1281,
+      "step": 19720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8244339224602693e-05,
+      "loss": 0.155,
+      "step": 19725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8243870423327553e-05,
+      "loss": 0.1448,
+      "step": 19730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8243401622052413e-05,
+      "loss": 0.1762,
+      "step": 19735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8242932820777276e-05,
+      "loss": 0.3514,
+      "step": 19740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8242464019502136e-05,
+      "loss": 0.3415,
+      "step": 19745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8241995218226996e-05,
+      "loss": 0.4549,
+      "step": 19750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8241526416951856e-05,
+      "loss": 0.2531,
+      "step": 19755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8241057615676716e-05,
+      "loss": 0.0812,
+      "step": 19760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8240588814401576e-05,
+      "loss": 0.0823,
+      "step": 19765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8240120013126435e-05,
+      "loss": 0.0879,
+      "step": 19770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8239651211851295e-05,
+      "loss": 0.0641,
+      "step": 19775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.823918241057616e-05,
+      "loss": 0.1483,
+      "step": 19780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.823871360930102e-05,
+      "loss": 0.1822,
+      "step": 19785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.823824480802588e-05,
+      "loss": 0.2754,
+      "step": 19790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8237776006750742e-05,
+      "loss": 0.3038,
+      "step": 19795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8237307205475602e-05,
+      "loss": 0.3856,
+      "step": 19800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8236838404200462e-05,
+      "loss": 0.2713,
+      "step": 19805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8236369602925322e-05,
+      "loss": 0.0589,
+      "step": 19810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.823590080165018e-05,
+      "loss": 0.0968,
+      "step": 19815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.823543200037504e-05,
+      "loss": 0.0974,
+      "step": 19820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.82349631990999e-05,
+      "loss": 0.119,
+      "step": 19825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8234494397824765e-05,
+      "loss": 0.1067,
+      "step": 19830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8234025596549625e-05,
+      "loss": 0.1,
+      "step": 19835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8233556795274485e-05,
+      "loss": 0.1671,
+      "step": 19840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8233087993999345e-05,
+      "loss": 0.2918,
+      "step": 19845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8232619192724205e-05,
+      "loss": 0.6152,
+      "step": 19850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8232150391449065e-05,
+      "loss": 0.1955,
+      "step": 19855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8231681590173928e-05,
+      "loss": 0.0708,
+      "step": 19860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8231212788898788e-05,
+      "loss": 0.0908,
+      "step": 19865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8230743987623648e-05,
+      "loss": 0.1555,
+      "step": 19870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.823027518634851e-05,
+      "loss": 0.1415,
+      "step": 19875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822980638507337e-05,
+      "loss": 0.1305,
+      "step": 19880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822933758379823e-05,
+      "loss": 0.2142,
+      "step": 19885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822886878252309e-05,
+      "loss": 0.2684,
+      "step": 19890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822839998124795e-05,
+      "loss": 0.2105,
+      "step": 19895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822793117997281e-05,
+      "loss": 0.4126,
+      "step": 19900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822746237869767e-05,
+      "loss": 0.2058,
+      "step": 19905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822699357742253e-05,
+      "loss": 0.0673,
+      "step": 19910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822652477614739e-05,
+      "loss": 0.1449,
+      "step": 19915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822605597487225e-05,
+      "loss": 0.0786,
+      "step": 19920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8225587173597114e-05,
+      "loss": 0.1939,
+      "step": 19925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8225118372321974e-05,
+      "loss": 0.1068,
+      "step": 19930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8224649571046834e-05,
+      "loss": 0.1644,
+      "step": 19935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8224180769771697e-05,
+      "loss": 0.1876,
+      "step": 19940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8223711968496557e-05,
+      "loss": 0.2146,
+      "step": 19945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8223243167221417e-05,
+      "loss": 0.4408,
+      "step": 19950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8222774365946277e-05,
+      "loss": 0.1771,
+      "step": 19955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8222305564671137e-05,
+      "loss": 0.0614,
+      "step": 19960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8221836763395997e-05,
+      "loss": 0.0667,
+      "step": 19965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822136796212086e-05,
+      "loss": 0.1274,
+      "step": 19970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822089916084572e-05,
+      "loss": 0.128,
+      "step": 19975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.822043035957058e-05,
+      "loss": 0.1434,
+      "step": 19980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.821996155829544e-05,
+      "loss": 0.2268,
+      "step": 19985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.82194927570203e-05,
+      "loss": 0.2602,
+      "step": 19990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.821902395574516e-05,
+      "loss": 0.2319,
+      "step": 19995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.821855515447002e-05,
+      "loss": 0.624,
+      "step": 20000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8218086353194883e-05,
+      "loss": 0.1856,
+      "step": 20005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8217617551919743e-05,
+      "loss": 0.0824,
+      "step": 20010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8217148750644603e-05,
+      "loss": 0.0766,
+      "step": 20015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8216679949369466e-05,
+      "loss": 0.1607,
+      "step": 20020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8216211148094326e-05,
+      "loss": 0.1132,
+      "step": 20025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8215742346819186e-05,
+      "loss": 0.1429,
+      "step": 20030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8215273545544046e-05,
+      "loss": 0.1928,
+      "step": 20035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8214804744268906e-05,
+      "loss": 0.2072,
+      "step": 20040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8214335942993766e-05,
+      "loss": 0.3352,
+      "step": 20045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8213867141718626e-05,
+      "loss": 0.5031,
+      "step": 20050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8213398340443486e-05,
+      "loss": 0.2016,
+      "step": 20055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8212929539168346e-05,
+      "loss": 0.0557,
+      "step": 20060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.821246073789321e-05,
+      "loss": 0.0958,
+      "step": 20065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.821199193661807e-05,
+      "loss": 0.0978,
+      "step": 20070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.821152313534293e-05,
+      "loss": 0.0867,
+      "step": 20075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8211054334067792e-05,
+      "loss": 0.147,
+      "step": 20080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8210585532792652e-05,
+      "loss": 0.1557,
+      "step": 20085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8210116731517512e-05,
+      "loss": 0.2568,
+      "step": 20090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8209647930242372e-05,
+      "loss": 0.3529,
+      "step": 20095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8209179128967232e-05,
+      "loss": 0.491,
+      "step": 20100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820871032769209e-05,
+      "loss": 0.1841,
+      "step": 20105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8208241526416955e-05,
+      "loss": 0.033,
+      "step": 20110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8207772725141815e-05,
+      "loss": 0.0934,
+      "step": 20115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8207303923866675e-05,
+      "loss": 0.059,
+      "step": 20120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8206835122591535e-05,
+      "loss": 0.1281,
+      "step": 20125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8206366321316395e-05,
+      "loss": 0.1659,
+      "step": 20130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8205897520041255e-05,
+      "loss": 0.1677,
+      "step": 20135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8205428718766115e-05,
+      "loss": 0.27,
+      "step": 20140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8204959917490978e-05,
+      "loss": 0.2848,
+      "step": 20145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8204491116215838e-05,
+      "loss": 0.5694,
+      "step": 20150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8204022314940698e-05,
+      "loss": 0.2125,
+      "step": 20155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820355351366556e-05,
+      "loss": 0.0379,
+      "step": 20160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820308471239042e-05,
+      "loss": 0.1238,
+      "step": 20165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820261591111528e-05,
+      "loss": 0.1555,
+      "step": 20170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820214710984014e-05,
+      "loss": 0.141,
+      "step": 20175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8201678308565e-05,
+      "loss": 0.0992,
+      "step": 20180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820120950728986e-05,
+      "loss": 0.1861,
+      "step": 20185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820074070601472e-05,
+      "loss": 0.2341,
+      "step": 20190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.820027190473958e-05,
+      "loss": 0.2778,
+      "step": 20195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8199803103464444e-05,
+      "loss": 0.4585,
+      "step": 20200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8199334302189304e-05,
+      "loss": 0.193,
+      "step": 20205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8198865500914164e-05,
+      "loss": 0.0783,
+      "step": 20210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8198396699639024e-05,
+      "loss": 0.1657,
+      "step": 20215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8197927898363884e-05,
+      "loss": 0.0852,
+      "step": 20220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8197459097088747e-05,
+      "loss": 0.1167,
+      "step": 20225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8196990295813607e-05,
+      "loss": 0.1107,
+      "step": 20230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8196521494538467e-05,
+      "loss": 0.1642,
+      "step": 20235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8196052693263327e-05,
+      "loss": 0.2792,
+      "step": 20240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8195583891988187e-05,
+      "loss": 0.37,
+      "step": 20245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.819511509071305e-05,
+      "loss": 0.3121,
+      "step": 20250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819464628943791e-05,
+      "loss": 0.2295,
+      "step": 20255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819417748816277e-05,
+      "loss": 0.0714,
+      "step": 20260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819370868688763e-05,
+      "loss": 0.1149,
+      "step": 20265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819323988561249e-05,
+      "loss": 0.1149,
+      "step": 20270
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819277108433735e-05,
+      "loss": 0.1848,
+      "step": 20275
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819230228306221e-05,
+      "loss": 0.1652,
+      "step": 20280
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.819183348178707e-05,
+      "loss": 0.1869,
+      "step": 20285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8191364680511933e-05,
+      "loss": 0.2933,
+      "step": 20290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8190895879236793e-05,
+      "loss": 0.2485,
+      "step": 20295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8190427077961653e-05,
+      "loss": 0.5297,
+      "step": 20300
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8189958276686516e-05,
+      "loss": 0.2442,
+      "step": 20305
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8189489475411376e-05,
+      "loss": 0.0492,
+      "step": 20310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8189020674136236e-05,
+      "loss": 0.1049,
+      "step": 20315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8188551872861096e-05,
+      "loss": 0.1435,
+      "step": 20320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8188083071585956e-05,
+      "loss": 0.1713,
+      "step": 20325
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8187614270310816e-05,
+      "loss": 0.1232,
+      "step": 20330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8187145469035676e-05,
+      "loss": 0.1662,
+      "step": 20335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.818667666776054e-05,
+      "loss": 0.2596,
+      "step": 20340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.81862078664854e-05,
+      "loss": 0.285,
+      "step": 20345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.818573906521026e-05,
+      "loss": 0.3881,
+      "step": 20350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.818527026393512e-05,
+      "loss": 0.2469,
+      "step": 20355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.818480146265998e-05,
+      "loss": 0.0816,
+      "step": 20360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.818433266138484e-05,
+      "loss": 0.159,
+      "step": 20365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8183863860109702e-05,
+      "loss": 0.0941,
+      "step": 20370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8183395058834562e-05,
+      "loss": 0.0982,
+      "step": 20375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8182926257559422e-05,
+      "loss": 0.1584,
+      "step": 20380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8182457456284282e-05,
+      "loss": 0.1844,
+      "step": 20385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8181988655009145e-05,
+      "loss": 0.2987,
+      "step": 20390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8181519853734005e-05,
+      "loss": 0.2073,
+      "step": 20395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8181051052458865e-05,
+      "loss": 0.4283,
+      "step": 20400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8180582251183725e-05,
+      "loss": 0.1415,
+      "step": 20405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8180113449908585e-05,
+      "loss": 0.041,
+      "step": 20410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8179644648633445e-05,
+      "loss": 0.0686,
+      "step": 20415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8179175847358305e-05,
+      "loss": 0.0919,
+      "step": 20420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8178707046083165e-05,
+      "loss": 0.1709,
+      "step": 20425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8178238244808028e-05,
+      "loss": 0.1945,
+      "step": 20430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8177769443532888e-05,
+      "loss": 0.2593,
+      "step": 20435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8177300642257748e-05,
+      "loss": 0.2522,
+      "step": 20440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8176831840982608e-05,
+      "loss": 0.2546,
+      "step": 20445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.817636303970747e-05,
+      "loss": 0.4386,
+      "step": 20450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.817589423843233e-05,
+      "loss": 0.1648,
+      "step": 20455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.817542543715719e-05,
+      "loss": 0.1356,
+      "step": 20460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.817495663588205e-05,
+      "loss": 0.0529,
+      "step": 20465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.817448783460691e-05,
+      "loss": 0.1376,
+      "step": 20470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.817401903333177e-05,
+      "loss": 0.1926,
+      "step": 20475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8173550232056634e-05,
+      "loss": 0.1433,
+      "step": 20480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8173081430781494e-05,
+      "loss": 0.0786,
+      "step": 20485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8172612629506354e-05,
+      "loss": 0.1545,
+      "step": 20490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8172143828231214e-05,
+      "loss": 0.2676,
+      "step": 20495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8171675026956074e-05,
+      "loss": 0.4635,
+      "step": 20500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8171206225680934e-05,
+      "loss": 0.1845,
+      "step": 20505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8170737424405797e-05,
+      "loss": 0.0334,
+      "step": 20510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8170268623130657e-05,
+      "loss": 0.0685,
+      "step": 20515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8169799821855517e-05,
+      "loss": 0.139,
+      "step": 20520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8169331020580377e-05,
+      "loss": 0.1517,
+      "step": 20525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816886221930524e-05,
+      "loss": 0.1412,
+      "step": 20530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.81683934180301e-05,
+      "loss": 0.1334,
+      "step": 20535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816792461675496e-05,
+      "loss": 0.3197,
+      "step": 20540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816745581547982e-05,
+      "loss": 0.3216,
+      "step": 20545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816698701420468e-05,
+      "loss": 0.6545,
+      "step": 20550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816651821292954e-05,
+      "loss": 0.1667,
+      "step": 20555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.81660494116544e-05,
+      "loss": 0.0827,
+      "step": 20560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816558061037926e-05,
+      "loss": 0.0818,
+      "step": 20565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816511180910412e-05,
+      "loss": 0.1156,
+      "step": 20570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8164643007828983e-05,
+      "loss": 0.114,
+      "step": 20575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8164174206553843e-05,
+      "loss": 0.1087,
+      "step": 20580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8163705405278703e-05,
+      "loss": 0.1839,
+      "step": 20585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8163236604003566e-05,
+      "loss": 0.3323,
+      "step": 20590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8162767802728426e-05,
+      "loss": 0.3873,
+      "step": 20595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8162299001453286e-05,
+      "loss": 0.45,
+      "step": 20600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8161830200178146e-05,
+      "loss": 0.2414,
+      "step": 20605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8161361398903006e-05,
+      "loss": 0.0611,
+      "step": 20610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8160892597627866e-05,
+      "loss": 0.124,
+      "step": 20615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.816042379635273e-05,
+      "loss": 0.0909,
+      "step": 20620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815995499507759e-05,
+      "loss": 0.1789,
+      "step": 20625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815948619380245e-05,
+      "loss": 0.1668,
+      "step": 20630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815901739252731e-05,
+      "loss": 0.1578,
+      "step": 20635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815854859125217e-05,
+      "loss": 0.253,
+      "step": 20640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815807978997703e-05,
+      "loss": 0.2484,
+      "step": 20645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.815761098870189e-05,
+      "loss": 0.4484,
+      "step": 20650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8157142187426752e-05,
+      "loss": 0.1914,
+      "step": 20655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8156673386151612e-05,
+      "loss": 0.1153,
+      "step": 20660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8156204584876472e-05,
+      "loss": 0.1582,
+      "step": 20665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8155735783601335e-05,
+      "loss": 0.1714,
+      "step": 20670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8155266982326195e-05,
+      "loss": 0.1086,
+      "step": 20675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8154798181051055e-05,
+      "loss": 0.1785,
+      "step": 20680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8154329379775915e-05,
+      "loss": 0.1845,
+      "step": 20685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8153860578500775e-05,
+      "loss": 0.3776,
+      "step": 20690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8153391777225635e-05,
+      "loss": 0.1668,
+      "step": 20695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8152922975950495e-05,
+      "loss": 0.5262,
+      "step": 20700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8152454174675355e-05,
+      "loss": 0.1545,
+      "step": 20705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8151985373400215e-05,
+      "loss": 0.1112,
+      "step": 20710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8151516572125078e-05,
+      "loss": 0.0721,
+      "step": 20715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8151047770849938e-05,
+      "loss": 0.1326,
+      "step": 20720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8150578969574798e-05,
+      "loss": 0.1883,
+      "step": 20725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8150110168299658e-05,
+      "loss": 0.0903,
+      "step": 20730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814964136702452e-05,
+      "loss": 0.2431,
+      "step": 20735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814917256574938e-05,
+      "loss": 0.301,
+      "step": 20740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814870376447424e-05,
+      "loss": 0.3377,
+      "step": 20745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.81482349631991e-05,
+      "loss": 0.3813,
+      "step": 20750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814776616192396e-05,
+      "loss": 0.154,
+      "step": 20755
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8147297360648824e-05,
+      "loss": 0.0773,
+      "step": 20760
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8146828559373684e-05,
+      "loss": 0.0761,
+      "step": 20765
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8146359758098544e-05,
+      "loss": 0.075,
+      "step": 20770
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8145890956823404e-05,
+      "loss": 0.2386,
+      "step": 20775
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8145422155548264e-05,
+      "loss": 0.1128,
+      "step": 20780
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8144953354273124e-05,
+      "loss": 0.1266,
+      "step": 20785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8144484552997984e-05,
+      "loss": 0.2653,
+      "step": 20790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8144015751722844e-05,
+      "loss": 0.2307,
+      "step": 20795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8143546950447707e-05,
+      "loss": 0.5244,
+      "step": 20800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8143078149172567e-05,
+      "loss": 0.2061,
+      "step": 20805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8142609347897427e-05,
+      "loss": 0.0903,
+      "step": 20810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814214054662229e-05,
+      "loss": 0.0667,
+      "step": 20815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814167174534715e-05,
+      "loss": 0.069,
+      "step": 20820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814120294407201e-05,
+      "loss": 0.0872,
+      "step": 20825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814073414279687e-05,
+      "loss": 0.183,
+      "step": 20830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.814026534152173e-05,
+      "loss": 0.1693,
+      "step": 20835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.813979654024659e-05,
+      "loss": 0.2612,
+      "step": 20840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.813932773897145e-05,
+      "loss": 0.2822,
+      "step": 20845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8138858937696313e-05,
+      "loss": 0.5785,
+      "step": 20850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8138390136421173e-05,
+      "loss": 0.2277,
+      "step": 20855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8137921335146033e-05,
+      "loss": 0.0781,
+      "step": 20860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8137452533870893e-05,
+      "loss": 0.0593,
+      "step": 20865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8136983732595753e-05,
+      "loss": 0.0789,
+      "step": 20870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8136514931320613e-05,
+      "loss": 0.1482,
+      "step": 20875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8136046130045476e-05,
+      "loss": 0.1335,
+      "step": 20880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8135577328770336e-05,
+      "loss": 0.1823,
+      "step": 20885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8135108527495196e-05,
+      "loss": 0.2789,
+      "step": 20890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8134639726220056e-05,
+      "loss": 0.2018,
+      "step": 20895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813417092494492e-05,
+      "loss": 0.4649,
+      "step": 20900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813370212366978e-05,
+      "loss": 0.2208,
+      "step": 20905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813323332239464e-05,
+      "loss": 0.0604,
+      "step": 20910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.81327645211195e-05,
+      "loss": 0.0778,
+      "step": 20915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813229571984436e-05,
+      "loss": 0.1314,
+      "step": 20920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813182691856922e-05,
+      "loss": 0.1132,
+      "step": 20925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813135811729408e-05,
+      "loss": 0.1354,
+      "step": 20930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.813088931601894e-05,
+      "loss": 0.208,
+      "step": 20935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8130420514743802e-05,
+      "loss": 0.26,
+      "step": 20940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8129951713468662e-05,
+      "loss": 0.281,
+      "step": 20945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8129482912193522e-05,
+      "loss": 0.4454,
+      "step": 20950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8129014110918382e-05,
+      "loss": 0.2346,
+      "step": 20955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8128545309643245e-05,
+      "loss": 0.0742,
+      "step": 20960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8128076508368105e-05,
+      "loss": 0.104,
+      "step": 20965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8127607707092965e-05,
+      "loss": 0.0568,
+      "step": 20970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8127138905817825e-05,
+      "loss": 0.1337,
+      "step": 20975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8126670104542685e-05,
+      "loss": 0.1325,
+      "step": 20980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8126201303267545e-05,
+      "loss": 0.1653,
+      "step": 20985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812573250199241e-05,
+      "loss": 0.2467,
+      "step": 20990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8125263700717268e-05,
+      "loss": 0.3002,
+      "step": 20995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8124794899442128e-05,
+      "loss": 0.3036,
+      "step": 21000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8124326098166988e-05,
+      "loss": 0.2193,
+      "step": 21005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8123857296891848e-05,
+      "loss": 0.0949,
+      "step": 21010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8123388495616708e-05,
+      "loss": 0.0853,
+      "step": 21015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812291969434157e-05,
+      "loss": 0.0995,
+      "step": 21020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812245089306643e-05,
+      "loss": 0.138,
+      "step": 21025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812198209179129e-05,
+      "loss": 0.1339,
+      "step": 21030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.812151329051615e-05,
+      "loss": 0.2001,
+      "step": 21035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8121044489241014e-05,
+      "loss": 0.2148,
+      "step": 21040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8120575687965874e-05,
+      "loss": 0.2632,
+      "step": 21045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8120106886690734e-05,
+      "loss": 0.4485,
+      "step": 21050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8119638085415594e-05,
+      "loss": 0.1182,
+      "step": 21055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8119169284140454e-05,
+      "loss": 0.0904,
+      "step": 21060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8118700482865314e-05,
+      "loss": 0.1157,
+      "step": 21065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8118231681590174e-05,
+      "loss": 0.1715,
+      "step": 21070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8117762880315034e-05,
+      "loss": 0.154,
+      "step": 21075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8117294079039894e-05,
+      "loss": 0.2591,
+      "step": 21080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8116825277764757e-05,
+      "loss": 0.2791,
+      "step": 21085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8116356476489617e-05,
+      "loss": 0.2793,
+      "step": 21090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8115887675214477e-05,
+      "loss": 0.2693,
+      "step": 21095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.811541887393934e-05,
+      "loss": 0.3771,
+      "step": 21100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.81149500726642e-05,
+      "loss": 0.1693,
+      "step": 21105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.811448127138906e-05,
+      "loss": 0.1137,
+      "step": 21110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.811401247011392e-05,
+      "loss": 0.0693,
+      "step": 21115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.811354366883878e-05,
+      "loss": 0.0632,
+      "step": 21120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.811307486756364e-05,
+      "loss": 0.1544,
+      "step": 21125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8112606066288503e-05,
+      "loss": 0.2066,
+      "step": 21130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8112137265013363e-05,
+      "loss": 0.2304,
+      "step": 21135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8111668463738223e-05,
+      "loss": 0.3096,
+      "step": 21140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8111199662463083e-05,
+      "loss": 0.302,
+      "step": 21145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8110730861187943e-05,
+      "loss": 0.3194,
+      "step": 21150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8110262059912803e-05,
+      "loss": 0.1765,
+      "step": 21155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8109793258637663e-05,
+      "loss": 0.0763,
+      "step": 21160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8109324457362526e-05,
+      "loss": 0.0571,
+      "step": 21165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8108855656087386e-05,
+      "loss": 0.1146,
+      "step": 21170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8108386854812246e-05,
+      "loss": 0.1685,
+      "step": 21175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810791805353711e-05,
+      "loss": 0.1763,
+      "step": 21180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810744925226197e-05,
+      "loss": 0.146,
+      "step": 21185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810698045098683e-05,
+      "loss": 0.232,
+      "step": 21190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810651164971169e-05,
+      "loss": 0.3637,
+      "step": 21195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810604284843655e-05,
+      "loss": 0.4796,
+      "step": 21200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810557404716141e-05,
+      "loss": 0.2576,
+      "step": 21205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810510524588627e-05,
+      "loss": 0.0738,
+      "step": 21210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810463644461113e-05,
+      "loss": 0.0893,
+      "step": 21215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810416764333599e-05,
+      "loss": 0.1347,
+      "step": 21220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8103698842060852e-05,
+      "loss": 0.1224,
+      "step": 21225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8103230040785712e-05,
+      "loss": 0.1012,
+      "step": 21230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8102761239510572e-05,
+      "loss": 0.1355,
+      "step": 21235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8102292438235432e-05,
+      "loss": 0.2893,
+      "step": 21240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8101823636960295e-05,
+      "loss": 0.2347,
+      "step": 21245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8101354835685155e-05,
+      "loss": 0.4496,
+      "step": 21250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8100886034410015e-05,
+      "loss": 0.2292,
+      "step": 21255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8100417233134875e-05,
+      "loss": 0.1001,
+      "step": 21260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8099948431859735e-05,
+      "loss": 0.0865,
+      "step": 21265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.80994796305846e-05,
+      "loss": 0.0803,
+      "step": 21270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.809901082930946e-05,
+      "loss": 0.1175,
+      "step": 21275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.809854202803432e-05,
+      "loss": 0.2219,
+      "step": 21280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.809807322675918e-05,
+      "loss": 0.2215,
+      "step": 21285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8097604425484038e-05,
+      "loss": 0.221,
+      "step": 21290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8097135624208898e-05,
+      "loss": 0.2907,
+      "step": 21295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8096666822933758e-05,
+      "loss": 0.5423,
+      "step": 21300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8096198021658618e-05,
+      "loss": 0.1497,
+      "step": 21305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.809572922038348e-05,
+      "loss": 0.1218,
+      "step": 21310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.809526041910834e-05,
+      "loss": 0.1345,
+      "step": 21315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.80947916178332e-05,
+      "loss": 0.0717,
+      "step": 21320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8094322816558065e-05,
+      "loss": 0.2388,
+      "step": 21325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8093854015282925e-05,
+      "loss": 0.1201,
+      "step": 21330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8093385214007784e-05,
+      "loss": 0.1755,
+      "step": 21335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8092916412732644e-05,
+      "loss": 0.2016,
+      "step": 21340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8092447611457504e-05,
+      "loss": 0.2375,
+      "step": 21345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8091978810182364e-05,
+      "loss": 0.6163,
+      "step": 21350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8091510008907224e-05,
+      "loss": 0.1855,
+      "step": 21355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8091041207632084e-05,
+      "loss": 0.1006,
+      "step": 21360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8090572406356947e-05,
+      "loss": 0.0872,
+      "step": 21365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8090103605081807e-05,
+      "loss": 0.0815,
+      "step": 21370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8089634803806667e-05,
+      "loss": 0.1751,
+      "step": 21375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8089166002531527e-05,
+      "loss": 0.1182,
+      "step": 21380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8088697201256387e-05,
+      "loss": 0.2401,
+      "step": 21385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808822839998125e-05,
+      "loss": 0.2828,
+      "step": 21390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808775959870611e-05,
+      "loss": 0.2181,
+      "step": 21395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808729079743097e-05,
+      "loss": 0.4759,
+      "step": 21400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.808682199615583e-05,
+      "loss": 0.0917,
+      "step": 21405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8086353194880694e-05,
+      "loss": 0.0687,
+      "step": 21410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8085884393605554e-05,
+      "loss": 0.0882,
+      "step": 21415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8085415592330413e-05,
+      "loss": 0.1867,
+      "step": 21420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8084946791055273e-05,
+      "loss": 0.2552,
+      "step": 21425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8084477989780133e-05,
+      "loss": 0.4176,
+      "step": 21430
+    },
+    {
+      "epoch": 1.0,
+      "eval_cer": 0.017639530100105133,
+      "eval_loss": 0.05617095157504082,
+      "eval_runtime": 420.9463,
+      "eval_samples_per_second": 45.255,
+      "eval_steps_per_second": 11.315,
+      "eval_wer": 0.151395023974025,
+      "step": 21431
+    }
+  ],
+  "max_steps": 214310,
+  "num_train_epochs": 10,
+  "total_flos": 2.574958852629261e+18,
+  "trial_name": null,
+  "trial_params": null
+}